Professional Documents
Culture Documents
Weronika Szubko-Sitarek
Multilingual Lexical
Recognition in the
Mental Lexicon of
Third Language Users
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Series editor
Mirosław Pawlak, Kalisz, Poland
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
The series brings together volumes dealing with different aspects of learning and
teaching second and foreign languages. The titles included are both monographs and
edited collections focusing on a variety of topics ranging from the processes under-
lying second language acquisition, through various aspects of language learning in
instructed and non-instructed settings, to different facets of the teaching process,
including syllabus choice, materials design, classroom practices and evaluation.
The publications reflect state-of-the-art developments in those areas, they adopt a
wide range of theoretical perspectives and follow diverse research paradigms. The
intended audience are all those who are interested in naturalistic and classroom
second language acquisition, including researchers, methodologists, curriculum and
materials designers, teachers and undergraduate and graduate students undertaking
empirical investigations of how second languages are learnt and taught.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Weronika Szubko-Sitarek
Multilingual Lexical
Recognition in the
Mental Lexicon of Third
Language Users
13
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Weronika Szubko-Sitarek
Department of English Language
and Applied Linguistics
Institute of English
University of Lodz
Lodz
Poland
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Acknowledgments
Professor Jan Majer has been my mentor throughout my entire academic career.
His patience, understanding, knowledge and ability to provide constructive
criticism—at the same time enhancing motivation—are second to none and I am
sure this book would never have been published had it not been for his profes-
sional and friendly assistance. Thanks are also due to Dr. Anna Parr-Modrzejewska
for her help with the visual elements contained in this work, as well as for gen-
eral encouragement. I also wish to acknowledge the help provided by Professor
Mirosław Pawlak, in particular his kind attitude, extraordinary patience and gener-
osity in enabling me to publish this volume as part of the series; without his con-
stant encouragement my effort would no doubt have gone to waste.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Contents
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
vii
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
viii Contents
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Contents ix
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
x Contents
Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Appendix C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Appendix D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
Appendix E . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Introduction
It is estimated that most language users in the world speak more than one
language, i.e. they are at least bilingual. In quantitative terms, then, monolingual-
ism may be regarded as the exception and multilingualism the norm (cf. Crystal
2003; Graddol 1997, 2006; Auer and Li Wei 2007; Aronin and Singleton 2012).
Nevertheless, it is only in the last two decades that issues of multilingualism started
to be examined closely and systematically. Indeed, it could no longer go unno-
ticed that a large part of the world’s population speaks several languages on a daily
basis and that a substantial proportion of language learners have some knowledge
of other languages beyond the second one. In fact, whether by living environment,
immigration or education, in many parts of the world the typical learner is no
longer a learner of a second language but rather of a third or additional language,
an L3 or an L4.
The wide spread of multilingualism, positively encouraged by people’s increas-
ing mobility across language borders, has led the ever-growing number of scholars
to investigate multilingual behaviour, as evidenced by the strong tradition of work
on sociolinguistic, educational and sociopolitical aspects of multilingualism (cf.
Cenoz and Genesee 1998; Cenoz et al. 2001b; Clyne et al. 2004; Edwards 1994,
2012; Kramsch 2006, 2014; Aronin and Singleton 2012). However, research on
the cognitive and psycholinguistic aspects of this young field of inquiry has been
much slower to appear. In fact, it was only recently that some researchers shifted
their interest to aspects relative to multilingual processing. Before describing the
structure of the present book, it is imperative to introduce some of the issues that
surround multilingualism. First, it will present some terminological inconsistencies
pertaining to the key notions in the field. Next, the existence of a mono- and bilin-
gual bias in multilingualism research will be discussed. An outline of the book
content concludes the present section, with a more precise introduction to the top-
ics discussed in each of the five chapters that follow.
As is usually the case with any other emerging field of inquiry, there is a
period of time in which the new area of research is perceived to be a mere exten-
sion of some related well-grounded fields. The most frequent tendency, then, is
to borrow the terminology already used in well-established fields such as SLA
xi
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
xii Introduction
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Introduction xiii
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
xiv Introduction
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Introduction xv
1994; Lemhöfer et al. 2008; Sánchez-Casas et al. 1992; Schwartz and Kroll 2006;
Schwartz et al. 2007; Voga and Grainger 2007). The faster production, recognition
and translation of cognates are usually attributed to a common set of form-based
representations (e.g., orthographic, phonological, morphological) that are used to
process them in both (or more) languages. These processing differences between
cognates and other words in multilinguals are acknowledged to indicate differ-
ences in the representations depending on the word type (cognate/noncognate).
In the Polish context research on the multilingual mental lexicon and the role
cognates play in its organization and processing is still rather limited. Hence the
need of the present author to explore this dynamically growing field and to verify
the applicability and generalisability of the research findings to the Polish setting.
According to statistical data published by the Central Statistical Office of Poland
(Pl. GUS), English is the first and German the second foreign language in Polish
schools (cf. Dmochowska 2010). Accordingly, taking this language constellation
as the basis for empirical research guaranteed a wide number of reliable respond-
ents. The other reason for choosing to use this pair of foreign languages through-
out the experiments is related directly to the ready availability both of natural
cognates and noncognates between German and English.
The main aim of this book is to report on a set of experiments which employed
both offline as well as online methodologies to investigate the role of cog-
nates in the internal structure of the mental lexicon of Polish–English–German
trilinguals. Additionally, it is hoped that the effects uncovered by this kind of
empirical work can also be used in foreign language instruction, the contention to
be further addressed in the final section of the present work. The structure of this
volume is organized into five chapters, the first four of which contain the discus-
sion of the relevant theoretical concepts and the last one reports on the findings
of a series of experiments conducted among Polish–English–German learners at
the tertiary level. Chapter 1 provides a comprehensive and up-to-date review of
studies into multilingualism and third language learning and teaching, including
examples mainly from a European background. It describes sociolinguistic, psy-
cholinguistic and educational aspects of multilingualism and emphasizes current
research trends in a fairly young area of multilingual teaching. More importantly,
the chapter introduces some basic differences between Second and Third or
Additional Language Acquisition and overviews the “no-difference” assumption
that shapes so many of the SLA and TLA studies currently available. Chapter 2
introduces the discussion on the structure of the monolingual mental lexicon and
word recognition models. It tackles issues related to lexical storage and process-
ing in the monolingual context. First, it describes various propositions concern-
ing the internal structure of the mental lexicon of monolingual speakers. Next, it
provides a brief review of models of lexical access with a special focus on visual
perception. Thereby, a broader, diachronic perspective on the general role of the
mental lexicon in lexical processing is presented. The main concern of Chap. 3 is
the presentation and discussion of research to date on the lexical structure of the
bilingual and multilingual lexicon, as well as on the hypothesized changes in lexi-
cal organization over time and the role of language proficiency in bringing about
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
xvi Introduction
these changes. The chapter additionally examines the evidence of integration and
separation of knowledge in the mind, verifies the arguments for and against non-
selective access hypothesis and accounts for issues related to task dependence
of multilingual word recognition studies. Much of Chap. 4 offers the theoretical
background to the empirical research reported in the last chapter by reviewing evi-
dence for the special status of cognates. It needs to be noted here that the term
cognate is viewed from a psycholinguistic perspective in the light of which it is
not etymological relation but formal and semantic resemblance that is of impor-
tance. Finally, Chap. 5 reports on the results of two off-line studies (similarity-
rating and translation-elicitation tasks) analysing the perception of cognates from
the form similarity perspective. The studies were conducted to identify German–
English cognates and noncognates as viewed by Polish learners of English (L2)
and German (L3). The outcome of these experiments was a list of German–
English cognates as identified by the speakers of Polish used in a series of four
experiments on a series of visual perceptual experimental studies conducted within
the lexical decision task paradigm whose aim was to test the assumptions concern-
ing the special position of cognates (the cognate facilitation effect) within a multi-
lingual mind and to answer the question whether trilinguals rely upon their second
language lexical knowledge when recognizing L3 words. The results of the experi-
ments attest to simultaneous activation and parallel processing as well as interac-
tion among all the three languages in a trilingual mind. At the same time, they
point to the fact that cross-linguistic lexical access and the source and strength of
transfer may be constrained by variables such as L1 dominance, language profi-
ciency linguistic typology as well as task demands. A detailed description of all
four experiments is offered in the main body of the chapter, whereas its final part
discusses the results, provides a brief synthesis of the general conclusions and
offers some recommendations for further research.
Notably, discussions in each chapter reflect the fundamental belief that research
on multilingual behaviour can offer some valuable insights into the processes
connected with non-native language acquisition. Therefore, the analysis and dis-
cussion of the results accrued in the experiments will be followed by some sugges-
tions which when applied in the multilingual educational context could contribute
to the increased efficiency of the process of third-language education.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Chapter 1
Beyond Bilingualism:
Issues in Multilingualism
1.1 Introduction
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2 1 Beyond Bilingualism: Issues in Multilingualism
The majority of people in the world speak several languages on a daily basis. It
can no longer go unnoticed that communicating in more than two languages has
become a dire necessity for the world today. Consequently, many language learn-
ers go on to study languages beyond the second one. As can be imagined, one
could name numerous reasons for the increasing interest in the phenomenon of
multilingualism, the most convincing of which postulates that humans are poten-
tially polyglot by nature since all of us “possess the capacity to learn several lan-
guages” (Mackey 1968, p. 555). Undoubtedly, there exists a multitude of more
pragmatic reasons accountable for that status quo, starting with the multilingual
situation of the majority of African countries and the rapid growth of international
cooperation and increasing contacts between the countries of the world; to, more
locally, the developing European integration. Another important factor is the mul-
tiplicity of a number of linguistic communities, especially in countries such as
Canada, Switzerland or Spain, where a vast number of research studies, especially
in the form of immersion programmes, have been conducted. Finally, the develop-
ment of English as world’s lingua franca (ELF) is yet another important variable
adding to the growth of studies into multilingualism. Interestingly, in contrast to
a widespread assumption that English in its role of a lingua franca constitutes a
serious threat to multilingualism, an ever-growing bulk of literature in this field
seems to suggest that ELF can be seen as one of the leading factors in the creation
of multilingualism today (cf. Hoffmann 2000; House 2003; Jessner 2006, 2013b;
Jenkins 2007; Seidlhofer 2004, 2011; Cogo 2008, 2012; Cenoz and Jessner 2009;
Archibald et al. 2011).
Nowadays it is universally recognized that to be multilingual is no longer an
aberration it was believed to be for years. Putting it bluntly, it is a dire necessity
for the majority of the world today (cf. Edwards 1994, p. 1). In fact, it has been
hypothesized that individual multilingualism is at least as frequent in the population
of the world as pure monolingualism, perhaps even more frequent (cf. Cook 1991;
Tucker 1998; Tokuhama-Espinosa 2003; Crystal 2003; Auer and Wei 2007; Aronin
and Singleton 2012). As Cook reports,
On some calculations there are more people in the world like the Cameroonian
(who speaks 4-5 languages) than like the Englishman; there are 3000-5000 languages in
the world but only about 150 countries to fit them all into (Cook 1991, p. 113).
Indeed, bearing in mind the statistical data concerning the distribution of lan-
guages within the countries of the world, gathered in Ethnologue, the most
comprehensive reference volume that catalogs all the known living languages
in the world today (cf. Lewis et al. 2014), the speculation that two-thirds of the
world’s children grow up in a multilingual environment seems quite plausi-
ble (Crystal 1997, 2003). And although this contention may be impossible to
document with full precision, many researchers refer to it as a powerful argu-
ment supporting the significance of multilingual research (cf. Cook 1992,
p. 578; Cenoz 2001a; Cenoz et al. 2003; Franceschini 2009). More importantly,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.2 The Multilingual World 3
And thus, for many decades it has been overlooked that in whatever form or con-
ditions the vast majority of the European population is multilingual. According to
Aronin’s (2005) diachronic approach to language use and behaviour, the attitude
towards individual multilingualism has since the 19th century gone through three
historical periods (or stages). They can be labeled monolingual (approximately
up to the late 1950s), bilingual (from the 1960s up to the late 1980s) and multi-
lingual (from the 1990s up to the present). The proposed diachronic division
implies that “in these periods particular language(s) arrangements were socially
significant and represented a dominant trend” (Aronin 2005, p. 10). Each of these
periods is also characterized by the prevailing societal attitude and interest
toward the phenomena and social response, like the world-wide spread of immer-
sion programs, or more recently, the rise of tertiary education1 and plurilingual
didactics (cf. Sect. 1.6.3).
1 The term “tertiary education” (also “tertiary languages”) has been adopted by some linguists
to refer to foreign languages learned after the first foreign language, i.e. as one’s second, third,
fourth, etc. foreign language (cf. Hufeisen 1991, 2004).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4 1 Beyond Bilingualism: Issues in Multilingualism
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.3 Multilingualism—Terminology and Definitions 5
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
6 1 Beyond Bilingualism: Issues in Multilingualism
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.3 Multilingualism—Terminology and Definitions 7
three or more. In practice, definitions used in the literature on the topic do not
stick to the “number of languages” rule. The terms bilingualism/multilingualism
and bilinguals/multilinguals are often used as synonyms. Obviously, many pro-
posed definitions of bilingualism restrict themselves explicitly to two languages.
Nevertheless, the implicit suggestion that several lingualisms can be subsumed
under the concept of bilingualism was already illustrated in early publications in
the history of multilingual studies, e.g., in the works of Skuttnab-Kangas (1984)
who allowed for more than two languages to be present in the person she defined
as bilingual. According to Oksaar, bilingualism was “the ability of a person to use
here and now two or more languages as a means of communication in most situa-
tions and to switch from one language to the other if necessary” (Oksaar 1983, p. 19).
In a similar vein, Grosjean saw bilingualism as “the regular use of two (or more)
languages, and bilinguals (as) those people who need and use two (or more) lan-
guages in their everyday lives” (Grosjean 1992, p. 51), Myers-Scotton (2002, p. 1)
proposed that the term “bilingual refers to persons who speak two or more lan-
guages”, whereas Cenoz and Genesee (1998) defined the term as the final outcome
of the process of acquisition of several non-native languages. It can be concluded
that as for the number of languages involved, the term bilingualism “(…) has so
many contradictory definitions and associations in popular and academic usage
that it seems best to avoid it whenever possible” (Cook 2002, p. 4). For that rea-
son, Jessner (1996, 2008a, b) and Hoffmann (2000) look at bilingualism from an
entirely new, one could say “reversed”, perspective and refer to the phenomenon
as a quantitatively specified subtype of multilingualism “albeit a common one”
(Herdina and Jessner 2000, p. 84). Following this line of reasoning, additional
terms that started to appear recently in the literature on multilingualism have been
trilingualism and quadrilingualism. Naturally enough, they are used in reference
to a precise number of languages involved.
Equally disputable as the definitions of bilingualism and multilingualism are
the notions of a second (L2) and a third language (L3). Firstly, it should be noted
that in the technical sense L2, L3 or L4 are not necessarily equal to language num-
ber two, three or four in order of acquisition; instead, the numbers point to the
level of proficiency following the equation: the higher the number, the less profi-
cient the speaker (cf. Hufeisen 1991, 2004). Secondly, analysing the literature on
multilingualism it becomes obvious that many scholars still approve of the bilin-
gual bias and do not differentiate between L2, L3 and Ln learners claiming that
the processes behind non-native language acquisition are essentially the same (cf.
De Angelis 2007). Thus languages that are acquired after the first language (or
first languages, in the case of simultaneous bilingualism) are commonly termed
second languages, L2s or Ln, whereas all non-native language learners are often
labelled L2 learners. This extension of meaning proves that many scholars even
perceive the distinction between L2 and L3 or Ln acquisition as unnecessary. By
way of example, Haugen referred to multilingualism as “a kind of multiple bilin-
gualism” (1956, p. 9) suggesting no qualitative differences between the two phe-
nomena. Similarly, Sharwood Smith opined that “[s]econd language acquisition
(SLA) normally stands as a cover term to refer to any language other than the first
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
8 1 Beyond Bilingualism: Issues in Multilingualism
More recently, Li Wei (2007, p. 7) pointed out that although the word bilingual is
primarily used in reference to someone using two languages, it can also be taken
to include the many people in the world who have varying degrees of proficiency
in and interchangeably use three, four or even more languages. Also Mitchell and
Myles (2004, p. 6) promoted the view that foreign languages should be included
under one more general term second languages claiming that the underlying pro-
cesses are essentially the same for all non-native languages.
Of late, however, a rationale underlying the interchangeability of multilingualism
and bilingualism as supplied by the foregoing researchers has been called into ques-
tion. Undoubtedly, these two phenomena share many similarities both at the psycholin-
guistic as well as at the sociolinguistic level; still, there seem to be many good reasons
to recognize that multilingualism has “characteristics of its own” (Hoffmann 2001a,
p. 3). Proponents of distinguishing studies on L3/Ln acquisition from the research on
bilingualism (cf. Cenoz and Jessner 2009; Aronin and Singleton 2012) note that factors
affecting third or additional language acquisition are much more numerous and much
more complex than those involved in the process of L2 learning. The variables include
among others: age of acquisition, sequence of acquisition of all non-native languages,
proficiency level in all languages or context and manner of acquisition. Most impor-
tantly, L2 is taught to monolinguals, whereas L3 to bilinguals. Claiming that processes
involved in learning L2 and L3 are the same would by extension imply that monolin-
guals do not differ from bilinguals (cf. Sect. 1.4).
According to Herdina and Jessner, “learning a third language differs essentially
from learning a second—something third language learners themselves intuitively
perceive” (Herdina and Jessner 2002, p. 96). As a matter of fact, in the light of
their Dynamic Model of Multilingualism (2002) learning any subsequent language
results in a qualitative change in the speaker’s language system as the acquisition
of a new language leads to the development of new skills (i.e. language learning
skills, language management skills and language maintenance skills; cf. Herdina
and Jessner 2002, pp. 92–93). Consequently, the process of L3 acquisition should
be depicted on the multilingual continuum, ranging from monolingual acquisition
(the acquisition of a foreign language based on the command of one language)
through balanced bilingualism to the command of three (or more) languages. Such
a perception on L3 learning supports the view that bilinguals should not be equated
with trilinguals as they are not extended bilinguals, just like a bilingual is not the
extension of a monolingual. As Cenoz and Genesee (1998) point out,
[m]ultilingual acquisition and multilingualism are complex phenomena. They implicate
all the factors and processes associated with second language acquisition and bilingual-
ism as well as unique and potentially very complex factors and effects associated with the
interactions that are possible among the multiple languages being learnt and the processes
of learning them (Cenoz and Genesee 1998, p. 16).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.3 Multilingualism—Terminology and Definitions 9
A further terminological refinement that needs to be noted concerns the levels and
breadth of proficiency needed to apply the term to persons using two or more lan-
guages. As far as definitions of bilingualism based on competence are concerned,
they range from “native-like control of two languages” (Bloomfield 1933, p. 56)
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
10 1 Beyond Bilingualism: Issues in Multilingualism
The same competence controversy holds for multilingual users. The question
arises whether the label multilingual should be reserved only for those individuals
whose proficiency is native-like and balanced across all their languages and across
the range of language skills (cf. Aronin and Singleton 2012). Following Cook’s
studies on multicompetence (e.g. Cook 2007a, b, Sect. 1.4). It seems to be more
likely that the proficiency levels necessary for a language user to be classified as a
multilingual cannot be specified.
Finally, another question is connected to the proficiency threshold level in
non-native languages. Namely, when do individuals start being multilingual?
How many hours/years of instruction do they need to become bi- or multilin-
gual users? How proficient do they need to be in their L2, L3 or Ln? And how is
their language proficiency to be assessed? How much is enough? (see Aronin and
Singleton 2012, pp. 2–3 for a comprehensive discussion).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.4 Towards a Holistic View of Multilingualism 11
linguistic features not necessarily compatible with those of bilingual speakers. The
assumption which is strictly related to the notion of multicompetence. The idea
of a ‘holistic’ interpretation of bilingualism as opposed to a monolingual ‘frac-
tional’ interpretation of bilingualism, initially put forward by Grosjean (1992,
2010), has over the last two decades attracted significant research attention in
the study of Third or Additional Language Acquisition. Similarly, the concept of
“holistic multicompetence” (cf. Cook 1992, 1995, 2008, 2012a, b) proposed by
Cook has been based on the essential tenet of holism which postulates that the
whole is more than the sum of its parts. By the same token, “a bilingual is NOT
the sum of two complete or incomplete monolinguals” (Grosjean 1992, p. 55, cap-
italized in the original), just like the trilingual is not the extension of the bilin-
gual speaker (Jessner 1996, 2006, 2008b; Herdina and Jessner 2002; Aronin and
Hufeisen 2009; Aronin and Singleton 2012). According to Cook’s pioneering defi-
nition relating to bilingualism, multicompetence is “the compound state of mind
with two grammars” (Cook 1992, pp. 557–558). Herdina and Jessner postulate
that multicompetence is an integrated entity “consisting of dynamically inter-
acting linguistic subsystems” (Herdina and Jessner 2002, p. 75), whereas Clyne
speaks about the multilingual’s “multilateral competence” (Clyne 1997) which
encompasses linguistic as well as procedural knowledge and finds evidence for
it in various aspects of multilingual, more precisely trilingual, behaviour. Finally,
taking the individual aspect of multilingual proficiency into account, Aronin and
Ó Laoire (2004) perceive the nature of multicompetence as an ecosystem or a
“bionic system” arguing at the same time that the study of multilingualism should
be based above all on the notion of identity. Unfortunately, space constraints pre-
vent this section from offering a detailed description of the multitude of works on
multicompetence. Therefore, in what follows, only a brief look at this fundamental
concept is presented.
First of all, Cook (1992) argues that what differs monocompetence from multi-
competence is not only the number of languages involved, but also the quality of the
linguistic knowledge. In his influential article presenting evidence for multicompe-
tence (1992), the author claims that since L2 users differ from L1 users in their L1
knowledge, metalinguistic awareness and cognitive processes, they are not simply
equivalent to two monolinguals but instead should be perceived as unique combina-
tions (Cook 1992). In the same vein, Cenoz and Genesee (1998, p. 19) indicate that
“multilinguals possess a configuration of linguistic competences that is distinct from
that of bilinguals and monolinguals competence”. Further, it is claimed that mul-
tilingual language competence contains the linguistic aspects, i.e. vocabulary and
grammar, from all the language systems of the learner, and also the pragmatic com-
ponent, consisting of sociolinguistic, discourse and strategic competences pertaining
to all the languages involved. Additionally, it includes the ability to function in mul-
tilingual contexts, which require decisions on code choice and code-switching. As
Grosjean postulates, bilinguals (and by extension multilinguals) have “a unique and
specific linguistic configuration” (Grosjean 1992, p. 55). And since for multilinguals
it is normal to move between different languages, switching, mixing and borrowing,
clearly, multilingual learner performance must exhibit linguistic features which are
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
12 1 Beyond Bilingualism: Issues in Multilingualism
While the previous sections endeavoured to shed some light on the terminological
problems related to specifying and differentiating the notions of bi- and multilin-
gualism by offering a concise review of definitions connected with these phenom-
ena, the subsequent paragraphs will focus on developmental patterns and cognitive
abilities of multilinguals. The following subsections will also seek to address
the pertinent issue of the qualitative differences that have been reported to occur
between multilingual and bilingual language users and which have been briefly
referred to in the foregoing discussion on multicompetence.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.5 Characterizing Multilingual Learners 13
For many years research has only applied the term multilinguality to cases in
which “learners are characterized as bilingual or multilingual as a result of natural
language acquisition processes and have a similar level of competence in all lan-
guages” (Cook 2000, p. 23). However, in recent years, the situation has changed
considerably. Consequently, with reference to the taxonomy of multilingual stud-
ies, many researchers distinguish, nowadays, between natural and formal multilin-
guality (cf. Fig. 1.1), at the same time emphasizing the complexity of both groups
and the variability among multilingual speakers themselves.
As shown in Fig. 1.1, natural multilinguality comprises citizens of bi- and
multilingual countries, children from mixed marriages and immigrants com-
ing to live in another country (both permanently or temporarily), whereas the
partially naturalistic setting of acquisition refers to natural bilinguals learning
a foreign language in the classroom context, e.g. a bilingual child brought up
with two school languages who learns another language in the formal setting. In
turn, formal multilinguality refers to monolingual or bilingual learners acquir-
ing their subsequent language(s) by means of formal instruction, especially in a
school context.
The complexity of the two former groups, is believed to be determined by
the social context, whose influence is difficult to measure since it is shaped by a
multitude of social, educational and political factors (cf. Barron-Hauwaert 2000;
Singleton et al. 2013). The formal type of multilinguality, on the other hand, the
one dependent on classroom instruction, is complex due to all possible language
configurations, typological and psychotypological proximity of the languages, the
learners’ motivation, the age factor, the previous linguistic knowledge and the prior
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
14 1 Beyond Bilingualism: Issues in Multilingualism
language learning experience, the metalinguistic and strategic awareness and, last
but not least, the instruction itself. These points granted, there is no denying the
fact that “the very nature of the process of formal multilinguality may create pat-
terns very different from those of the natural multilingual context” (Gabryś-Barker
2005, p. 19).
In bilingual acquisition two basic orders of linguistic development have been
distinguished. These are: compound (L1 → L2) and coordinate (L1 + L2) bilin-
gualism. As suggested above, in the case of multilingualism the situation is far
more complex since the number of languages involved in multilingual acquisi-
tion multiplies their possible acquisition orders, which in turn further compli-
cates the study of the relationships among the languages involved. In her account
of the research on multilingual acquisition Jasone Cenoz distinguished four dif-
ferent paradigms to be followed by a trilingual subject; all of them amassed in
Table 1.1. The first paradigm (L1/L2/L3) illustrates a fully naturalistic situation
typical of multilingual African countries where three or even more languages
are used by individuals on a daily basis. Type b (L1/L2 → L3) is referred to as
partially naturalistic and includes, among others, children from mixed marriages
learning another language at school or bilingual immigrants acquiring the lan-
guage of the country they had come to live in. Finally, types c (L1 → L2/L3)
and d (L1 → L2 → L3), apart from being valid for immigrants, show fully for-
mal contexts of language acquisition by monolingual speakers learning their for-
eign languages in the classroom context either simultaneously or consecutively
(sequentially).
Due to such a big number of potential patterns of multilingual development,
and often contradictory opinions on the level of proficiency necessary to classify
someone as a bi- or multilingual, it seems almost impossible to design general
models and hypotheses concerning multilinguals. Therefore, every single study
needs to precisely specify what multilingual profile it pertains to. Otherwise, its
results may turn out not to be applicable to other multilinguals. In the research
reported in Chap. 5 special attention will be given to group d as it is the most
representative category of the Polish educational context (cf. Arabski 2002;
Widła 2005; Gabryś-Barker 2005; Siek-Piskozub et al. 2008) and the one ena-
bling to analyze the importance of prior L1 and L2 language knowledge on the
process of instructed third language acquisition (which is the underlying aim of
the present work).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.5 Characterizing Multilingual Learners 15
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
16 1 Beyond Bilingualism: Issues in Multilingualism
score and intelligence of the tested bilingual children, e.g., tests by Goddard (1917;
in Romaine 1995) or Brigham (1923; in Romaine 1995), who concluded that the
language handicap is responsible for mental retardation, in other words for lower
intelligence of the examined bilinguals. It was only much later that their results
were highly criticized, methodological shortcomings and biased interpretation of
the data being at the foot of this criticism.
The period from the late 1950s to the early 1960s is usually referred to as the
neutral period in the history of research on bilingualism and intelligence. During
this time methodological weaknesses of previous research were brought into light
and allowed for the development of the positive period. The development of cog-
nitivism and cognitive methodology in the late 1960s allowed researchers to look
at the processes involved in language acquisition and various aspects of facilitation
bilingual competences perform, thus initiating the period of additive effects of bilin-
gualism (cf. Hakuta 1986). Since then, bilinguals have been compared to monolin-
guals with respect to their cognitive abilities in a multitude of psycholinguistic and
neuropsychological studies. A contrasting attitude to those of Jespersen (1922) and
Weisgerber (1966) towards bilingualism and its influence on children’s development
came from what is now regarded as a landmark in the studies on bilingualism and
intelligence, namely Peal and Lambert’s article on “The Relation of Bilingualism to
Intelligence” published in 1962. The researchers found that bilingual children per-
formed better than monolingual children in both verbal and non-verbal intelligence
tests. They additionally noticed that bilinguals were significantly more successful at
certain types of tasks which required mental manipulation, all of which suggests the
positive influence of bilingualism on the children’s cognitive development. Having
studied all available evidence, Peal and Lambert argued that “intellectually [the
bilingual child’s] experience with two language systems seems to have left him with
a mental flexibility, a superiority in concept formation and a more diversified set of
mental abilities” (Peal and Lambert 1962 p. 20). It needs to be noted that theirs was
one of the earliest studies which accounted for variables like age, type of bilingual-
ity or social status in the development of individual bilinguals. Since then the view
that bilingualism promotes cognitive development and facilitates third language
acquisition has been reported in a number of studies (for a review see Cenoz 2003).
Bilinguals have been repeatedly reported to present higher scores in tests of diver-
gent thinking or creative thinking (cf. Baker 2006; Cenoz 2003), to obtain bet-
ter results in tasks demanding a high level of analysis (Bialystok 2001). They are
also more sensitive to the communicative needs of their interlocutors and use more
varied communication strategies (Thomas 1992). Carringer (1974) emphasizes the
greater cognitive flexibility of bilinguals which enables them to better separate form
and content because they have two terms for one referent. Of late, bilingualism has
been even reported to “(…) protect[s] against age-related cognitive decline” and
“postpone the onset of symptoms of dementia” (Craik et al. 2010).
Considering the above, the question arises whether knowledge of a second lan-
guage further enhances cognitive development and aids in the acquisition of lan-
guages beyond the L2. The majority of research studies in this area have confirmed
that prior language knowledge and prior learning experience facilitate the acquisition
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.5 Characterizing Multilingual Learners 17
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
18 1 Beyond Bilingualism: Issues in Multilingualism
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.6 Research Themes in the Study of Multilingualism 19
Table 1.2 Types of studies into multilingualism and MLA (based on Marx and Hufeisen 2004,
pp. 146–150)
General category Areas of research
Models and state-of-the-art Research desiderata, models of multilingualism; projects
articles promoting multilingualism and TLA (e.g., EuroCom)
Empirical studies TLA/MLA acquisition by children—case studies; accounts of
the negative and positive effects of previous foreign language
learning experience; phenomena of transfer; studies on individ-
ual differences in multilingual contexts (e.g., learning strategies,
motivation, metalinguistic awareness); discourse analysis; stud-
ies on the multilingual mental lexicon; neurolinguistic studies
Sociocultural aspects of Students’ attitudes; multilingual societies; environmental
multilingualism influences
Educational aspects of Pedagogical implications of multilingual instruction; lifelong
multilingualism language learning; the teaching of tertiary languages; curricular
suggestions; didactical and methodical issues concerning L3
teaching; different multilingual classroom concepts, cross-
linguistic influence (CLIL)
Political aspects of language Language-political aspects; political dimension of
learning multilingualism
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
20 1 Beyond Bilingualism: Issues in Multilingualism
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.6 Research Themes in the Study of Multilingualism 21
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
22 1 Beyond Bilingualism: Issues in Multilingualism
but that these are more complex when three languages are involved” (Clyne 1997,
p. 112). More importantly, the results obtained from the research lead him to con-
clude that “trilinguals (are) trilingual in different ways” (Clyne 1997, p. 113).
Clyne speculates that some trilinguals are more like bilinguals who have two dis-
tinct standard languages and “a non-standard variety regarded as part of one of
them in some ways” (ibid.), whereas other trilinguals are more like double bilin-
guals, “with two pairs of languages where each pair has a special relationship
with their L3 (English) rather than with the other languages” (ibid.).
One of the most dynamic areas of bilingualism, and more recently, multilingual-
ism research involves the psycholinguistic study of both children and adults. It
needs to be noted that the majority of studies within this domain predominantly
focus on bilingual users, however, a steady growth of research dealing with more
than two languages cannot go unnoticed. Perhaps the most numerous group
of studies exploring the psycholinguistic aspects of multilingualism has been
conducted into the area of multilingual language processing. The mechanisms
involved in comprehension and production in two or more languages have been
examined with reference to phonetics, lexis, and grammar. The comprehensive
amount of studies concern spoken language processing including speech per-
ception and comprehension (Grosjean 2008, 2010) as well as speech production
(de Bot 1992, 2004; Costa et al. 2005). The central question posed here relates to
the dynamic character of multilingual speech production enabling bilinguals and
multilinguals “to generally produce language in the ‘intended’ language”. Another
group of studies deals with written language processing. Here special attention has
been given to visual word recognition (Dijkstra 2005, cf. Chap. 3), far less to the
processing features of bilingual text production (Manchón 2013). As regards mul-
tilingual recognition, the leading question posed by researchers has been whether
visual word recognition is restricted to the contextually appropriate language or
whether the other languages are also involved. Overall, the research results dem-
onstrate a considerable amount of interaction between the languages known by
a multilingual, which has led the majority of researchers to believe that lexical
access in multilinguals is basically nonselective with respect to language. Of late,
however, a growing amount of research has been conducted to verify whether lan-
guage nonselectivity is maintained or eliminated in context as the empirical data
gathered thus far show that when words are processed in sentence context, their
processing seems to be sensitive to the semantic and syntactic aspects of the sen-
tence (cf. Hartsuiker et al. 2004, 2008; in Riehl 2010). Research on visual word
recognition can pride itself on a relatively long tradition and throughout many
years of studies a number of factors affecting written lexical access have been
enumerated and explored (cf. Warren 2013). They constitute yet another area of
research into multilingual processing.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.6 Research Themes in the Study of Multilingualism 23
Within the psycholinguistic area much research has been dedicated to simultaneous
(Yip and Matthews 2007, 2010) and successive language acquisition (Bialystok 2001,
De Angelis 2007, Pavlenko 2009) in multilinguals. Here, a recurrent question
addressed by psycholinguists relates to cross-linguistic influences. In fact, the study of
different forms of cross-linguistic interaction, including code switching and code-mix-
ing, has had a long tradition in research on multilingualism. Interestingly, recent stud-
ies on third or additional language acquisition have confirmed the multidirectionality
of cross-linguistic interaction. Moreover, outcomes of numerous experiments indicate
that there could be closer links between users’ foreign languages than between these
languages and the mother tongue, which seems to be most probable in the case of
typologically related languages learned after L1 (see, e.g., De Angelis 2007). Cross-
linguistic interaction has also been explored in the context of the early acquisition of
two or more languages (Paradis 2007).
Yet another group of studies analyzing multilingualism from the psycholinguis-
tic perspective covers the bilingual (multilingual) brain and cognition. Here special
attention has been devoted to the bilingual lexicon. Different models of organiza-
tion of lexical word forms and their conceptual representations in the bilingual and
multilingual lexicon have been devised and tested in practice (cf. Chap. 3; De
Groot 2002, 2011, Pavlenko 2009). The central question posed by researchers deal-
ing with the multilingual brain concerns the multilingual lexical storage and retrieval
(Dijkstra 2003a, b) both in production and comprehension. It is worth mentioning that
new possibilities of examining the multilingual brain have opened along with the use
of neuroimaging techniques (MRI, fMRI, PET) and methods to analyze the electrical
activity of the brain (ERP, EEG). They offer the opportunity to relate language pro-
cessing to different parts of the brain and to explore some characteristics of bilingual
and, more recently, multilingual processing with more accuracy and from different
angles. Further, there are also studies that investigate the implications of being bilin-
gual on various aspects of non-linguistic cognition. Many researchers (Bialystok and
Craik 2009, Bialystok et al. 2009, 2010) work on verbal abilities of bilingual children
and adults in terms of vocabulary size, metalinguistic awareness and learning to read.
Finally, mention must be made of research projects on the relationship between
multilingualism and conceptualization (Pavlenko 2009, 2011). Whereas some
psycholinguists promote the view that the conceptual base of monolinguals and
multilinguals is identical (de Bot 2008), others support the position that there are
some significant qualitative differences between mono and multilingual concep-
tualisation (Kecskes 2010). Many of them explore the relationship between the
acquisition of additional languages and conceptual development and restructuring.
Interesting findings have been reported on areas such as visual perception, inner
speech, and gesturing (Cook and Bassetti 2011; Jarvis and Pavlenko 2008).
In the foregoing section a brief and rather eclectic look at the different areas of
the psycholinguistics of multilingualism has been offered. It needs to be noted that
studies on visual word recognition are of crucial importance for the experimental
part of this volume and as such they will be further elaborated on in the following
chapters entirely devoted to a more meticulous survey of experimental studies and
theoretical models of multilingual visual comprehension.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
24 1 Beyond Bilingualism: Issues in Multilingualism
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.6 Research Themes in the Study of Multilingualism 25
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
26 1 Beyond Bilingualism: Issues in Multilingualism
The claim that plurilingualism is innate in all individuals, because even in their
mother tongue they shift back and forth between a number of often clearly dif-
ferentiated language variants appeared already in late 1970s in the publications
by Mario Wandruszka who distinguished two types of this phenomenon. The
first type, referred to as “internal plurilingualism”, concerns the already men-
tioned language variants such as standard language, dialect, colloquial lan-
guage, specialist jargon or knowledge of earlier linguistic forms of their own
language that are nonetheless linked with each other. To this was added indi-
viduals’ ability to learn other languages in addition to their own in the course of
their intellectual development. The latter type was called “external plurilingual-
ism”. In fact, the term plurilingualism in its modern usage requires some clari-
fication since in the last few decades it has come to denote a range of different
meanings.
To date research has typically applied the term plurilingualism, as synony-
mous to individual multilingualism (cf. Sect. 1.3.2), to cases in which learners
are characterized as plurilingual as a result of natural language acquisition pro-
cesses and have a similar level of competence in all languages (cf. Hufeisen and
Neuner 2004). However, plurilingualism as it has emerged in the current discus-
sions of foreign language teaching and as it is also used in Council of Europe
documents is an utterly different concept. Using the metaphor of a “threshold
level” Christ (2001, p. 3) suggests that a person is plurilingual if, “with respect
to a number of languages, he/she has learned to cross the threshold into these
different language houses.” Indeed, as a result of the circumstances of their lives
and the demands of their jobs, plurilingual learners usually reach different lev-
els with respect to the four competences. More significantly, plurilingualism
is at present defined and distinguished from multilingualism as a plurilingual
competence that does not simply add on the skills in one language to those of
another, but rather combines and interrelates them in a variety of ways. This is a
plurilingualism that is inter alia characterized by interlinguistic and intercultural
awareness (cf. Abendroth-Timmer and Breidbach 2000; Jessner 2006; Aronin
and Hufeisen 2009). Beyond this, the plurilingual approach emphasizes the fact
that as an individual person’s experience of language in its cultural contexts
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.6 Research Themes in the Study of Multilingualism 27
expands they do not keep their languages and cultures in strictly separated men-
tal compartments, but rather “builds up a communicative competence to which
all knowledge and experience of language contributes and in which languages
interrelate and interact” (Council of Europe 2001, 4; cf. Sect. 1.4). Basing on the
principles of Cook’s multicompetence (1992, 2007a, b) and Grosjean’s holistic
view of bilingualism, Neuner enumerates (2004, pp. 15–16) the following fea-
tures of plurilingual learners, emphasizing the specificity of the contemporary
interpretation of the term in question. Firstly, the level of competence and the
language profile in each language learned can be very different. Secondly, there
is no need to achieve the ideal of “near nativeness” in each of the languages to
be learned. Lastly, when several languages are learned, the learner does not begin
“at zero” in each case, but rather the existing language possession is continually
extended by each new language.
As regards types of plurilingual learners, the same author singles out three
types of the phenomenon in question: retrospective, retrospective-prospective
and prospective (cf. Neuner 2004, p. 15). Retrospective plurilingualism means
that learners bring their plurilingualism into the classroom. They are (to a
large extent) bilingual, with considerable knowledge of L2, the language being
taught, and thus have a definite advantage in terms of knowledge and skills in
this language over the other learners. Retrospective-prospective plurilingual-
ism refers to learners who bring their plurilingualism into the classroom and
therefore have a substantial lead in linguistic knowledge over the other learners,
but neither of these two languages is the subject being taught. Through teach-
ing in an L3 (or Ln) the learners are extending their plurilingualism. Finally,
prospective plurilingualism is used in relation to the learner who arrives in the
foreign language classroom as a monolingual and first begins to develop and
extend their plurilingualism in the teaching of the foreign language. This is the
situation that is assumed to be typical for foreign language teaching in the first
foreign language. The research presented in the final chapter of this volume
focuses on retrospective-prospective plurilinguals. It aims at analyzing their
ability to use the already acquired lexical knowledge in the process of visual
word recognition.
For many decades it has been overlooked that in whatever form or conditions the
vast majority of the European population is multilingual. In fact, it was not until
the development of the European integration that the monolingual bias began to
disappear. Nowadays, as some researchers believe, “the concept of the monolin-
gual native speaker has no place in the new Europe” (Cheshire 2002, p. 33). Today,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
28 1 Beyond Bilingualism: Issues in Multilingualism
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.6 Research Themes in the Study of Multilingualism 29
Fig. 1.2 Key areas for multilingual development (based on a new framework strategy for multi-
lingualism, 2005)
2 Cf. The European Commission's White Paper Teaching and Learning: Towards the Learning
Society (1995) considers that European citizens should be proficient in three community lan-
guages and recommends foreign language teaching at pre-school level in order to allow for sec-
ond foreign languages in secondary school.
3 Cf. A New Framework Strategy for Multilingualism [Brussels, 22.11.2005 com (2005) 596
final].
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
30 1 Beyond Bilingualism: Issues in Multilingualism
When it comes to language teacher training, Cenoz and Gorter (2013) recently
proposed an innovative pulrilingual approach to the teaching of English, which
as they claim softens the boundaries between languages (Cenoz and Gorter 2013,
p. 591). The authors point out that learning English in Europe cannot be separated
from the use of other languages in education, and argue for a language policy that
moves from the traditional monolingual ideology towards adopting holistic pluri-
lingual approaches in the teaching of second and foreign languages. They believe
that a plurilingual approach highlights how learners relate the languages in their
repertoire to each other when learning English as an additional language and when
they use their languages in a social context (Cenoz and Gorter 2013, p. 597). They
conclude that,
it is time for TESOL professionals to use the opportunity to accelerate the learning pro-
cess by using plurilingualism as a resource and not as an obstacle by setting attainable
goals, using the learners’ plurilingual repertoire, integrating syllabi, and using learners’
linguistic creativity as a resource (Cenoz and Gorter 2013, p. 597).
1.7 Conclusion
In this chapter an attempt has been made to explore different aspects of the con-
cept of multilingualism. The presentation of numerous definitions of bilingualism
and multilingualism differing as regards content and scope proves how complex
and diversified the perception of these phenomena is. The picture that emerges
from the foregoing discussion is rather fragmentary and incomplete since attempts
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
1.7 Conclusion 31
to define a multilingual user are characterized by diversified, and quite often con-
tradictory, opinions and research findings. As Jessner (2006) puts it, multilingual-
ism and multilinguality are multifaceted constructs which can be viewed from a
variety of angles and whose study has only just begun.
As regards the multilingual research, it cannot go unnoticed that the body
of multilingual literature written in the last few decades has been growing fast.
However, there is no denying the fact that further extensive studies are still needed
before all the terminological inconsistencies are solved and concrete models are
forwarded. Marx and Hufeisen (2004) propose the following desiderata for the
study of multilingualism: more longitudinal studies; the use of more learner-ori-
ented data such as learner questionnaires, introspection and retrospection or think-
aloud protocols, as well as more in-depth research on the learner’s metalinguistic
awareness or research into positive results of multilingualism and TLA.
The above discussion has shed some light on terminological complexities and
definitional problems. Numerous theoretical issues related to the phenomenon of
multilingualism have been tackled including deliberation on such concepts as mul-
tilingual cognitive skills, multicompetence or metalinguistic awareness. Special
reference has been made to Cook’s (e.g. Cook 2007a, b) notion of holistic mul-
ticompetence which posits a very high degree of integration of language compe-
tence across languages. In the final part of the chapter the discussion has revolved
around selected research areas. Since the experimental part of the work concerns
research on the multilingual mental lexicon, special attention has been paid to the
presentation of studies dealing with the psycholinguistic aspect of multilingualism.
Now that the main issues concerning the concept of multilingualism have been
delineated, it is time to outline influential theoretical models and the underlying
research experiments concerning the structure of the mental lexicon of mono- and
multilingual users.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Chapter 2
Modelling the Lexicon: Some General
Considerations
2.1 Introduction
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
34 2 Modelling the Lexicon: Some General Considerations
The mental lexicon includes a large number of lexical entries containing all the
information on individual words. But what precisely are these individual words
and what do they consist of? Within the psycholinguistic tradition two proposi-
tions concerning the issue of the internal structure of the lexical entry merit atten-
tion; the first, whose most fervent supporter is Levelt (1989, 1993) and the other,
put forward by Bierwisch and Schreuder (1992).
Many linguists (cf. Aitchison 2003a, 2012; Levelt 1989, 1993) support the view
that all the information “behind” a word can be allocated to two separate compo-
nents: a semantic component called a lemma1 (including the information on the
word’s meaning, its connotations, style, and its syntactic pattern) and a formal one,
frequently referred to as a lexeme (including the word’s morphology, phonology
and orthography). According to Levelt, each lemma has a lexical pointer which
“indicates an address where the corresponding word-form/information is stored”
(Levelt 1989, p. 165). Levelt enumerates four main internal features of a lexical
item (cf. Fig. 2.1): meaning, defined as the semantic information which lists “a set
of conceptual conditions that must be fulfilled in the message for the item to
become selected” (Levelt 1989, p. 165), syntax (including syntactic arguments and
other properties), morphology and phonology. He also points to some stylistic,
pragmatic and affective attributes of a word.
Another influential perception of the internal structure of a lexical entry was
proposed by Bierwisch and Schreuder (1992). There are many similarities between
Levelt’s and Bierwisch and Schreuder’s models since both approaches argue in
favour of similar internal features of the lexical entry. Bierwisch and Schreuder
enumerate the phonetic form, the grammatical form, the argument structure and the
semantic form. There is, however, one fundamental difference between the models.
Fig. 2.1 The structure of
lexical representations in the
mental lexicon (adaptation
based on Levelt 1989)
1 The term lemma was first used by Kempen and Hujbers (1983) in reference to the part of the
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.2 The Internal Composition of a Lexical Item 35
Namely, the manner in which the meaning is represented. The key question is
whether semantic representations of items are identical with general world knowl-
edge or whether it is possible to draw a direct line between word meanings and con-
cepts which represent encyclopedic information. Similarly to other psycholinguists
and unlike Levelt, who is a fervent proponent of the holistic approach to meaning
representation, Bierwisch and Schreuder (1992) believe that the internal structure
of the lexical meaning of the entry is a composition of more primitive units. In con-
trast, Levelt asserts that the meaning of lexical items is “represented as a whole
which cannot be decomposed into separate elements” (Levelt 1993, p. 28).
The findings so far, then, are that although there exists a general agreement as
to the constituents of a word, two conflicting approaches concerning the issue of
the representation of meaning are still under discussion. The first approach,
called the one-level model or the network model (cf. Levelt 1989), considers
semantic and conceptual knowledge identical. The other approach, known also as
the two-level model (cf. Bierwisch and Schreuder 1992), differentiates between a
word’s semantic meaning and the more general conceptual knowledge the item
refers to. It needs to be noted that this latter theory relates to the proposition
advocating the modularity of mind (cf. Fodor 1983, 1989). In the light of this
approach a human linguistic system forms a closed mental module which does
not depend on other mental faculties but is to some extent interconnected with
them. Hence a lexical item is connected with the more general conceptual
domain of world knowledge.2
All in all, the commonly adopted approach to the structure of a lexical entry
and to the representation of meaning is the compromise between the two presented
options. On the one hand, this approach draws a boundary between semantic and
conceptual knowledge and conceives of them as non-identical. On the other hand,
it admits that these two types of knowledge are strongly related (cf. Aitchison
2003a, 2012; Pickering and Garrod 2013; Randall 2007).
The following sections pertain to a number of issues concerning the mental lexi-
con. Firstly, an attempt has been made to review various definitions of the phe-
nomenon in question from a diachronic perspective. Subsequent sections relate to
the internal organization of the mental lexicon; to be more precise to the actual
number of storage systems. In brief, questions arise how many lexicons are to be
found in the brain and how the semantic and formal (morpho-phonological and
orthographical) components of lexical entries are stored. Are they stored together
in a unitary modality-neutral lexicon or rather separately within two different
modality-specific lexicons, and, if so, are there any direct links between the two?
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
36 2 Modelling the Lexicon: Some General Considerations
The term mental lexicon was introduced by Oldfield in 1966 (Oldfield 1966; in
Singleton 1999) and since then it has been the focus of attention of a number of
psycholinguists all over the world. It has been researched and re-defined from
various perspectives many times. One of the early definitions was proposed by
Fay and Cutler who attempted to describe the mental lexicon in terms of the lexi-
con metaphor as “the listing of words in the head” (1977, p. 509). The evidence
they cite to support their claim demonstrates that the majority of words, excluding
onomatopoeias, are characterized by arbitrary sound-meaning relations. Fay and
Cutler (1977, pp. 508–509) offer the following description of the mental lexicon:
What is this mental dictionary, or lexicon, like? We can conceive of it as similar to a
printed dictionary, that is, as consisting of pairings of meanings with sound representa-
tions. A printed dictionary has listed at each entry a pronunciation of the word and its
definition in terms of other words. In a similar fashion, the mental lexicon must represent
at least some aspects of the meaning of the word, although surely not in the same way as
does a printed dictionary; likewise, it must include information about the pronunciation of
the word although, again, probably not in the same form as an ordinary dictionary.
While some linguists compare the mental lexicon to a written dictionary, others
describe it as a network of interconnected nodes similar to bundles of neurons in
the brain. Aitchison (2003a, p. 248) rightly argues that “the mental lexicon is (…)
concerned above all with links, not locations” and observes that “the lexical con-
nections in the mind are far from what we normally imagine a dictionary or lexi-
con to be”. When a word is activated, other words of similar form (Stamer and
Vitevitch 2012), meaning (Mirman 2011), syntax (Kim and Lai 2012), orthogra-
phy (Carreiras et al. 2013) or emotional content (Bayer et al. 2012) are also acti-
vated, suggesting that the mental lexicon is complex and highly interconnected.
Emmorey and Fromkin (1988) propose to view the mental lexicon as
that component of grammar in which information about individual words and/or mor-
phemes is entered, i.e. what a speaker/hearer of a language knows about the form of the
entry (its phonology), its structured complexity (its morphology), its meaning (its seman-
tic representation), and its combinatorial properties (its syntactic, categorical proper-
ties) (…) also orthographical or spelling representation (Emmorey and Fromkin 1988; in
Gabryś-Barker 2005, p. 38).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.3 Towards the Model of the Mental Lexicon 37
Before anything more is said about the structure of the mental lexicon, it is
imperative to realize that finding common patterns in language errors is believed
to provide valuable information about the nature of the internal lexical storage sys-
tem. Thus, error analysis constitutes a basis and seems to be a perfect source of
data in research on language processing (cf. Fromkin 1973). Admittedly, errors in
any language system have an incalculable explanatory value. The evidence from
word searches and “slips of the tongue”, selection errors known as malapropisms,
but also psycholinguistic experiments and research with aphasic patients, show that
lexical items in the mental lexicon are interconnected in a wide variety of ways.
Fay and Cultler (1977) based their model of the mental lexicon on malaprop-
isms (cf. Vitevitch 1997; Goldrick et al. 2010). These are speech or writing errors
“in which a word similar in sound to the intended one is uttered as in The cold
is being exasperated by the wind instead of The cold is being exacerbated by the
wind” (Aitchison 2003b, p. 71). However, there are three basic conditions an erro-
neous word needs to meet in order to function as a malapropism. Firstly, the mean-
ing of the error and the target word needs to be unrelated. Secondly, erroneous
intrusion should sound similar to the intended word. Thus, using tattoo instead of
book cannot be classified as a malapropism; whereas substituting tattoo for taboo
would be. Lastly, the word becomes a malapropism providing it has the so-called
recognized meaning in the user’s language. Consequently, coining a non-existent
or ungrammatical word by adding some affixes does not make a word a malaprop-
ism. Moreover, Fay and Cutler claim that
(…) the malapropisms, have some interesting properties. First, the target and the error
are of the same grammatical category in 99 % of the cases. Second, the target and the
error frequently have the same number of syllables (87 % agreement in our list). Third,
they almost always have the same stress pattern (98 % agreement) (Fay and Cutler 1977,
pp. 507–508).
On the basis of their findings, Fay and Cutler proposed a model which assumes
that lexical storage is phonologically governed. The mental lexicon is conceived
of as a network which “lists entries that have similar phonological properties near
each other” (Fay and Cutler 1977, p. 512). Recapitulating, words beginning with
the same phoneme are listed together, whereas words sharing the same second
phoneme are grouped in a subcategory of that class and so on. It needs to be added
that Fay and Cutler do not exclude the possibility of arrangement by syntactic cat-
egory. Nevertheless, they do not provide any further details of such a concept.
Admittedly, there are considerable similarities between a traditional dictionary
and the human mental lexicon. They are both organized along some underlying
principles based on the characteristics that words share. Clearly, in the case of a
written dictionary, the basic criterion of organization is orthography. Words in a
book lexicon are always stored in alphabetical order. Consequently, if we want to
look up a word we need to identify its initial letter, find words beginning with that
letter and, finally, again in alphabetical order, exhaust the possibilities until the
right entry has been found. Locating the word enables us to gain access to all the
related data hidden “behind”—the semantic, phonetic, and pragmatic information.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
38 2 Modelling the Lexicon: Some General Considerations
As Bakhtin (1981) formulates it, “every word smells of the context (…) in which
it has lived its intense social life” (Bakhtin 1981; in Gass and Selinker 1994,
p. 276). In communication, language users depend on the contexts in which words
appear to a significant degree, inferring word senses on the basis of linguistic as
well as non-linguistic data, the latter being frequently even more informative.
However influential the lexicon metaphor can be, many cognitive psychologists
and psycholinguists reject it claiming that the mental lexicon is much more than
just a repository of lexical items. The advocates of the cognitive approach posit
that the mental lexicon consists of concepts and their linguistic realizations, both
phonological and orthographic. They conceive of it is as a conceptual system. As
Gabryś-Barker puts it,
A mental lexicon should be seen more as a conceptual system than a pure inventory of
entries, a system which is composed of concepts and their linguistic realisations both pho-
nological and orthographic, and with strong emphasis put on lexical processing (…) that
is to say, access and retrieval as evidence of the working structure of the mental lexicon
(Gabryś-Barker 2005, p. 39).
Notably, the standard position in language processing is that the mental lexicon
is a largely fixed resource, acquired during early development. Although people
can of course add new lexical entries during their adult life, this is generally seen
(2003b, p. 57).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.3 Towards the Model of the Mental Lexicon 39
as a marginal activity. Studies of processing assume that people already know the
language that they use and that there is a clear demarcation between acquisition
and processing. (cf. Aitchison 2003a, 2012; Cutler 2005). In addition, the lexicon
is treated as a store that principally consists of small units (either words or mor-
phemes) and that knowledge of larger units is largely limited to idioms, which are
regarded as fairly peripheral to “core” language processing.
More recently, Pickering and Garrod (2013) proposed an alternative view
of the mental lexicon that is consistent with the Dynamic Systems Theory
(cf. Briggs and Peat 1989). They based their proposition on the evidence from
dialogues which shows that interlocutors make use of fixed or semi-fixed expres-
sions during a particular conversation with meanings that are established through
that conversation. They also argued that language users “routinize” (Pickering
and Garrod 2005, p. 87) these expressions by storing them in the mental lexicon,
normally for that conversation alone. This requires a conception of the lexicon
in which complex expressions (of all kinds, not just established idioms) can be
stored alongside more traditional lexical units. On this view, the lexicon can be
constantly and dynamically updated, and the strict division between acquisition
and adult usage is removed.
The final paragraphs of this section attempt to shed some light on the research
concerning the much debatable problem of the size of the mental lexicon. It is
generally believed that the mental lexicon is comprised of a huge number of lexi-
cal entries; however, its exact size remains undefined. In the early research con-
ducted by Seashore and Eckerson in 1940 (in Aitchison 2012) the number of
words stored in the mental lexicon of an educated adult was estimated at about
150 thousand receptive words with 90 % available for production. A similar study
carried out by Diller in 1978 resulted in an unpredictably high number of about
250 thousand words, whereas the more recent work by Levelt (1989) rated the pro-
ductive vocabulary of an educated adult at no more than 30 thousand word fami-
lies. According to Clark (1993), on the other hand, adult speakers of a language
have at their disposal between 20 and 50 thousand productive words, the amount
of receptive vocabulary being “considerably larger” (1993, p. 13). All things con-
sidered, average educated adult language users have at their disposal a production
vocabulary of between 20 and 50 thousand words and comprehension vocabulary
of between 150 and 250 thousand words.
Why are the research results so diverse? Many linguists postulate that such
sharp differences are connected with the failure to distinguish between produc-
tive and receptive vocabulary. Consequently, different experiments employ either
active vocabulary exclusively or involve both passive and active words. Some
researchers concentrate on active vocabulary (thus achieving lower numbers);
other experimenters use both passive and active words (those used exclusively
for comprehension and those for both comprehension and production). Another
typically cited explanation for such a discrepancy in the results are various, not
infrequently, incompatible methodologies. Nonetheless, whatever the answer to
the question about the amount of mental word-stores, the actual number seems to
exert little impact on the way the lexicon functions.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
40 2 Modelling the Lexicon: Some General Considerations
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.3 Towards the Model of the Mental Lexicon 41
With regard to organization, then, Fay and Cutler (1977) believe that there is one
single mental lexicon for both production and comprehension instead of two sepa-
rate lexicons. This assumption has been based on the analysis of common speech
errors such as malapropisms or slips of the tongue. By contrast, Garman’s model
(1990) accounts for the existence of two separate specialized stores: one for gen-
erating and one for identifying words. Here the evidence supporting this view
comes, above all, from neuropsychological research which has proved a number of
discrepancies between comprehension of spoken and written input and production
of spoken and written output. According to Ellis and Young’s model (1988, 1996),
on the other hand, there is one semantic lexicon incorporating four-modality spe-
cific interconnected sub-lexicons.
The structure of the lexicon is not the only debatable issue concerning the human
word-store. Equally controversial is the matter of the relations within the mental
lexicon. A highly advanced classification of various internal connections occurring
in the mental lexicon was proposed by Levelt (1989), who distinguished between
intrinsic and associative links. Intrinsic relations occur when items are linked
through at least one component of the fourfold information on a word—meaning,
morphology, syntactic category or phonology. Associative relations, on the other
hand, hold between words which show no direct semantic, phonological or mor-
phological links, but which frequently co-occur in speech or writing.
Lexical items can be intrinsically related through their meaning. A word is
linked with its hyperonym (banana—fruit), co-hyponyms (banana—apple), near-
synonyms (wide—broad), antonyms (wide—narrow) etc. All these interrelated
links form a network called a semantic field. Another form of intrinsic links are
morphologically-determined relations between derivatives of one item, which
simultaneously share some semantic features (e.g., govern, government, govern-
mental, governor). Evidence supporting the existence of such types of relations
between individual lexical entries, again, comes from the analysis of speech
errors. Fay and Cutler (1977) and much later Fikkert (2005, 2007) point to yet
another type of intrinsic relation—the one based on phonological features which
may be responsible for substitution errors such as the already discussed mala-
propisms. The authors claim that “words with the same initial or final segments
seem to be connected as they cause errors in speech production such as week for
work” (Fay and Cutler 1977, p. 514). Finally, there is some evidence on syntacti-
cally conditioned connections between entries coming from research with aphasic
patients who have lost access to the entire class of words (cf. Haverkort 2005).
The second type of connections between entries in the mental lexicon are the
associative relations. This kind of link occurs between entries which do not share
any semantic, phonological or morphological features but which tend to co-occur
in language use. The existence of associative relations has been evidenced in a
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
42 2 Modelling the Lexicon: Some General Considerations
One of the most hotly disputed controversies connected with the mental lexicon
seems to be the issue of whether words are stored as whole units or as roots plus
affixes. The following paragraphs address two fundamental questions concerning
the lexical storage of polymorphemic words. Prior to the presentation of two influ-
ential hypotheses seeking to explain the storage of morphologically complex words,
a question addressing the problem of what precisely is stored will be tackled.
The issue of lexical storage is strongly related to the phenomenon of word
primitives which are commonly defined as the smallest meaningful elements
stored in the mental lexicon. For many decades linguists have tried to determine
how words consisting of more than one morpheme (e.g., government) are stored
within the lexicon. Are they stored as independent units, or, as many linguists have
suggested, are complex words decomposed into their constituent elements (e.g.,
govern and ment), which would support the morphemic organization of the lexi-
con? Depending on the perception of word primitives, linguists advocate in favour
of one of the following theories.
The Full Listing Hypothesis was first proposed by Butterworth in 1983 and
since then it has gained a number of supporters (cf. Henderson et al. 1994). In
the light of this theory, derivations are stored, similarly to a written dictionary, as
separate, independent entries (e.g., go and goer are stored as independent units).
Consequently, both for comprehension and production they are accessed sepa-
rately. In the light of more recent studies (cf. Vigliocco and Hartsuiker 2005), the
only advantage of this hypothesis seems to be the so-called access efficiency.
The alternative proposition, known also as the Decompositional Hypothesis,
has gained far more advocates (cf. Levelt 1989; Taft 2004; Frost and Ziegler 2007)
and for this reason the idea of the morphemically-governed organization of the
lexicon will be elaborated further in what constitutes the final paragraphs of the
present section.
In the Decompositional Theory, words are seen as bundles of morphemes,
and since morphemes are believed to be the smallest meaningful units of
4 For
a thorough discussion of this methodology see Gabryś-Barker (2005), Fitzpatrick (2007)
and Roux (2013).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.3 Towards the Model of the Mental Lexicon 43
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
44 2 Modelling the Lexicon: Some General Considerations
of the aurally presented verbs (govern, exist, decide), MacKay noticed that RTs
varied significantly depending on the “complexity” of the derivation. And thus,
government was identified as the fastest item (no phonological changes), exist-
ence was slower (resyllabification), while decision turned out to be the slowest
(two phonetic changes). The interpretation put forward by MacKay strongly sup-
ports the Decompositional Theory. He claims that the results confirm the assump-
tion that subjects are able to make those changes when producing morphologically
complex words, which means that such words are not stored simply as independ-
ent units. However, the popular criticism of the research was linked to the very
form of the experiment. Namely, the participants were explicitly instructed to
derive morphologically complex nouns from a presented list of verbs. The task
itself required derivational processing which may not occur normally. Thus, many
linguists find MacKay’s research unreliable.
A similar lexical decision task paradigm was adopted by Taft and Forster
(1975) and Taft (1981), who worked on the “prefix-stripping” strategy in word rec-
ognition. They concluded that in lexical decision tasks words with prefixes had
greater RTs than words without them. Thus, remind (prefix re-) was identified
faster than relish (“pseudoprefix”). The proposed interpretation of the obtained
results is as follows:
Morphological processor automatically “strips off” anything that looks like a prefix (e.g.,
“RE”), then searches for the base in the lexicon. With words like REMIND, it will find
MIND (real word), but with words like RELISH, will not find *LISH and will have to
restart the search for the whole string (Taft and Forster 1975, pp. 642–643).
It needs to be stressed, however, that the presented results and Taft and Forster’s
interpretation were also rejected by many who, as in the previous case, criti-
cized the methodology of the conducted experiment. The opponents claimed that
although the participants were not instructed to strip off prefixes, “maybe they were
implicitly told this by the kind of word list they got” (cf. Rubin et al. 1979, p. 760).
Aitchison, who is also an ardent advocate of the decompositional approach, to
support her theory uses error analysis of a spoken discourse—he example she gives
is She wash upped the dishes instead of She washed up the dishes (Aitchison 2003a,
p. 65). In her interpretation, the error may be suggestive of the organization of the
internal lexical storage. In other words, the error has been committed since the brain
has accessed the preposition up instead of the verb wash. The researcher believes that
such errors verify the Decompositional Hypothesis and prove the fact that words are
stored as morphemes. To generate this sentence our brain needs to access the verb
wash, the preposition up and the past tense morpheme -ed. If derivations were stored
as in the Full Listing Hypothesis, such an error would not occur—our brain would
store the word washed as a separate item. We would not have to go into the process
of building a word; instead, the already prepared items would wait to be accessed.
Yet another source of evidence supporting the hypothesis under discussion are the
results of experiments with patients suffering from Broca’s aphasia (cf. Tyler et al.
1995). On the basis of these and other findings, models of word recognition were
created which typically include a processing stage in which complex words are split
into their constituent morphemes before meaning-based representations are accessed.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.3 Towards the Model of the Mental Lexicon 45
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
46 2 Modelling the Lexicon: Some General Considerations
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.4 Theories of Semantic Representation 47
Fig. 2.2 A hierarchical network model of semantic information related to animals (adapted from
Collins and Quillian 1969)
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
48 2 Modelling the Lexicon: Some General Considerations
To check their model Collins and Quillian employed sentence verification tasks
(cf. Collins and Quillian 1969, 1970). They assumed that it takes longer to verify a
sentence containing information from the most remote nodes in the hierarchy, e.g.,
A bear is a mammal, than a sentence using information from closer nodes, such as
A bear is an animal, since the lower levels inherit the information from the higher
levels. This kind of familiarity effect has not, however, been confirmed in empirical
research (cf. Table 2.1 below). The model has also been criticized for the invalid-
ity of accommodating the typicality effect which posits that all the words from the
same level of a given hierarchy, e.g., robin, ostrich, canary etc. are to be considered
equal. Hence the relation between robin and bird and ostrich and bird should be
perceived as equal. Nevertheless, the postulate has not been evidenced in sentence
verification tasks. Conversely, the research carried out proved the predictions of the
hierarchical framework inaccurate. Table 2.1 below compiles the basic problems
the model does not account for. For instance, in contrast to Collins and Quillian’s
assumptions rejecting the familiarity effect, familiar words are indeed recognized
faster than unfamiliar words irrespective of their position in the hierarchy.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.4 Theories of Semantic Representation 49
More importantly, the model also seeks to account for the problem of semantic
economy. Whereas the hierarchical model assumed that the word’s semantic prop-
erties were stored, for the reason of economy, at the highest possible nodes thus
eliminating the redundancy, the revised theory has it that certain features which
are typically associated with a given word are stored with the semantic represen-
tation of that word, against cognitive economy, quite redundantly. Collins and
Loftus’ model also encompasses the typicality effect as developed by the prototype
theory (cf. Sect. 2.4.3). Hence the distance between two nodes is conditioned by
the typicality of these words and not by the hierarchy of organization; e.g., the
connection between bird and penguin is weaker than between bird and pigeon. To
test the efficacy of their model, Collins and Loftus employed the semantic priming
paradigm. The obtained results appeared to support the assumption of the auto-
matic spreading activation mechanism to be found in the processing of semantic
representations.
The semantic feature view stands in contrast to the hierarchical network model
(cf. Smith et al. 1974). This kind of approach, also termed the componential
approach, proposes that words can be decomposed into a bundle of primitive
semantic elements. As a result, words similar in meaning share some of their
semantic features known as defining features, but they also incorporate some char-
acteristic features specific only to them. This theory is connected with two influ-
ential categorisation theories to be discussed below.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
50 2 Modelling the Lexicon: Some General Considerations
Aristotle enumerates two aspects of a thing: essence described as “the parts which
are present in such things, limiting them and marking them as individuals, and by
whose destruction the whole is destroyed (…)” (Metaphysics 5.8.3.) and accidents
referred to as “that which attach to something and can be truly asserted, but neither
of necessity nor usually, e.g., if someone in digging a hole for a plant has found
treasure” (Metaphysics 5.30.1.). The above-mentioned aspects can be explained on
the example of the word flower. The essence of a flower is that it is a plant; its col-
our or smell is just accidental and does not influence the judgment of whether the
entity is a flower or not.
However influential the classical theory of categories may be, it has been “reap-
praised in all of the cognitive sciences” (Lakoff 1982, p. 3). In the 1970s a com-
peting theory of natural categorisation was proposed by Eleonor Rosch (1975).5
Since the theory centered on the so-called prototypical members of the group of
possible referents of a given word, it was labeled the prototype theory. It can be
briefly described as “a hypothesis that people understand the meaning of words by
reference to a highly typical example” (Aitchison 2003a, p. 94). In a short time the
theory gained a wide group of supporters, including Bolinger (1977), Lakoff
(1982, 1987), Wierzbicka (1985) or Langacker (1987) and more recently Smith
and Minda (2002) or Taylor (2003). Unlike the purely theoretical argumentation of
the objectivist metaphysics and psychology, the prototype theory is based on
empirical evidence, “experimental results and the interpretation of these results”
(Lakoff 1982, p. 8).
The subsequent paragraphs provide a short presentation of the number of dif-
ferences concerning these two highly influential theories. The first difference to
be discussed is the so-called componential analysis. In the classical view catego-
ries are defined in terms of a conjunction of necessary and sufficient conditions.
Entities can be described in terms of smaller parts—components or features of
binary structure (present [+] or absent [−]). All members of the category have to
share the same necessary and sufficient features (cf. Taylor 1990). It needs to be
noted that categories are homogenous; i.e. that all members have equal status and
5 The theory found its philosophical grounds in the works of Ludwig Wittgenstein (1953).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.4 Theories of Semantic Representation 51
they need to share the same features. Thus, there are no worse or better examples.
In other words, no single cat is more cat-like than others (cf. Lakoff 1987). In the
prototype theory, on the other hand, entities belonging to one category do not have
to, and rarely do, possess the same inventory of features. Wittgenstein (1953) pre-
sents it as follows:
Consider for example the proceedings that we call: ‘games’. I mean board-games, card-
games, ball-games, Olympic games, and so on. What is common to them all? (…) For if
you look at them you will not see something that is common to all, but similarities, rela-
tionships, and a whole series of them at that (Wittgenstein 1953, p. 31).
Thus, it can be concluded that the underlying principle behind the categorisa-
tion is family resemblance (Wittgenstein 1953). However, not all members pos-
sess the same inventory of features still forming one common category. Rosch
and Mervis refer to a category as to “a set of items of the form AB, BC, CD,
DE. That is, each item has at least one, and probably several, elements in com-
mon with one or more other items, but no, or few, elements in common to all
items” (Rosch and Mervis 1975, p. 66). An item classified as belonging to a
given category shares features with a few others, but not necessarily with all,
members of the same category. Categories are not homogenous, which means
that e.g., some birds are more birdy, like the robin, while some are less birdy
like the penguin (cf. Rosch 1975). It could be concluded that the theory
accounts for worse and better members of one category. The most representa-
tive entities for the entire category are called prototypes. Prototypes have a
privileged place in memory as they occupy the central role in the category and,
consequently, are retrieved quicker.
Another discrepancy between the two theories concerns category boundaries.
In the classical view category boundaries are clear-cut and stable. And thus, the
decision whether an entity belongs to the category or not is based on objective
features. Moreover, no factors can influence those categories. As Lakoff points
out, “category boundaries do not vary. Human purposes, features of context, etc.
do not change the category boundaries” (Lakoff 1982, p. 15). Thus, they would
often demand redefinition or the creation of new categories. Internal definition is
the only factor affecting the category, the structure of a category is context inde-
pendent, no subjective factors can affect the category and thus, psychological
factors seem to be unimportant. No matter how humans perceive a given item, it
is categorized regardless of the subjective interpretation. In contrast, the proto-
type theory shows that there are no clear-cut boundaries. Instead, any boundaries
are described as flexible, susceptible to subjective factors such as human pur-
poses. Many experiments proved (cf. Black 1949; cited in Ungerer-Schmidt
1996; Labov 1973) that prototype-based categories merge into each other, and
their boundaries instead of being clear can be described as fuzzy. In his publica-
tion Labov (1973) elaborates on an experiment in which the subjects were asked
to name various containers (e.g., cup, bowl). The results showed that the labels
provided by the participants varied substantially. Furthermore, the same partici-
pants were not consistent in their responses. Labov later concluded that a word
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
52 2 Modelling the Lexicon: Some General Considerations
has its core meaning which is most central and invariant as well as its peripheral
meanings.6 As a consequence, advocates of this theory emphasize that the mean-
ing of a word should be analyzed on a continuum.
To conclude, the objective of the present section was to discuss the two most
influential theories related to lexical meaning. As has been shown, the theories are
very different. Many scholars emphasize the fact that the prototype theory being
based on empirical evidence seems to be far more convincing than the classical
one based on non-empirical speculations (cf. Lakoff 1987). In the following this
section various models of lexical access will be examined.
6 Cf.Kellerman`s famous study of core and peripheral meanings of the German verb brechen
(break) (Kellerman 1978).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.5 Models of Lexical Access in the Mental Lexicon 53
However complex and demanding the research on the mental lexicon might be,
psycholinguistic literature abounds with models of lexical access. There are many
properties according to which the models can be grouped. Firstly, some models
focus on word recognition, others on production. There are also models which try
to combine these two processes. Another distinctive property is the type of search
involved in lexical processing: here serial (indirect) or parallel (direct) models are
distinguished. The serial models have it that words are accessed individually, one
by one, at the phonological, orthographic and semantic levels. The parallel mod-
els, on the other hand, postulate that words are searched simultaneously. A further
property is interactivity—the question of whether lexical information can travel
backwards and forwards between different levels of lexical representation and
affect their retrieval.
What differentiates the models to be examined in the paragraphs to follow is
the sequence of interaction; a property which divides the models into direct and
indirect ones (cf. Garman 1990, p. 260). The first category of models distinguished
by Garman are indirect access models, which depict lexical processing as “look-
ing up a word in a dictionary” or “finding a word in a library” (Singleton 2000,
p. 170). The indirect access models, also known as multi-step models, are predi-
cated on two-stage access “via a search procedure and then a retrieval procedure”
(Singleton 1999, p. 84). A much-debated representative of the indirect kind of
model to be described below is Forster’s serial search model functioning within a
modular system paradigm.
When referring to direct access models Garman uses the metaphor of a “word-
processing package which allows items stored by name to be accessed simply by
the typing in of as many letters as are sufficient to distinguish the relevant name
from all other stored names” (Singleton 2000, p. 170). In other words, direct mod-
els view lexical processing as a one-stage phenomenon. Two oft-cited representa-
tives of the direct type of model to be discussed below are the logogen model and
the cohort model.
The best known and one of the most influential indirect models is Forster’s auton-
omous search model (Forster 1976; Murray and Forster 2004). The processes
of access and retrieval described in this model resemble looking up a word in a
written dictionary or looking for a book in a library, the only difference being the
organizing principle, which in the case of a written dictionary is alphabetically-
governed, while in the mental lexicon it is claimed to be frequency-dependent.
This is how Garman summarizes Forster’s suggestion:
We enter, looking for a particular book; we do not go straight to the main shelves where
the books are located, since there are simply too many of them to permit efficient search
of them in this direct fashion. So we go instead to the catalogue. Searching through the
catalogue, we find something that matches what we are looking for; but we retrieve from
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
54 2 Modelling the Lexicon: Some General Considerations
Fig. 2.4 Forster’s serial search model of lexical access (adaptation based on Foster 1976)
this stage of the process, not the book itself, but an abstract location marker, telling us
where to find the book on the shelves. Armed with this, we implement the second stage
of the process, by using the marker to guide us to the right book on the shelves (Garman
1990, pp. 266–267).
A similar metaphor has been adopted by Singleton, who compares the first stage to
finding the right page in a dictionary, the major difference between these two pro-
cesses being the aforementioned principle behind the organization of the entries.
The moment the “page” has been found (on the basis of the initial properties of
the signal), the search goes on through the “page” governed by word frequency.
Once the abstract location marker has been found, the second stage starts. Forster
assumes that lexical entries are searched sequentially until the appropriate word
is selected and believes that the mental lexicon consists of two levels: one con-
taining access files and the other the master file (the lexicon proper; cf. Forster
1976; Fig. 2.4). There are two stages of word processing. In the first stage, follow-
ing perceptual processing, the serial search of access files starts. The only infor-
mation on a word available in the access files is its address in the master file. To
put it differentially, access files comprise the stimulus features of a word, i.e. its
access code and the pointer to the matching entry in the master file. The master
file includes all the information on a given word—phonological, orthographic,
morphological, semantic and syntactic data. It needs to be noted that not only is
the master file a complete representation of each individual lexical item, but it also
includes cross-references between all the items stored in the master file, thus pro-
viding for the semantic priming effect. In order to cater for different modalities,
through which lexical items are perceived and generated, and two directions of
lexical access Forster proposed three separate access files which organize words
either by orthographic, phonological or syntactico-semantic properties and are
linked with the master file by pointers. These discrete operating subsystems pro-
cess lexical information independently of one another. A schematic visualization
of the model has been presented in Fig. 2.4.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.5 Models of Lexical Access in the Mental Lexicon 55
Garman (1990) notes that Forster’s search model incorporates two key features
a good model of the mental lexicon should have. It is characterized by the diver-
sity of access and, at the same time, the unity of storage. Whatever the channel of
communication, it is always the same entry in the master file. Access to each word
depending on the channel leads always through a suitable access file:
If one is listening to speech, one processes each spoken word by going first to the pho-
nological access file; if one is reading written language one goes first to the orthographic
access file; and if one is producing language on the basis of particular meaning intentions,
one goes first to the syntactic/semantic access file. The access file in question then facili-
tates access to the master files (Singleton 2000, p. 174).
For convenience, the access files are divided into separate bins based on the initial
sound or letter. The words in a bin are arranged in a descending order of frequency
so that more frequent words can be searched faster and matched with the acoustic
string prior to low frequency words. In this way, Forster eventually managed to
accommodate the frequency effect within his model.7 The effect has been con-
firmed by a substantial body of evidence from lexical decision tasks where high
frequency words were identified faster than low frequency items. The model also
includes the lexicality effect assuming that there occurs exhaustive search for non-
words and terminating one for actual words. Before rejecting a nonword the brain
needs to search through the entire master file only to find an empty entry. This has
been confirmed by numerous lexical decision tasks which show that it takes about
150 ms longer to identify nonwords than actual words.8
Not only does the model account for the frequency effect and the lexicality effect,
but it is capable of accommodating the priming effect as well. Forster’s model is
not interactive in that it does not allow for the cross-referencing of access files and
the master file. Words in the master file are accessed only through one file at a time.
However, once an entry in the master file has been accessed, cross-references are
observed. Thus, the model is able to accommodate the effect of semantic priming.
If an individual sees the target word for doctor and then subsequently is shown the
word nurse, the response time for the latter word is rightly expected to decrease.
However detailed the model seems to be, there is still a substantial number of
controversies it has been unable to end. Firstly, the model faces the problem
of capacity limitations. The evidence from lexical decision tasks supports the
idea of empty entries for nonwords, which, if really there, would occupy a lot
of space redundantly. Secondly, speech seems to be much too rapid to accept the
idea that words are searched sequentially; the model allows for only one entry to
be searched and matched with the input at a time. Another repeatedly criticized
issue is the fact that the model does not allow for the influence of context on the
process of recognition. It also does not give an account of form-based priming
7 The original version of the model which presented it as a direct access model (cf. decision
trees; Forster 1976, p. 258) failed to incorporate the results supporting both the lexicality and the
frequency effect and proved quite to the contrary.
8 Again this effect was impossible to implement within the original version of the model.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
56 2 Modelling the Lexicon: Some General Considerations
effect and it cannot explain the role of similarity neighbourhood. Finally, the
model does not explain the influence of context on speech production (a phenom-
enon confirmed by the repetition priming effect). Due to the above limitations
the early version of the model underwent extensive changes (cf. Forster 1989).
For instance, in the revised version, Forster introduced a discrete comparator
for each lexical entry, thus solving the problem of limited capacity (Murray and
Foster 2004). He also proposed various models of activity among lexical entries.
However, it seems that the presented changes have actually transformed the
model in the direction of spreading activation models.
The logogen model, in contrast to its serial search equivalent, assumes one-stage
parallel processing. Initially, the model was proposed by the British psychologist
John Morton in 1969 to account for visual word recognition. Only later was it
developed and revised to incorporate both written word recognition and word
selection in speech production. The model comprises three elements: the logogen
system, the cognitive system and the response buffer. However, its key feature is
the logogen system which is defined to be a part of the nervous system responsible
for lexical processing (in the initial version of the model it was described as a neu-
ral unit, to be later changed to the more technical term logogen9). According to
Singleton, the logogen system is “a set of mechanisms (…) specialized for collect-
ing perceptual information and semantic evidence concerning the presence of
words to which the logogens correspond” (Singleton 1999, p. 86), whereas
Coltheart et al. (2001) describe it as an “information-gathering” device.
Initially, Morton postulated that there was a unitary logogen system, but due
to some empirical evidence he revised the idea and divided the system into three
parts. He distinguished two specialized input logogen systems: visual and audi-
tory and the output logogen system (cf. Morton and Paterson 1980). It has to be
stressed that none of these units includes semantic information about words. This
information is stored in the cognitive system, which includes “a collection of
semantic information of various kinds” (Singleton 1999, p. 86). The system may,
but does not have to, be incorporated in the lexicon itself. What merits special con-
sideration, however, is that meaning is not stored as a single unit for each word;
instead, it is computed when required.
In the logogen system every single item is represented by the corresponding
logogen which comprises the word’s features (phonological and orthographic
characteristics). The moment the acoustic or visual input reaches the logogen, it
is changed into appropriate phonological or orthographic representation which
9 logogen, from Greek logos, “word” and Greek and Latin gen—“birth”; “to bring to life” (after
Singleton 2000, p. 171).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.5 Models of Lexical Access in the Mental Lexicon 57
launches the process of word finding. The next step is passing the information
to the cognitive system which specifies its semantic and conceptual aspects, and
finally to the logogen output system. It needs to be stressed that the links between
the elements of the model are bidirectional.
It has to be noted that one of the key features of the model is the so-called
threshold level. Each logogen has a “resting” threshold level. Once sufficient evi-
dence has been introduced into the system, the threshold level is reached and the
word is activated. This means that, e.g., in the case of a written word, even before
all the letters are identified, the word can already be recognized and its code sent
to the cognitive system. Clearly, threshold levels differ in value depending on the
words’ frequency. Thus a high frequency word has a much lower threshold than
a low frequency word and less activation will be needed to “fire” and thus access
this word (cf. Harley 2008). In this way the model accounts for the frequency
effects of words in a different way, by using the activation and raising of potentials
within different words.
Summing up, in this model word recognition is seen as a process of accumulat-
ing sufficient information to ultimately access a given word. Once enough infor-
mation has been gathered, the logogen’s threshold is exceeded, the code is passed
to the cognitive system and to the suitable output logogen. The key features of the
model are: directionality of access—every word has its logogen, interactivity—it
allows for the interaction of semantic and perceptual aspects, and parallel process-
ing—incoming information is checked against logogens when it reaches threshold.
The model is versatile in that it accounts for both visual and auditory processing.
Nevertheless, it is perceived as very complex and difficult to test experimentally. It
also leaves many empirical findings unaccounted for, one of them being the effect
of neighbourhood size.
The cohort model is another variant of the direct access model. It was first pro-
posed by the British psycholinguist William Marslen-Wilson (1973, 1975) and
later revised many times to incorporate new findings of psycholinguistic research
(Marslen-Wilson and Welsh 1978; Marslen-Wilson and Warren 1994; Marslen-
Wilson 2007). The model focuses on explaining the process of spoken word recog-
nition and does not account for other aspects of lexical access, namely visual word
recognition and word production. The model is based on the assumption that
words are recognized by their onsets in the left-to-right fashion of processing.
Once the initial segments of the word are uttered and received, all potential lexical
candidates commencing with that very sound(s) are activated and form an initial
cohort. This assumption is supported by the tip-of-the-tongue phenomenon (TOT),
according to which lexical access is possible when the beginning sounds of a word
become accessible (Biedermann et al. 2008). Spoken word recognition is assumed
to constitute three stages: access, selection and integration. In the access stage the
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
58 2 Modelling the Lexicon: Some General Considerations
10 To recognize a monosyllabic word it takes about 300 ms from the word onset and about
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.5 Models of Lexical Access in the Mental Lexicon 59
information has no impact on the selection of the word-initial cohort, (…) once the
cohort has been established, word candidates which are inconsistent with the con-
text can begin to be deactivated” (Singleton 1999, p. 94).
What is interesting to note is that whereas the previously described logo-
gen model allowed for various levels of activation, the early version of the cohort
model, on the contrary, postulated the existence of the binary membership. It
asserted that an item is either active (on) or not (off). The first alternative refers
to the situation when a word still belongs to a cohort of word-candidates, the lat-
ter describing the situation when the word has already been eliminated from the
cohort. Ultimately, later versions of the model rejected the binary membership
and postulated the gradual membership. It was suggested that words which do not
receive further verification from the incoming acoustic representation have their
level of activation gradually lowered; they cannot, however, be eliminated from
the cohort. Conversely, they can be activated again if an appropriate signal occurs.
Consequently, in the more recent versions of the model (1990, 1993) the importance
of input goes beyond the uniqueness point and the deactivation of word candidates
is reversible; the alterations which have made the model maximally efficient.
Finally, it needs to be stressed that however controversial the model might be,
there is still a fair amount of experimental evidence to be found in favour of its
main assumptions. Singleton (2000), for instance, refers to the recognition of non-
words. The recognition of nonwords is shorter, he says, in those cases where rec-
ognition points come early in words. As Singleton reports, “the more contextually
predictable a word is, the shorter the sequence of sounds required to reduce the
cohort to a sole candidate” (Singleton 1999, p. 95). The recognition time is much
longer when recognition points appear later within a word. On the other hand, the
model is still criticized, the basic criticism being connected with the fact that the
model accounts for only one type of modality and fails to explain effects of fre-
quency or neighbourhood density. Additionally, some researchers hold it as highly
unlikely that we recognize words on the basis of “the noisy and ambiguous acoustic
signal which is speech” (Marcus and Frauenfelder 1985, p. 164; in Garman 1990).
2.5.4 Computational Models
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
60 2 Modelling the Lexicon: Some General Considerations
be going on in the boxes and later successfully work out differential predictions of
the models (cf. Norris 2005). For the reasons indicated above, there is a common
agreement among psycholinguists that computational models are to be preferred
over traditional “box-and-arrow” models. However, some obvious limitations all
the current models share, the main being their focus on a single domain of behav-
iour, should not be left unnoticed. Indeed, Norris raises a valid point when he
remarks that there is still a need for more integrated theories of word recognition
(2013, p. 523).
Table 2.2 presents a selection of the most influential computational models of
visual word recognition and points to the basic phenomena the models have been
developed to explicate. As for the modelling style of framework within which the
models have been created, the most influential style of computational models are
connectionist models, the earliest example of which is the Interactive Activation
model (IA),11 first proposed by McClelland and Rumelhart in 1981. This type of
modelling has for many years been favoured among researchers mainly due to the
fact that it is relatively “brain like” (Clark 1993) and relatively easy to understand.
An alternative style of modelling—mathematical or computational one—exploits
computational procedures or mathematical formulae. It needs to be noted,
11 McClelland and Rumelhart’s model will be discussed in details in Sect. 2.6.2 below.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.5 Models of Lexical Access in the Mental Lexicon 61
however, that the interactive activation model (McClelland and Rumelhart 1981),
the Spatial Coding Model (Davis 2010) or the dual-route cascaded model
(Coltheart et al. 2001) which are typically visualized as connectionist models can
also be expressed mathematically (cf. Norris 2013).
This section provides a contrastive description of the two hypotheses on the lin-
guistic storage (the modularity hypothesis and connectionism), with particular
emphasis on their strengths and weakness.
The modular view has it that the mind is “divided into separate compartments,
separate modules, each responsible for some aspect of mental life” (Cook and
Newson 1996, p. 31). The modularists claim that linguistic meaning is clearly
separated from other varieties of meaning and is represented and processed within
the language module (cf. Emmorey and Fromkin 1988). The proposed processing
is sequential (i.e. one thing at a time—an assumption which makes the process-
ing slow), symbolic (i.e. one token equals one concept) and procedural (linguistic
behaviour is governed by rules). The basic problem with this theory, however, is
its inflexibility.
Cognitive theories, which are commonly taken to be antipathetic to the modular
view, adopt the analogy of brain-style neuronal interactions and depict the mind
as a single system—an interactive network. They describe linguistic processing in
terms of connection strength rather than rules or patterns. It needs to be noted,
however, that in the recent decades the most current models have sought to com-
bine both the modular computational and the connectionist theory (cf. Dell 1988).
Customarily, the origins of the modularity theory can be traced as early as in the
18th century, when a German anatomist Franz Josef Gall “developed the view that
each intellectual and behavioural attribute was controlled by a specific location
in the human brain” (Singleton 1999, p. 111). The current version of the hypoth-
esis became one of the most influential cognitive perspectives of the late 1960s.
The major proponents of this modular view of the mind are theoretical linguist
Noam Chomsky (1988) and psycholinguist Jerry Fodor (1983, 1989). Whereas
Chomsky’s interest in modularity is related exclusively to language acquisition
processes, Fodor’s work concentrates on aspects that are processing-oriented.
Since the focus of the present chapter is centered around issues relating to lan-
guage processing, only the Fodorian perspective is discussed.
The modularity hypothesis, according to Fodor (1983), postulates that “the
entire language faculty is a fully autonomous module [comprising] a number of
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
62 2 Modelling the Lexicon: Some General Considerations
12 This opinion does not equal with the claim that the language module has absolutely no con-
nection with other cognitive operations (cf. Aitchison 2003a).
13 The detailed description of the features has been based on Fodor (1983, 1985).
14 It is assumed that learned systems do not display these particular regularities (cf. Singleton
2000).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.6 Views on Language Processing 63
To conclude, from the presented evidence it can be inferred that in the Fodorian
model lexical knowledge is represented in the network of interconnected nodes.
It is perceived as a central part of a larger system which operates independently
from other systems. The underlying assumption posits that the mind is modular
and comprises special-purpose perceptual processors called modules.
15 It seems justified to emphasize that Fodor’s description of interlexical excitation shows direct
similarities with some claims of spreading activation theory (cf. Sect. 2.3.2).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
64 2 Modelling the Lexicon: Some General Considerations
2.6.2 Connectionism
The connectionist theory dates back to the works of McCulloch and Pitts, who in
the 1940s presented the first mathematical model describing the functioning of a
neuron (McCulloch and Pitts 1943; after Singleton 2000). However, the first influ-
ential models of the lexical processing within the connectionist paradigm were
proposed only after a long period of silence in the 1970s and 1980s. To explain
lexical processing, connectionism adopts the “brain metaphor” (Rumelhart and
McClelland 1986, p. 75), which is based on neurophysiological activity in the
brain. And thus, the basic feature of all connectionist models, known also as inter-
active network models, is an analogy between the human brain and the connection
of neurons. They all depict the mental lexicon as a network of nodes which have
various degrees of activation and perceive lexical processing as activation spread-
ing along a network of interconnected units. In fact, one of the major interests of
connectionism is to compute the algorithm that reflects how activation spreads
around the network and triggers individual nodes.
It is worth noting that the connectionist approach to lexical processing belongs
in a much broader parallel processing perspective which stands in stark contrast to
the aforementioned modular theory deriving from the serial processing tradition
(cf. Sect. 2.4.1). Firstly, whereas the parallel processing perspective advocates the
independence of processing operations (i.e. that many activated items can be han-
dled simultaneously), the serial perspective describes lexical processing as organ-
ized in stages (i.e. that activated items can be handled in the one-at-a-time order).
Secondly, the connectionist models call into question the Chomskyan/Fodorian
perception of language and the mind by rejecting the so-called symbolic para-
digm which posits that “mental operations involve the manipulation of symbols”
(Singleton 2000, p. 179). Instead, the connectionist paradigm seeks to describe
information processing in terms of the strength of connections between units in a
network rather than in terms of rules. As Singleton puts it, “it is not patterns that
are stored (…) but rather the connection strengths between elements at a much
lower level that allow these patterns to be recreated” (Singleton 2000, p. 180).
Proponents of parallel processing models stress that an obvious advantage of
these models over serial search models is that they can explain the enormous com-
plexity of the information processing in the brain. On the other hand, connection-
ist models are frequently criticized for their inability to account for syntactic and
semantic aspects of language processing. Indeed, the current versions concentrate
mainly on the lexical level.
In the subsequent paragraphs two approaches representative of the connection-
ist tradition will be briefly outlined: localist connectionism and distributed con-
nectionism, also referred to as parallel distributed processing (PDP).16 In the
16 Both localist connectionism and distributed connectionism approach will be further discussed
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
2.6 Views on Language Processing 65
localist models each item is represented by a single unit (node) which is symbolic
in nature and has a functional value (cf. McClelland and Rumelhart 1981;
Stemberger 1992; Dell 1988; Roelofs 1992, 1999). In contrast, distributed con-
nectionism models assume the existence of distributed representations which are
processed in parallel and where units do not bear any functional value. The most
representative example of the latter type is the parallel distributed processing
model designed by Seidenberg and McClelland (1989). The basic difference
between the localist and distributed connectionism models lies in the representa-
tion of words. In contrast to the localist connectionism models which assume
one-to-one correspondence of lexical units and their mental representations, in
the PDP models “knowledge of words is embedded in a set of weights on connec-
tions between processing units encoding orthographic, phonological, and seman-
tic properties of words, and the correlations between those properties”
(Seidenberg and McClelland 1989, p. 560). In these models there are no entries,
bins or logogens. The models do not accommodate the mental lexicon in the tra-
ditional sense, nor do they account for the traditional lexical access. As Singleton
observes, “different portions of information are simultaneously processed inde-
pendently of one another (‘in parallel’) on different levels (‘distributed’)”
(Singleton 2000, p. 179).
One of the first parallel processing models (pre-connectionist model) is
the Interactive Activation Model which was put forward by McClelland and
Rumelhart (1981). The model postulates that perceptual processing takes place
simultaneously at a more than one level (parallel processing). McClelland and
Rumelhart (1981) distinguished the feature level, the letter level, the word level
and higher levels responsible for top-down input to the word level. The model
is not only parallel but it also accounts for interactive processing, which means
that in the process of word comprehension there are two co-occurring factors,
namely lexical knowledge and the incoming information from the perceived
stimulus. Thus, the processing is both top-down (conceptually driven) and
bottom-up (data-driven) at the same time. As for the representation of words,
the model posits that lexical units have their corresponding nodes which are
stored in levels (localist tradition) and are linked with other nodes. It needs to
be stressed that nodes are connected bidirectionally with other nodes at different
levels of the network.
The model also accommodates the frequency effect. Nodes have their acti-
vation levels which are modified by the amount of activation they receive from
other nodes (neighbours). Nodes corresponding to frequently or recently used
lexical items have a lower level of activation and thus are selected faster than
nodes which represent words of lower frequency. Communication between
nodes is possible due to the spreading activation mechanism. McClelland and
Rumelhart (1981) posit that there are two types of connections within the sys-
tem of nodes: excitatory and inhibitory ones. The former is responsible for
increasing the activation level of connected nodes, the latter for decreasing
the level.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
66 2 Modelling the Lexicon: Some General Considerations
2.7 Conclusion
This chapter has been meant to serve as a background for a more thorough con-
sideration of various theoretical issues and empirical investigations concerning the
multilingual context which will be discussed in the subsequent chapter of the pre-
sent work. The underlying assumption has been that presenting and explicating the
most significant concepts pertinent to the modelling of the mental lexicon would
enable the reader to interpret and evaluate the research projects whose design and
findings are presented in Chaps. 4 and 5. In order to do so, an attempt has been
made to review the classical theories and models concerning the organization of
the monolingual mental lexicon. The chapter started with a short discussion on the
internal structure of a lexical entry. Subsequently, it provided a brief overview of
various definitions of the mental lexicon as an entity, from those depicting it as
a dictionary to the ones which view it as a network. Apart from tackling termi-
nological issues, the present chapter dealt with the phenomenon of the represen-
tation of meaning in the human mind. The discussion revolved around the most
influential models concerning the internal structure of the monolingual lexicon, as
well as the numerous models of lexical processing. With regard to lexical process-
ing, the chapter summarized and assessed the better-known psycholinguistic mod-
els concerning the organization and functioning of the mental lexicon (Forster’s
lexical search model, Morton’s logogen model, Marslen-Wilson’s cohort model),
and gave a brief account of most recent computational models of visual word
recognition. Finally, special consideration was given to modular and connection-
ist perspectives on lexical processing. It needs to be noted, however, that despite
its substantial size, the section on lexical processing has not exhausted the vast
scope of visual word recognition study. In particular, a lot more can be said about
the achievements of computational modelling (cf. Norris 2005, 2013). However,
as stated above, the chapter has been meant to serve as a background for a discus-
sion of the multilingual mental lexicon, which constitutes the main concern of the
present work.
Now that the main issues concerning the concept of the mental lexicon have
been delineated, it is time to outline the most prominent hypotheses and models
of language storage, processing and retrieval in relation to the mental lexicon of
multilingual speakers. More precisely, the following chapter will be devoted to the
presentation and discussion of the issues of single vs. multiple lexicons and lan-
guage selective vs. language nonselective lexical access.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Chapter 3
Modelling Multilingual Representation
and Processing
3.1 Introduction
In the case of multilingual speakers, who are in the focus of attention of the
present work, the complexity involved in L1 lexical storage and processing, dis-
cussed in the previous chapter, is further multiplied by the complications added by
other lexical systems, those of L2, L3, Ln. Probably, the most enduring questions
in psycholinguistic research on bilingual and, more recently, multilingual mental
lexicon that have been repeatedly posed by researchers since the 1960s (that is,
since the beginning of extensive psycholinguistic research on bilingual and, later,
multilingual mental representation and processing), concern separate/integrated
storage and selective/non-selective access and as such, are both linked to the dis-
cussion about dependence/interdependence dichotomy. As Kroll and Tokowicz
rightly observe, “life experience offers ample support for each alternative” (2005,
p. 531). On the one hand, multilinguals are able to function in each of their lan-
guages independently, on the other, instances of code-switching are not uncommon
in their discourse. And thus, two alternative theories concerning bilingual lexical
representation have been proposed: one is the common memory theory (commonly
referred to as one-store hypothesis), which postulates a single integrated memory
system for both languages. The other, known as the multiple-memory theory (two-
store hypothesis), claims that words from each language are represented separately.
Whereas the results of tests involving word association or comprehension of
code-switched sentences speak in favor of a two-store hypothesis, studies on
transfer and interference support the hypothesis of one store.1 These two theoreti-
cal models, however, appeared too simple to explain all the experimental findings
(e.g., the result that cognates were recognized with the same speed in both
1 A review of these early tests and their results is proposed by Kroll and De Groot (1997).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
68 3 Modelling Multilingual Representation and Processing
A key issue that arises with reference to bilingual and, by extension, multilingual
storage is whether language systems are stored together or separately. Weinreich’s
compound-coordinate distinction was the first statement of the shared and separate
store hypotheses of bilingual memory which provided the basis for the subsequent
research in this field. On the basis of the Saussurean (1959) distinction between the
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.2 Mental Representation of Bilingualism 69
Fig. 3.1 Weinreich’s
three types of bilingual
representation: a coordinate,
b compound, c subordinate
(adaptation based on
Weinreich 1953)
signifier and the signified, Weinreich proposed three types of bilingual lexical organi-
sation2 i.e., coordinate, compound and subordinate (cf. Weinreich 1953; cf. Fig. 3.1).
In 1963, Kolers related the models of the organisation of bilingual memory to
models of the nature of representation. He adapted Weinreich’s compound sys-
tem and labelled it the shared-store model (Kolers 1963). In this model, all lexical
knowledge is stored in one lexicon, concepts are represented in some sort of non-
linguistic, abstract form. Kolers also described what was essentially the coordinate
system and labelled this the separate-store model. Here, the bilingual’s representa-
tions of words encoded in a specific language are stored in a form that in some
way is specific to that language.
A few decades later, Obler and Gjerlow (1999, p. 12) proposed a range of dif-
ferent models representative of possible connections existing in the internal lexi-
cons of multilingual language users. They distinguished between separationist
(Fig. 3.2a), interdependent (Fig. 3.2b, c), or partial overlap views (Fig. 3.2d) of
the multilingual mental lexicon and their conceptual basis. The model depicted in
Fig. 3.2a assumes the independence of particular lexicons (cf. Weinreich’s coor-
dinate model in Fig. 3.1a), whereas Fig. 3.2b presumes their interdependence (cf.
Weinreich’s subordinate model in Fig. 3.1c). The latter model resembles a bilin-
gual dictionary in which the meanings of lexical items from L2 are direct trans-
lations of L1 words. Interestingly, the equivalence of the words from subsequent
2 A more detailed description of the types of lexical organisation described by Weinreich will be
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
70 3 Modelling Multilingual Representation and Processing
lexicon (in this case L3) seems to be received through the subordinate language,
most likely the one of higher proficiency.
Further, Fig. 3.2c presents a fully integrationist model (cf. the concept of multicom-
petence discussed in Cook 1993, 1996, 2007a, b) where there is one conceptual store
with different language forms. The model assumes the cross-consultation between the
languages (cf. Singleton 1999) and resembles Kroll and Stewart’s Revised Hierarchical
Model (1994), discussed in detail in Sect. 3.2.3.2. Finally, Fig. 3.3d shows partial over-
lap of particular lexicons in their relationship to a shared conceptual store, thus refer-
ring to Kroll and De Groot’s Distributed Feature Model (1997; cf. Sect. 3.2.3.1).
Clearly, the issue of bilingual and, more recently, multilingual storage is the
subject of a multitude of empirical studies on the basis of which many models and
hypotheses have been developed. Some of them argue for the separate and others
for the shared organisation of the multilingual mental lexicon. Successful, in part,
seems to be an attempt to resolve this conflict made by those psycholinguists who
opt for the mixed structure of the multilingual store. The two following sections
provide a critical overview of studies and arguments supporting either the separa-
tist or integrationist view of the mental representation of multilingualism, whereas
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.2 Mental Representation of Bilingualism 71
Fig. 3.3 The distributed feature model (adaptation based on Van Hell and De Groot 1998a)
Sect. 3.2.3 will be devoted to the more detailed presentation and discussion of the
hypotheses and models supporting the mixed-storage view.
Grainger and Beauvillain (1987) provided experimental support for the model using
a lexical decision task3 where context effects were investigated. In their study sub-
jects were asked to decide if a given letter string is French or English. Lists of letter
strings were either mixed (included both English and French items) or blocked for
language (with words either exclusively English or French). In the study it took
longer to decide about language affiliation of the stimuli if they were mixed across
languages rather than when they were presented in monolingual lists, which has
been interpreted as a fact supporting the separatist view. More interestingly, reaction
times were the longest for those items which lacked language specific orthographic
3 Lexical decision tasks as well as other tasks typically used in the research on multilingual stor-
age and processing are further delineated in Chap. 4.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
72 3 Modelling Multilingual Representation and Processing
patterns and were presented immediately after items from the other language. By
way of example, it took longer for the participants to recognize the English word
time when it was preceded by the French lire than when it was preceded by the
English word life. This effect of the preceding word was absent if the target was
white, whose spelling is illegal in French (cf. Grainger and Beauvillain 1987).
Another pool of evidence for the separation of language-specific lexicons comes
from the modularity hypothesis (cf. Chomsky 1988 after Singleton 1999; Fodor
1983; Sect. 2.6.1); according to which “in the case of post-pubertal L2 learner, L1
and L2 lexical operations proceed in absolute isolation from each other” (Singleton
1999, p. 167). Also, neurological research studies applying imaging techniques
prove that different brain areas are activated when the informant performs in L1 and
when he or she switches to L2 (cf. Arabski 2004; Field 2004; Kovelman et al. 2008).
Substantial arguments supporting the separation theory have also been pro-
vided by research studies with aphasic patients (e.g., Albert and Obler 1978;
Arabski 2004; Cutler 2005; Ibrahim 2009; Paradis 2004, 2009). By way of exam-
ple, Edwards (1994) reports on bilingual patients who lose both their languages
and then recover them one by one, not necessarily in the order of their acquisition,
which leads him to a conclusion that recovery patterns for each of the languages
may not be identical. In a similar vein, Smith (1997) summarizes her findings col-
lected in language impairment studies concluding that
a.) For some bilinguals, there has been a reported loss of one language, with no impair-
ment in functioning in the second language; b.) different types of aphasia have been found
to occur in the two languages of a bilingual; and c.) the recovery patterns in aphasia often
differ for each of the several languages known to the patient (Smith 1997, p. 149).
It needs to be stressed, however, that Smith (1997) does not postulate a neuroana-
tomic separation of the systems but proposes a difference in the levels of activation
of the two lexicons.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.2 Mental Representation of Bilingualism 73
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
74 3 Modelling Multilingual Representation and Processing
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.2 Mental Representation of Bilingualism 75
are learned via translation equivalents and the neural connections between these
equivalents are more strongly developed, whereas in uncontrolled acquisition situ-
ations the links between concept and lexical lemma might be stronger.
The significance of bilingual participants’ proficiency level was further recog-
nized by van Hell and Dijkstra who aptly state that the relative fluency in their two
languages
will affect the bilingual’s sensitivity to L1 interference when he/she is processing in L2,
and the sensitivity to L2 interference when processing in L1. The underlying rationale
here is that less activation is needed to recognize words that are used relatively frequently,
as are words in a language in which the speaker is relatively proficient (2002, p. 782).
Accordingly, it can be concluded that “bilingual memory representation is not
static but rather a dynamic system that can be influenced by language usage”
(Heredia 2008, p. 57).
Other studies that have indicated a proficiency effect on multilingual lexi-
cal organisation are reviewed by Christoffels et al. (2006) and include those by De
Groot and Comijs (1995), Kroll and Curley (1988) and Chen and Leung (1989).
They are all focused on adult bilinguals of different levels of proficiency and all
use picture-naming and word-translation methodologies. It needs to be noted, how-
ever, that they all adopt various proficiency measures, from the number of years of
instruction (Cieślicka 2000; van Hell and Dijkstra 2002) and self-assessment of the
language skills (Haigh and Jared 2007) to on-line vocabulary tests (Lemhöfer et al.
2004), and tend to allocate bilinguals into separate proficiency groups rather than to
investigate gradual changes that appear along with the increasing L2 (an important
exception being the research by Chambers and Cooke 2009). A further complicating
factor seems to be the fact that bilinguals from different proficiency groups may also
belong to different age groups and thus their bilingual representation may also be
dynamic. Here a vast number of current empirical studies pointing to the changing
patterns of bilingual representation of early and late bilinguals comes from the field
of neurolinguistics (cf. Wattendorf et al. 2001; Wartenburger et al. 2003; Ullman
2001; Kroll and Tokowicz 2005; Miozzo et al. 2010).
Another aspect concerning the organization of the multilingual lexicon, also
underpinned by neuroimages of the multilingual brain, is a great inter-subject vari-
ability between speakers. This is also reflected in the dynamic systems theory, as
suggested by de Bot et al. (2007). The proposed framework implies that all vari-
ables are interrelated so that changes in one variable will have an impact on all
other variables in the system. And thus, a major question raised by de Bot and his
colleagues (2007) is whether, as commonly assumed, individuals really have simi-
lar L1 systems. Further, in their view it is also possible that in multilingual repre-
sentation there might be different systems according to the individual’s linguistic
biography and acquisition contexts: different numbers of subsystems and different
modes of access (cf. de Bot et al. 2007).
Indeed, the question of mental representation of multilingualism cannot be
answered for all speakers in the same way. This contention has been confirmed
by Cieślicka (2000) who conducted the lexical decision task experiment on a
group of fluent and non-fluent Polish-English bilingual language users who were
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
76 3 Modelling Multilingual Representation and Processing
The Distributed Feature Model assumes language independent lexical and concep-
tual levels of representation with a language specific lemma level. It accounts for
the results obtained by a number of researchers in a variety of experiments such
as translation production, translation recognition, lexical decision, primed lexi-
cal decision, Stroop-type picture-word production and word association or (cf.
Grainger and Beauvillain 1987; Altarriba and Mathis 1997; Costa and Caramazza
1999; De Groot 1992; De Groot et al. 1994; Van Hell and De Groot 1998a, b).
Interestingly, all these experiments proved that it is not pure language mixing that
increases the processing time but the extent to which stimuli used in experiments
activate both languages, which points to the influence of task dependency on lexi-
cal processing.
Within the early version of the model, De Groot (1992, p. 1011) assumed
different storage patterns only at the semantic/conceptual level and proposed
that “words in the bilingual’s mental lexicon activate conceptual features which
are assumed to be distributed such that particular concepts correspond to sets of
activated features”. Van Hell and De Groot (1998a) later proposed to extend the
Distributed Feature Model and suggested the distribution of features not only at
the conceptual but also at the lexical level (cf. Fig. 3.3). It is imperative to indi-
cate, however, that the version of the model presented in Fig. 3.3 depicts only one
layer of lexical features. This simplification was done for the sake of clarity and
not to imply that multilayered representation of various aspects of lexical form is
excluded from the model.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.2 Mental Representation of Bilingualism 77
The Revised Hierarchical Model was proposed by Kroll and Stewart (1994) to
“characterize the consequences of differential expertise in the two languages for
the connections between words and concepts” (Kroll and Stewart 1994, p. 157).
The model includes independent lexical representations for each language, with
L1 assumed to be larger than L2, and a shared conceptual representation. Of inter-
est is that it focuses on the connections between words and concepts, not on the
structure. The model is based on the classic study of Potter et al. (1984) who refer-
ring to the empirical findings supporting the assumption that words may be stored
separately from concepts (cf. Potter 1979), proposed two separate models of
4 Empiricalevidence for concept mediated translation of cognates was found also by Kroll and
Stewart (1994) and will be referred to in Chap. 4.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
78 3 Modelling Multilingual Representation and Processing
5 For detailed description of empirical data used in the original research by Potter et al. (1984)
see the review by Kroll and Tokowicz (2005).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.2 Mental Representation of Bilingualism 79
Fig. 3.5 The revised
hierarchical model of lexical
and conceptual representation
in bilingual memory
(adaptation based on Kroll
and Stewart 1994)
from L1 to L2 (depicted with a dotted line). By and large, this asymmetry in the
strength of connections between words in the two languages and meaning is a
characteristic feature of this model.
The model represents the consequences of the learning history of the late sec-
ond language learner for whom lexical and conceptual representations are already
in place for L1 when L2 learning begins. Early in the acquisition, L2 words are
strongly linked with the L1 items and only weak links connect L1 words with
those of L2. Kroll (1993) emphasizes the fact that
the strength of the connections between the L1 and L2 lexicons and between each lexicon
and conceptual memory is hypothesized to vary as a function of relative fluency in L2 and
language dominance (Kroll 1993, p. 70).
In other words, with the increasing proficiency, stronger and stronger semantic
links between concepts and L2 items are developed.
What is interesting to note is that the empirical observation that led to the RHM
was initially the finding that translation from L1 to L2 is typically slower and
more error-prone than translation from L2 to L1. It took less time for proficient
participants to translate words from L2 to L1 (backward translation) than from L1
to L2 (forward translation). To explain these findings Kroll and Stewart claimed
that backward translation operates within the lexical level (hence faster), unlike
forward translation which is conceptually mediated (slower). By analogy, Kroll
and Dussias assume that
[i]n the L2 to L1 direction, the strongly associated translation equivalents will be assessed
directly. In the L1 to L2 direction, the bias to activate the meaning of the L1 word will
encourage reliance on a translation route that engages semantics. The latter process will
require additional processing and also the potential negotiation of lexical competition
prior to selecting an L2 response. The L1 to L2 direction is hypothesized to be particularly
difficult for less proficient bilinguals for whom the concept to L2 links are relatively weak
(Kroll and Dussias 2004, p. 176).
6 For a critical review and assessment of the model see Kroll et al. (2010).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
80 3 Modelling Multilingual Representation and Processing
Fig. 3.6 The revised
hierarchical model (RHM)
with integrated lexicons
and distributed semantic
representations; Dutch-
English stimulus example
(adaptation based on Desmet
and Duyck 2007, p. 177)
of Duyck and Brysbaert’s (2004) study contrast with the RHM’s developmental
hypothesis, which states that L2 lexico–semantic mappings may only develop in
very high proficiency levels.
A critical review of the model was also proposed by Desmet and Duyck (2007)
who noted its two important problems. Firstly, as the researchers rightly note, the
model still contains two separate lexicons, which is clearly at odds with the large
body of evidence against lexical autonomy. Secondly, Desmet and Duyck suggest
that the RHM should be more specific about the nature of semantic representa-
tions, assuming gradual semantic feature overlap depending on word variables
(similar to the way semantics are conceived in the Distributed Feature Model; cf.
Sect. 3.2.3.1). According to Desmet and Duyck (2007), an updated version of the
RHM with the modifications concerning the above mentioned arguments would
probably look like the model presented in Fig. 3.6.
The model aptly illustrates the current mixed storage theories where words are
stored in a language specific manner and concepts, which underlie the words, form
a single language-independent store.
The preceding sections focused on the lexical storage in the bilingual mental lexi-
con. The vexing question that arises in the context of the present work, however,
is what happens to the lexical storage system when more than two languages are
involved. In the following paragraphs an attempt will be made to delineate one of
the most prominent issues in the multilingual lexicon studies, namely, the degree
to which lexical operations are separate or integrated within a multilingual mind.
In turn, the problem of multilingual processing will be tackled in Sect. 3.7.
As already mentioned in Sect. 3.3, early studies (e.g., Meara 1982) assumed that
the L1 mental lexicon was qualitatively distinct from the mental lexicons of other
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.3 Modelling Multilingual Representation 81
Fig. 3.7 Model of
multilingual memory
representation (adaptation
based on Gabryś-Barker
2005, p. 64)
languages, the difference in quality implying the separate storage. Another typical
argument against integration of lexical systems of many languages in one mind was
put forward by the proponents of the modularity hypothesis (cf. Sect. 2.6.1). Strong
evidence supporting the separatist view came also from studies of aphasia and lan-
guage loss in multilinguals (cf. Sect. 3.2.1). Interestingly enough, some early research-
ers referred to the psychotypology effect arguing that it may support selectivity which,
by implication, may suggest separate storage (cf. Singleton 2003 for review).
In contrast, more recent studies promote the idea of the high degree of intercon-
nectivity within a multilingual mind. One of the most fervent supporters of this
view is Vivian Cook (e.g. 1996, 2007a, b) who has for many years been advocating
for the complete integration of language competence across languages—the holis-
tic multicompetence hypothesis (cf. Chap. 1). Further, extensive evidence in favour
of integration comes from Franceschini et al.’s research findings, comprehensively
reviewed in the authors’ contribution on a range of brain-imaging studies published
in what seems one of the first volumes devoted entirely to the multlingual acquisi-
tion, storage and processing (Franceschini et al. 2003; in Cenoz et al. 2003). The
results of the conducted studies, all lead to the conclusion that lexical-semantic
aspects of the processing of all languages known to an individual are subserved by
essentially the same areas of the cortex (cf. Franceschini et al. 2003). Importantly,
the authors of the study account for the role of the level of proficiency and the age
of acquisition in the organisation of the lexicon. Their research findings imply that a
high level of proficiency acquired in a late learned L2 may mask the differences con-
nected with the onset time of acquisition (Franceschini et al. 2003, p. 163). The con-
tention which, if confirmed, seems to be of importance for the pedagogical context,
which for many years has been strongly promoting early language education.
The role of language proficiency for the multilingual lexicon structure was also
confirmed in the research by Gabryś-Barker (2005). Basing on the data gathered in
a series of association tests conducted on Polish-English-German and Portuguese-
English-German trilinguals, Gabryś-Barker put forward a multilingual adaptation
of Kroll and Stewart’s Revised Hierarchical Model. In this model, depicted in
Fig. 3.7, items in the multilingual memory are interconnected by way of two types
of links, lexical and conceptual. The lexical links exist “within the same language
referring to the form as a factor and across languages using translation equivalents”
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
82 3 Modelling Multilingual Representation and Processing
(2005, p. 64), whereas the conceptual links exist “within the same language and
across languages manifested as a semantic field search, e.g., coordination or super-
ordination” (2005, p. 64). It is believed that, in the multilingual memory words are
accessed either via lexical links or conceptual links depending on a set of factors,
such as language dominance in the multilingual competence and performance of a
learner, language proficiency in all the languages, the form of a linguistic task, and
the type of a linguistic stimulus (Gabryś-Barker 2005, p. 64).
Gabryś-Barker concludes that the gathered research findings confirmed “a con-
ceptually based L1 mental lexicon, separation of the lexicons with growing lexi-
cal competence (L2) and strong connections between high proficiency L1 and L2,
and L3 (the lowest level proficiency) lexicons, in other words, integration between
these lexicons” (2005, p. 73).
Important evidence for the full integration of language systems within the mul-
tilingual mind can be found in the research conducted by Dijkstra (2003a, b) and
Dijkstra and Van Heuven (e.g., 1998, 2002a) who have devoted much time to the
investigations of the multilingual processing. Since the present section is pre-
dominantly focused on mental representation, both bilingual as well as multilin-
gual models proposed by Dijkstra will be delineated in Sect. 3.7, which is entirely
devoted to the problem of multilingual lexical retrieval. The present discussion
will instead offer evidence of a high degree of cross-lexical connectivity (also
referred to in the practical chapters of this work), which is what most researchers
on the multilingual lexicon mean by integration.
Psychologists, psycholinguists as well as neurolinguists opine that the most
important principle for the organization of the mental representation is the ability
to categorize new information basing on similarity to existing knowledge repre-
sentations (cf. Rosch 1975; Fay and Cutler 1977—discussed in Sect. 2.4.3 of the
previous chapter). Consequently, many theories aimed at designing a multilin-
gual mental representation model refer to the exploitation of similarity between
new lexical input and previous lexical knowledge. An important example is the
Parasitic Model originally devised to explain aspects of L2 vocabulary acquisi-
tion and later extended to incorporate the L3. The model makes strong hypotheses
about the initial stages of the cognitive processes involved in additional language
vocabulary development and concentrates on the automatic, unconscious detec-
tion and adaptation of similarity between novel lexical input and the information
already stored in the mental lexicon (cf. parasitic learning strategy; Hall 2002; Hall
et al. 2009). According to the Parasitic Model, “new lexical representations will be
integrated where possible, into the rest of the network via connections with preex-
isting representations (…), at points of similarity or overlap between them” (Hall
and Ecke 2003, p. 72)—the phenomenon documented in a number of studies on the
role cognates in CLI (e.g., Ringbom 1987, 2001, 2007; Cenoz et al. 2001a). The
model predicts that new words are integrated into the existing lexical network with
the least possible redundancy and as rapidly as possible. As Hall and Ecke (2003,
p. 71) declare, the ability to integrate new information within the existing knowl-
edge based on the criterion of similarity is “essential for the development of con-
ceptual relations and networks as well as for the acquisition and organization of the
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.3 Modelling Multilingual Representation 83
mental lexicon”. By way of illustration, when a new L3 verb has a cognate in one
of the previous languages learned, but not in the other, learners will tend to assume
that the new verb is used in the frame of its cognate in the L1 or L2 (cf. Hall and
Ecke 2003). For example, the L3 German verb brechen, at the early stage of lexical
acquisition, is connected parasitically to the L2 English form break, rather than the
L1 Spanish form romper, because of the cognate relationship between the L2 and
L3 forms. Thus, the learner will be led to assume that the German form brechen is
used nonreflexively, like its English L2 cognate counterpart break, rather than with
a reflexive pronoun, as in the case of the Spanish L1 translation romper (cf. Hall
et al. 2009, p. 163). Overall, the model adopts the view of the mental lexicon as “an
interconnected network of processing units through which the activation spreads”
(Hall and Ecke 2003, p. 7). Items in the lexicon are connected on the basis of
similarity—both within a single language and between languages. They are repre-
sented in the form of a lexical triad (cf. Triad Model; Hall 1992 in Hall et al. 2009)
which comprises three levels: form (or lexeme) which is a representation of the
phonological and/or orthographic features of a word, syntactic frame (associated
with Levelt et al.’s, (1999) notion of lemma) which includes a specification of a
syntactic category and any other feature regarding sentential deployment of a word,
and concept (representation of meaning).
The role of typological or rather psychotypological similarity for the integra-
tion of linguistic systems in the multilingual mind resonates also in the model
proposed by Herwig. In her view, the mental lexicon comprises “dynamically
interacting subsystems of a common linguistic system, subject to individual vari-
ation and change over time” (Herwig 2001, p. 115). To shed some light on the
multilingual mental lexicon, Herwig adopts Paradis’ Subset Hypothesis for TLA,
advocating for three languages to form a single system with three identical, lan-
guage-specific networks of connections. In the case of SLA, in the initial stages
of bilingual development, there is an extended system where the second language
is strongly related to the L1 items. With progress in L2 proficiency, the L2 net-
work becomes less L1-dependent to the extent of becoming a fairly independent
subsystem. According to Herwig, the same pattern of development is repeated in
TLA. The point is that it is not necessarily L1 that undergoes the initial extension.
Instead, it seems that the language that is typologically closer to L3 will undergo
the initial extension of the network. As Herwig puts it, “the structure of the lexi-
con, the connectivity of its subsystems and their interrelation depend on a number
of factors, such as perceived linguistic distance, proficiency of the user, or method
of acquisition” (Herwig 2001, p. 115).
The effect of psychotypology was also exploited by Singleton, who in 2002
proposed a model of a trilingual speaker based on Weinreich’s assumptions.
Adopting Weinreich’s models for trilinguals, Singleton concentrated predomi-
nantly on the subordinate organisation of languages. The key question he asked
was to which of the previously acquired languages the L3 would become sub-
ordinate. On analysing empirical data he concluded that L3 will be subordinate
towards the language which is perceived as typologically closer (in his example
Italian to Spanish). He also hypothesized that with progress in the acquisition of
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
84 3 Modelling Multilingual Representation and Processing
The previous sections focused on the issue of bi- and multilingual representation and
showed that the majority of current research arguments in favour of the mixed stor-
age and high interconnectivity of language systems within one mind. To get a better
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.4 Modelling Bilingual and Multilingual Processing 85
insight into how languages are interconnected, the present sections takes a closer
look at the mechanisms of how the languages are accessed. However, before some
information on lexical access is provided, it is worth mentioning that research stud-
ies that address the question of whether lexical candidates from different languages
are activated during bilingual word recognition basically exploit three main types
of stimulus materials: interlingual homographs, also referred to as homographic
noncognates or more typically false friends (words with the same form but differ-
ent meaning across languages such as fin meaning end in French), homographic
cognates (words sharing both form and meaning across languages such as wolf in
English and German) and interlingual neighbours which resemble interlingual hom-
ographs except that their forms do not overlap completely between the languages
(such as Dutch mand and English sand; example after De Groot 2013a, p. 76). These
three types of stimuli are typically used since it is believed that if reaction time dif-
ferences between these words and matched control words arise, this will be probably
due to their existence in two or more languages (cf. Dijkstra 2005; De Groot 2013
for a thorough discussion of all the types of stimuli).
The debate over the selective/non-selective access dichotomy has not ceased to
inspire theoretical considerations and empirical investigations. In a number of early
studies on lexical access, language selective position was favoured. On the basis of
their research conducted with a group of Spanish-English bilinguals, Gerard and
Scarborough (1989) concluded that the participants were accessing each of their two
lexicons selectively when they performed a monolingual task. Similarly, in a pleth-
ora of early research studies (cf. Caramazza and Brones 1979; in Singleton 1999;
Soares and Grosjean 1984; in Dijkstra 2005) no clear reaction time (RT) differences
were observed between test items and controls. Also some later studies replicated the
RT null results under similar experimental circumstances (cf. De Groot et al. 2000;
a lexical decision experiment with interlingual homographs and cognates). These
research findings were interpreted as constituting support for language specific
access as the participants of the studies seemed to have been operating in a language
specific manner without any influence from their L2. It has to be stressed, however,
that the early understanding of language selective access implied partition of the lex-
icon by language. Clearly enough, this assumption has evolved in the course of psy-
cholinguistic research and it is now generally agreed that language selective access
does not have to imply the existence of separate lexicons for each of the languages,
but it may be the result of different activation patterns for L1, L2 and Ln (cf. Costa
and Caramazza 1999; Paradis 2004, 2009; De Groot 2011).
In recent years, more and more studies reported evidence in support of lan-
guage nonselective access with respect to form (orthographic and phonological),
as well as semantic representations. Obviously enough, this does not mean that
words from the two or more languages cannot be distinguished anymore; rather,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
86 3 Modelling Multilingual Representation and Processing
7 The notion of the neighbourhood was introduced in Sect. 2.5.3 of the previous chapter, where
competition between the activated words constituting the fundamental tenet of the cohort model
was discussed.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.4 Modelling Bilingual and Multilingual Processing 87
In yet another study, Lemhöfer and Dijkstra (2004) found that cognates were rec-
ognized faster than the matched English and Dutch controls. Because at the same
time the homographs (having an identical orthographic form across languages)
showed no effects (relative to Dutch controls), the effect for cognates seemed to
be determined at least by the fact that their meanings overlap across languages.
The obvious interpretation was that the cognates’ semantics in both languages
must have been co-activated. Also, the data coming from cross-linguistic priming
and repetition effect tasks (cf. Woutersen 1997; Jiang and Forster 2001; Basnight-
Brown and Altarriba 2007) clearly support the nonselective access view. Primed
lexical decision tasks revealed that semantically related words prime each other,
even when prime and target appear in different languages (cf. Duyck 2005).
Lexical decisions were facilitated by cross-linguistic orthographic and semantic
similarity relative to control words that belonged only to English.
Overall, the research results in multilingual word recognition demonstrate a
considerable amount of interaction between the languages known by a multilin-
gual, which has led the majority of researchers to believe that lexical access in
multilinguals is basically nonselective with respect to language (Caramazza and
Brones 1979; De Groot et al. 2000; Kroll and Dijkstra 2002; De Groot 2013b),
automatic (i.e., not under control of the reader/listener), and, although task
dependent, its first processing stages might remain unaffected by nonlinguistic
factors (cf. Dijkstra 2003b). In this view, lexical word representations from both
(or more) languages are activated even in situations where only one language
is relevant. Consequently, the group of word candidates that compete for selec-
tion within the word recognition or production process is not restricted to one
language.
Importantly, it has to be noted that all the above mentioned experiments con-
cern out-of-context word recognition performance. The empirical data reviewed
in the previous paragraphs indicate that language nonselectivity is a compelling
feature of this type of recognition. This means that word candidates form differ-
ent languages initially become active on the presentation of a letter string. This
nonselectivity seems to hold for all representations that characterize words (e.g.,
orthographic, phonological, and semantic codes). Further, multilingual word
recognition also seems to be automatic in the sense that the process takes place
relatively unaffected by nonlinguistic contextual factors. This applies not just to
words from the native language (L1), but also to words from the L2. At the same
time, De Groot (2013) raises a valid point when she remarks that some research
is still needed to verify whether language nonselectivity is maintained or elimi-
nated in context since the empirical data gathered thus far show that when words
are processed in sentence context, their processing seems to be sensitive to the
semantic and syntactic aspects of the sentence (cf. Hartsuiker et al. 2004, 2008; in
Riehl 2010).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
88 3 Modelling Multilingual Representation and Processing
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.5 Models of Bilingual Visual Word Recognition 89
perceptual input (cf. Forster 1976, 1989; Sect. 2.6.1). Another advocated the view
of a set of word “detectors” that compete to collect evidence that their word was
present in the input (cf. Morton 1969; in Singleton 1999; Sect. 2.5.2). Both theo-
ries have been criticized for being little more than verbal descriptions and more
recently they have been replaced by more realistic computational models which
“force clarity on theories because they require previously vague descriptive
notions to be specified sufficiently for implementation to be possible” (Thomas
and Van Heuven 2005; cf. Sect. 2.5.4).
Within the bilingual domain, a number of computational models have been
used. Notably, many computational models of bilingual word comprehension
have worked within the connectionist tradition. In fact, bilingual researchers have
appealed to two different types of connectionist models in studying processes
of comprehension: localist and distributed. Localist models are seen as direct
descendants of the original word detector models proposed in the 1970s. These
models do not incorporate change according to experience, their focus within mul-
tilingual research has been to investigate the static structure of the word recog-
nition system in the adult multilingual. One good example of a localist model is
the Bilingual Interactive Activation [BIA] model and its revised version known
as BIA+. Another important representative of this group of models is Semantic,
Orthographic, and Phonological Interactive Activation [SOPHIA] model (Dijkstra
and van Heuven 2002). The Bilingual Interactive Model of Lexical Access
[BIMOLA] (Léwy and Grosjean 1997, 2008) also belongs to the group of localist
models but focuses on spoken word recognition.
Distributed models, on the other hand, tend to represent individual entities (like
words) as patterns of activity spread over sets of units. The entity represented by a
network cannot, therefore, be identified by looking at a single unit, but only as a
code over several units. Consequently, they are able to account for both language
independence and language interaction within a single network. More interest-
ingly, distributed models tend to focus on experience-driven change, specifically
on learning to map between codes for different types of information (such as a
word’s spoken form and its meaning). Since these models incorporate changes
according to experience they are often applied to issues of language acquisition
and change in language dominance over time. Some prominent examples of the
distributed developmental models are: the Bilingual Single Network [BSN]
(Thomas 1997); the Bilingual Simple Recurrent Network [BSRN] (French 1998)
or the Self-Organizing Model of Bilingual Processing [SOMBIP] (Li and Farkaš
2002), and its later version known as the Developmental Lexicon Model [DevLex]
(Li et al. 2007).8
Importantly enough, the distinction between the localist and distributed mod-
els is not a dichotomous one, but rather a continuum. According to Van Heuven
and Dijkstra (2010, p. 105), although localist and distributed models have different
advantages for studying various phenomena of bilingual language processing, the
8 For a thorough discussion of the computational bilingual models (see Warren 2013,
pp. 231–235).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
90 3 Modelling Multilingual Representation and Processing
The Bilingual Interactive Activation Model (BIA) by Dijkstra and Van Heuven
(1998) is an implemented localist-connectionsit model (processing units are separate
and unitary nodes in a network) that extends the McClelland and Rumelhart (1981)
Interactive Activation Model (IA; cf. Chap. 2) to the bilingual case. Similarly to
monolingual word recognition, processing is assumed to be initiated bottom-up upon
presentation of visual input and nonselectively so that all information resembling the
input is activated. The BIA differentiates between four hierarchically organized levels
of different linguistic representations: letter features, letters, words, and language tags
(or language nodes). The model hypothesizes that all lexical representations from the
same language are linked bi-directionally to a single language node at a superior rep-
resentational level. This means that activation from letters spreads to words and then
to language nodes and vice versa. Consequently, the activation levels of the language
node can influence representations at the word level. As depicted in Fig. 3.8, when a
string of letters is presented to the model, it activates features at each letter position,
which subsequently leads to the activation of letters possessing those features, and,
simultaneously, to the inhibition of other letters. Thus activated letters excite ortho-
graphic forms of both languages which contain those letters in appropriate positions,
which in turn leads to competition among words sharing overlapping or identical
orthography in the bilingual’s two languages (e.g., cognates). In Fig. 3.8 such excita-
tory connections are indicated by arrows, with arrowheads pointing in the direction
of activation spread. To allow for language selection, the model introduces a sepa-
rate level of representation within the lexicon—the level of language nodes. It is at
this level that words from the non-selected language are inhibited in such a way that
the inhibition increases the likelihood of selection of the intended language words,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.5 Models of Bilingual Visual Word Recognition 91
Fig. 3.8 The bilingual
interactive activation model
(adaptation based on Dijkstra
and Van Heuven 1998,
p. 207)
which, however, does not affect the initial activation of words in both languages.
In Fig. 3.8 inhibitory connections are indicated by lines with closed circles.
Extensive evidence for the BIA model can be found in studies in which aspects
of word type have been experimentally manipulated, like in language-specific and
language-general decision tasks. A growing number of recent studies using this
methodology have provided support for the claim that lexical access is nonselec-
tive and that bilinguals cannot help but respond as if information in both languages
was active. Kroll and Dussias (2004) offer a comprehensive review of these stud-
ies which include the use of interlingual homographs (cf. Dijkstra and Van Heuven
1998; De Groot et al. 2000, Smits et al. 2006), cross-language neighbours (cf.
Jared and Kroll 2001; Kerkhofs et al. 2006; Dijkstra et al. 2010) and, most impor-
tantly for the present work, cognates (cf. Dijkstra and Van Heuven 1998; Van Hell
and Dijkstra 2002; Kroll and Sunderman 2003; Hall et al. 2009).
Despite substantial and convincing evidence supporting the validity of the model,
certain doubts concerning some aspects of bilingual word recognition that are not
fully accounted for by Dijkstra and Van Heuven’s proposition have been raised by
the authors themselves. Notably absent in the BIA model is the representation of
semantics. The assumption at this level of analysis is that lexical form properties
of words in the bilingual’s languages are activated in a bottom-up fashion and only
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
92 3 Modelling Multilingual Representation and Processing
later in processing does the output of the lexical identification system interact with
semantics and higher-level context. Another line of criticism concerns the lack of
phonological representation, which explains why the model fails to cater for results
of between-language phonological priming found e.g., by Brysbaert et al. (1999)
or Brysbaert and Dijkstra (2006). Most importantly, the model did not account for
task and context effects, although there is clear evidence in the literature that task
demands can influence bilingual performance to a considerable extent (cf. De Groot
et al. 2000). The more recent variations of the BIA model: SOPHIA and later BIA+
seek to fill in some of these theoretical gaps (cf. Sect. 3.5.3).
As indicated above, the BIA model failed to incorporate the distinction between the
actual word identification system and a task/decision system. It the light of numer-
ous studies confirming that bilingual word processing often depends on the task that
is to be performed and the context in which it is performed, this is a rather obvious
shortcoming. Indeed, it seems impossible to talk about multilingual word recogni-
tion in general without describing the conditions under which it takes place and the
goal that needs to be achieved. The Inhibitory Control Model (IC) of bilingual word
recognition offered by Green (1986, 1998) and presented in Fig. 3.9 successfully
meets these ends. It accounts for the fact that bilingual language processing always
takes place within a particular task context and with certain goals in mind. As the IC
model has been more often applied to bilingual language production and translation
tasks than to visual word recognition processes, the following discussion will be lim-
ited to those aspects of the model which are of importance for the experimental part
of present text, namely task dependency.
Fig. 3.9 The inhibitory
control model (adaptation
based on Green 1998)
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.5 Models of Bilingual Visual Word Recognition 93
The topmost part of the model and its key element is the “conceptualizer”,
which is language independent and builds conceptual representations on the basis
of information stored in the long-term memory. These representations are driven
by the “goal”, i.e. an intention to produce a word in a particular language. The
supervisory attentional system (SAS), which is the main element of the model,
along with the lexico-semantic system and a set of language task schemas, medi-
ate the planning. For Green (1998, p. 72) the term schema refers to “mental
devices or networks that individuals may construct or adapt on the spot in order
to achieve a specific task and not simply to structures in long-term memory”. In
various language tasks different schemas (e.g. translation schemas or word pro-
duction schemas) compete for output. Once the schema is specified, it is retrieved
from memory and modified if the task demands it. A given language task schema
regulates the outputs from the lexico-semantic system whose representations fol-
low Levelt’s (1989) and Levelt et al.’s (1999) model where each lexical concept
is associated with a lemma containing syntactic specification of the concept.
Importantly, each lemma is tagged for language and the language tag is also a part
of the conceptual representation. In order to achieve a linguistic goal, a particular
task schema is activated under the control of SAS. Next, the task schema regulates
activation and suppression of tags at the lemma level. Activated lemmas send acti-
vation to the associated word forms in a given language and, finally, output in the
appropriate language is produced.
Despite its numerous merits, Green’s model is not free of certain shortcomings,
one of them being the fact that it does not fully account for the way the recogni-
tion of items within the lexico-semantic system takes place. It is worth mention-
ing that this criticism points to an aspect that the BIA has also failed to address,
namely a more precise description of the interactions at the sub-lexical level,
including some phonological representations.
Due to its similarity, the common basis and the same developers, the last two
models will be discussed together. As already stated in Sect. 3.5.1, an important
problematic issue criticized in the BIA model was the lack of semantic and pho-
nological levels of representation. These gaps have been addressed by Van Heuven
and Dijkstra (2001) in the semantic, orthographic and phonological interactive
activation model (SOPHIA). Accordingly, the architectural structure of SOPHIA
depicted in Fig. 3.10 incorporates the BIA system but also includes levels for
semantics and phonology. More specifically, nodes at one particular level can
activate and inhibit units in neighbouring levels via excitatory and inhibitory con-
nections, respectively. Moreover, representations at a particular orthographic or
phonological level mutually inhibit each other via lateral inhibition. However, the
connections between an orthographic level and its analogous phonological level
are of an excitatory nature only (i.e. they do not inhibit each other). For instance,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
94 3 Modelling Multilingual Representation and Processing
Fig. 3.10 The semantic,
orthographic, and
phonological interactive
activation (SOPHIA)
model of word recognition
(adaptation based on Van
Heuven and Dijkstra 2001)
the presentation of the visual stimulus ‘cat’ will activate its orthographic word
node for ‘cat’ and the corresponding phonological node [k t].
Another alteration introduced to SOPHIA is the inclusion of a semantic system
which can directly interact with remaining structural levels in the model. Finally,
and most importantly, another modification to the SOPHIA model concerns the
language nodes. In the BIA word nodes were connected to their corresponding
language node via excitatory pathways. In turn a language node exerted inhibitory
influence on all the word nodes from the opposing language. The latter inhibitory
connections have been removed from SOPHIA.
SOPHIA did not, however, resolve all the problems identified in the BIA,
in particular it did not cater for lack of context and task demands system.
Consequently, within the new version of the BIA model—BIA+, Dijkstra and
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.5 Models of Bilingual Visual Word Recognition 95
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
96 3 Modelling Multilingual Representation and Processing
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.6 Models of Multilingual Visual Word Recognition 97
Fig. 3.12 Trilingual
interactive activation model
(adaptation based on Dijkstra
2003a)
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
98 3 Modelling Multilingual Representation and Processing
Apart from bottom-up factors that influence the speed of processing (such as the
language-specific cues9), there are top-down factors which affect activation of
items belonging to the languages known to the multilingual speaker. The top-down
factors are twofold, either non-linguistic, such as context, task demands, and stim-
ulus list compositions, or linguistic, such as morphological representations, sen-
tence level information, or language membership information. Strong evidence for
the influence of the linguistic, sentential context on lexical processing has been
found in the monolingual domain but, again, it is not certain if this could be zto
multilingual processing. As far as language membership is concerned, Dijkstra
et al. (1999) found evidence against its facilitative effect. They tested bilingual
participants on a set of Dutch-English homographs embedded in mixed language
lists. It was found that participants were unable to exclude effects from the non-target
language on interlingual homographs identification.
Summing up, it can be seen that word recognition is aided by bottom-up and
top-down factors. Neighbourhood density, storage patterns within the mental lexi-
con, as well as linguistic and non-linguistic factors appear to speed up the pro-
cess of word recognition, which helps multilinguals to cope with the processing
load related to the greater number of items that they have in their mental word
stores. The model is of special importance as the research findings will be ana-
lyzed within its framework.
3.7 Conclusion
Whereas the vast majority of accounts on the mental lexicon of foreign language
users have been concerned mainly with the processing of two languages, the pre-
sent chapter has gone beyond this, paying special attention to the interaction of more
than two languages and thus offering theoretical basis to the multilingual analysis
described in Chaps. 4 and 5. As has been demonstrated throughout the chapter, the
issue of multilingual storage is the subject of a growing range of empirical studies
on the basis of which many models and hypotheses have been developed. Some of
them argue for the separate and others for the shared organisation of the multilingual
mental lexicon. Successful, in part, seems to be an attempt to resolve this conflict
made by those psycholinguists who opt for the mixed structure of the multilingual
store. It has also been shown that lexical storage and access depend on a number of
different variables both user- and word type-determined.
The present chapter was divided into two main parts. The first of these con-
sidered multilingual mental representation theories with the special focus on the
separate/shared dichotomy. The second part was devoted to the presentation and
9 Itis suggested that the recognition might be facilitated by language specific cues such as dia-
critical markers in French or Polish words or onset capitals in German nouns. However, Kroll
and Dijkstra (2002) admit that this aspect of processing awaits further research as evidence for
the aforementioned facilitation is still limited.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
3.7 Conclusion 99
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Chapter 4
Visual Word Recognition in Multilinguals
4.1 Introduction
Now that the main issues concerning the multilingual mental lexicon storage and
retrieval have been delineated, it is time to outline the most essential aspects con-
cerning visual word recognition in multilinguals. Even a brief inspection of articles
on visual word recognition reveals that three groups of words typically exploited
in this kind of research are interlingual homographs, interlingual neighbours and
cognates (cf. Sect. 3.4). Through such words, a wealth of studies in the last decade
have confirmed that during the initial stages of word identification by multilinguals,
word candidates from several languages are often co-activated. In accordance with
these results, several word recognition models propose that multilingual word rec-
ognition involves an initial language nonselective access into an integrated lexicon.
The present chapter focuses on the last category of words which have also been
used in the empirical experiments described in the final chapter of the book. Cross-
language cognates are of special interest for designing a model of the multilingual
lexicon, because there is a possibility that at least part of their representations is
shared between two languages. The cross-linguistic form overlap of cognates has
also been used by researchers to explore whether words from different languages
are co-activated during the reading, listening, and speaking of multilinguals. If
responses to such “special” items differ from those to language-specific control
items, this can be seen as evidence that both readings of the cognates have become
active and affect each other. A number of experimental bilingual, and more
recently multilingual, studies have demonstrated a substantial difference in the
processing of cognates and monolingual control words (cf. Caramazza and Brones
1979; Dijkstra et al. 1998; Dijkstra et al. 1999; De Groot et al. 2002; Lemhöfer
and Dijkstra 2004; Lemhöfer et al. 2004, 2008; Van Heuven and Dijkstra 2010;
Van Heuven et al. 2011). In all these studies, lexical decisions on cognates were
faster and/or more accurate than those on control words. Moreover, cognates have
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
102 4 Visual Word Recognition in Multilinguals
been found to be translated faster (cf. Sanchez Casas et al. 1992; De Groot et al.
1994), to be more effective as masked or unmasked primes in visual lexical deci-
sion than noncognate translations (cf. Cristoffanini et al. 1986; Gollan et al. 1997),
to be easier to learn (cf. De Groot and Keijzer 2000), and to be named faster in
word naming (cf. De Groot et al. 2002) or in picture naming (cf. Costa et al. 2000;
Cieślicka and Kujałowicz 2006; Hoshino and Kroll 2008). Interestingly, the cog-
nate effect has also been observed in bilinguals using different scripts (cf. Gollan
et al. 1997; Kim and Davis 2003; Hoshino and Kroll 2008). Studies using electro-
physiological and neuroimaging methods (cf. De Bleser et al. 2003; de Bruin et al.
2013) also indicate that cognates are processed differently from controls.
Much of the present chapter offers theoretical background to the empirical
research documented in Chap. 5 by reviewing evidence for the special status of
cognates. First, the chapter will present word stimuli used in visual word recog-
nition studies with a special focus on the terminology and typology used in con-
nection to cognates viewed from the psycholinguistic perspective. Second, special
attention will be given to factors affecting visual word recognition in bilinguals
elaborating on their mutual interaction and its role in the visual recognition. Here,
much space will be devoted to the cognate facilitation effect, the essence of which
is that bilinguals produce and recognize cognates faster than noncognates. Further,
factors affecting cognate recognition will be delineated together with models of
the bilingual lexicon accounting for the cognate facilitation effect. Next, a brief
presentation of the more recent line of studies devoted to bilingual word recog-
nition in a sentence context will follow. In the final subchapter the effect of the
cognate status on the foreign language vocabulary learning will also be explicated.
4.2.1 Interlingual Homographs
Interlingual homographs are words that exist in two languages but their meanings
are different in these languages. In other words, they share the written or spoken
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.2 Word Stimuli Used in Visual Word Recognition Studies 103
form (homographs and homophones, respectively) but not the meaning. Many
studies have employed interlingual homographs to confirm the nonselectivity of
the multilingual lexical retrieval processes. It has been believed that if the inter-
lingual homographs and their matched controls are recognized in the same time
it means that the other meaning of the homograph is not activated which in turn
means that the lexical access is selective. If, however, the interlingual homograph
is recognized faster than its matched control it means that the access is nonselec-
tive. Such a difference in decision times and error rates between the homographs
and control words is commonly referred to as “the homograph effect”.
One of the first works in which interlingual homographs were used was
Beauvillain and Grainger’s primed lexical decision task study. The study exploited
French words to prime English targets. The results showed that a priming effect
occurred in the case of homographic primes which proves that both meanings of
a homograph are activated nonselectively. However, considering the outcomes of
the study in the light of Grosjean’s language mode hypothesis (e.g. 1982, 2008), it
must be noted that Beauvillain and Grainger’s research did not give clear evidence
about the nonselective access. According to Grosjean’s theory, both language sub-
systems were activated since the stimulus material included items from both lan-
guages. Consequently, later studies were unilingual, which means that the research
material came from only one of the bilingual’s languages. The basic question the
studies tried to address and ultimately managed to confirm was whether both lan-
guages can be activated if the experimental material from only one language is
used (e.g. Dijkstra et al. 1998).
As far as the factors affecting the homograph effect are concerned, the most
influential ones include task dependency, level of proficiency and word frequency.
Indeed, it has been confirmed that the type of task influences the direction of the
homograph effect. In a language neutral decision task it is facilitatory as responses
to homographs are faster, whereas in a language specific task the effect is inhibi-
tory, which means that responses to homographs take longer to occur (cf. De
Groot et al. 2000). Further, many studies using interlingual homographs in the
lexical decision task (Jared and Szucs 2002) confirm that in unbalanced bilinguals
the stronger language can be immune to an influence of the weaker language. Also
the frequency of the selected interlingual homographs seems to be an important
factor as reviewed in De Groot et al. (2000). All these examples clearly show
that the occurrence of the homograph effect is not that easy to predict and if the
answer to the question whether language access is selective or not was only based
on the research exploiting interlingual homographs, it would not be unequivo-
cal. Instead it depends on a number of factors such as the experimental material
(Dijkstra et al. 1998), prior activation of the non-target language (Jared and Szucs
2002), the type of task (Dijkstra et al. 1998), the proficiency level (Jared and
Szucs 2002) or, even the linguistic context (Elston-Güttler et al. 2005). Finally, it
should be noted that the effects obtained with interlingual homographs and homo-
phones have been far less consistent than those observed for cognates (cf. sect.
4.2.3). As De Groot (2013, p. 82) observes they do not rule out either one of the
two theoretical positions regarding bilingual word recognition (selective vs. non-
selective access).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
104 4 Visual Word Recognition in Multilinguals
4.2.2 Interlingual Neighbours
Another group of words typically used in the visual recognition studies are inter-
lingual neighbours. In the research on monolingual visual recognition it was
proved that orthographically similar words are already activated on the presenta-
tion of only one of those words (Andrews 1989). Importantly, the time necessary
to recognize a given word depends on the number and frequency of its ortho-
graphic neighbours—those with more neighbors are recognized slower, those
which do not have many neighbours are, in turn, recognized faster.
In 1992 Grainger and Dijkstra conducted one of the first bilingual studies using
the interlingual neighbours and proved that the neighbourhood effect also works
beyond languages. The study by Van Heuven et al. (1998) confirmed that the
recognition of target words in L1 and L2 is affected by similar word forms both
within the target language and within the other language. The authors not only
noticed that the identification latency for L2 words depended on the number of L1
neighbours, but also that the identification latency for L1 words was influenced by
the number of L2 neighbours, which was interpreted as suggesting that the influ-
ences of L2 on L1 visual word recognition work at proficiency levels lower than
generally assumed. It needs to be noted that, so far, the interlingual neighbourhood
effect was confirmed for studies using many different types of tasks and paradigms
(e.g., lexical decision or priming tasks).
4.2.3 Cognates
The third group which will be discussed in more detail are cognates. Within the
three types of word stimuli used in the visual word recognition studies, cognates
constitute a very special group. The traditional definition of cognates has its roots
in historical linguistics which sees them as lexical units (words) in two or more
languages that are “part of a relation defined in terms of direct descent from a
common word (or morpheme) belonging to a given shared ancestral language”
(Carroll 1992, p. 100). In other words, in the traditional linguistics approach,
cognate pairs are to be found in etymologically related languages. They are lexi-
cal items of similar form (which can be explained in terms of regular phonologi-
cal change in each language) and having the same (or similar) meaning (e.g.,
German Hand and English hand, German trinken and English drink, German rot
and English red, etc.). Due to their common etymological origins, cognate pairs
are very important in historical linguistics. By comparing the pronunciation of
etymologically related words, linguists can postulate tendencies in the change
of the phonological properties of words from the common ancestor language to
the present-day languages of the language family. In view of the above, cognates
can only occur naturally in languages with some shared roots. Interestingly, in
unrelated languages, some “cognates” may also exist (cf. Kim and Davis 2003;
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.2 Word Stimuli Used in Visual Word Recognition Studies 105
cf. Sect. 4.4.1) but strictly speaking they are borrowings: words taken from a for-
eign language, usually to fill a newly-formed linguistic need as a result of cultural
contact or, increasingly, technological development, and often phonologically
or orthographically adapted into the language, such as tişört (Turkish, T-shirt)
and futball (Hungarian, football). Such words are definitely quite different from
English wolf and German Wolf, where both the language-specific forms have
evolved from a common origin and long been part of the language.
However, considered from the point of view of lexical storage and processing
as well as language acquisition, the linguistic definition of cognates based on ety-
mological bonds is more than unsatisfactory. Perceived from the psycholinguis-
tic perspective, that is, how and why language users or learners identify words
as being the “same” in two languages, etymological relatedness is “neither a suf-
ficient, nor a necessary condition” (Carroll 1992, p. 98). Indeed, it seems highly
unlikely for this information to be encoded in the lexical representation of a word
in our mind. Instead, metalinguistic information, like the linguistic history of a
word, is postulated to be represented in memory separate from lexical knowledge
and is not supposed to be “part of the automatic processing of structural informa-
tion in word activation” (Carroll 1992, p. 103). Thus, in psycholinguistic literature,
cognates are any two words with shared aspects of spelling, sound, and meaning
across two or more languages, whereas cognate-pairing is usually described as
a form of automatic cross-linguistic activation of lexical addresses that have the
same or similar formal properties in two languages (ibid.). To recapitulate, in lin-
guistics, cognates are often defined as words that share a common etymological
origin. Within psycholinguistic research, however, focus is given not to etymo-
logical relations but to language processing, since only aspects of the underlying
mental representation (orthography, phonology, and semantics) are hypothesized
to affect processing. Historical linguistic roots of a word are not represented in
the mind of the language user, and thus the linguistic definition of cognates based
on etymology does not capture the form-meaning overlap that psycholinguists as
well as language practitioners are interested in (cf. Carroll 1992; Ringbom 2007).
Accordingly, in the present publication, which seeks to explore issues related to
multilingual storage and processing—both independent of metalinguistic informa-
tion—the psycholinguistic definition will be employed.
As described in previous sections, during lexical processing several units can
be activated simultaneously in the mental lexicon. In multilinguals, formally
similar lexical items in two or more languages can also receive parallel activa-
tion (cf. Grainger and Dijkstra 1992). Apparently, in the same way as formally
similar words in one language (e.g., stay, steak, stage) can prime each other (that
is, they make each other more easily accessible), interlingual cognate pairs (e.g.,
G. singen—E. sing; G. blau—E. blue) seem to prime each other between languages
(cf. Costa et al. 2000; Costa et al. 2005). Moreover, lexical items in two languages
do not need to be completely identical in form for neighbourhood activation. On
this account, research most often reports on cross-language cognates, meaning both
formally identical or almost identical words in two languages (cf. Beauvillain and
Grainger 1987; Gerard and Scarborough 1989; Sect. 4.4.1). Consequently, many
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
106 4 Visual Word Recognition in Multilinguals
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.2 Word Stimuli Used in Visual Word Recognition Studies 107
English land or German Frieden and English freedom), are called deceptive false
cognates, whereas word-pairs that share no semantic features and cannot be trans-
lation equivalents in any context are typically referred to as accidental (or formally
determined) false cognates (e.g. German Rock meaning skirt and English rock).
In the above categorization etymological relationship, or the lack of it, is still a
fundamental aspect in the definition of cognate types. In cognitive psycholinguis-
tics, however, the genetic relatedness of words is not taken into account and the
type of the cognate-pairing is determined by the differences in the mental repre-
sentations of formally similar words and the nature and degree of formal resem-
blance. Indeed, the speed and accuracy of cognate recognition depends, above all,
on their cross-linguistic form similarity. Hence in terms of the nature of the simi-
larity of form, psycholinguistic typology concentrates on orthographic and pho-
nological cognates. There are many studies that have examined the influence of
phonological and orthographic overlap of cognates in comprehension tasks. As
Browne (1982) rightly observes, visual accessing of the lexicon can give rise to
different cognate pairs from those arising from auditory processing. Moreover,
there is now quite some evidence that visually presented words are not recognized
on a visual basis alone. Readers rely on the spoken—phonological—informa-
tion enclosed in the orthographic stimulus when they identify written words, and
researchers no longer quarrel about whether phonology is involved in visual word
recognition but to what extent it is involved (cf. De Groot 2013a).
Another distinction important for lexical recognition research is the one based
on the degree of similarity (identical vs. non-identical cognates). For instance,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
108 4 Visual Word Recognition in Multilinguals
Dijkstra et al.’s (2010) study, more fully described in Sect. 4.4.1, analyzed how
cross-linguistic similarity affects Dutch-English identical and non-identical cog-
nate recognition. Results showed that the emergence and the directionality of
cross-language form similarity effects can depend not only on task demands but
also on cognate type (identical vs. non-identical).
Research on visual word recognition can pride itself on a relatively long tradi-
tion and throughout many years of studies a number of factors affecting written
lexical access have been enumerated and explored (cf. Warren 2013). Before the
most important, and most commonly analysed factors affecting visual word rec-
ognition are delineated, it seems justified to note that the influence of many of
those effects was primarily investigated within the monolingual domain and only
more recently has it been explored in the multilingual context. It also needs to be
observed that the multitude effects influencing lexical recognition do not exist in
separation. Instead, they interact making visual word recognition a highly complex
phenomenon. In the multilingual context, the complexity caused by interaction of
the various effects is further complicated by the occurrence of additional language
or languages. Accordingly, the following paragraphs elaborate on the most typical
factors affecting visual word recognition trying to contextualize them within the
multilingual domain.
The word superiority effect was described as one of the first factors influencing
visual word recognition. In 1887 Cattell confirmed that identification of words
is easier and faster than the identification of the so-called nonwords—words that
from the phonological perspective can occur in a language but do not exist in a
given language. Cattell proved that the identification of letters within a string of
letters creating a meaningful word is easier than identifying words within a mean-
ingless stream of words. The results of Cattell’s and a number of later studies
prove that word recognition is not based on a letter-by-letter sequence. Instead,
the effect accounts for the existence of some top-down processing from words to
letters. More recent studies distinguish between legal and illegal nonwords (word
and wlod, respectively; cf. Warren 2013). Interestingly enough, the word superi-
ority effect concerns both types of words form this category. In the experiments
presented in the practical chapter, nonwords are used along with bi- and trilingual
cognates and control words.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.3 Factors Affecting Visual Word Recognition in Bilinguals 109
Another important factor influencing our visual lexical recognition that has been
recurrently explored in the monolingual domain is quite physical and refers to
the very length of the word. Word length can be based on orthographic measures
(number of letters) or phonological measures (number of phonemes and sylla-
bles). Importantly, as New et al. (2006, p. 45) report in their review article on the
word length effect in written lexical access, these different measures are gener-
ally highly interconnected, and they also correlate with other variables (such as the
number of orthographic neighbors and the printed frequency) that influence word
recognition. Although it seems reasonable that our recognition should slow down
together with the word length, a variety of tasks (such as perceptual identifica-
tion, lexical decision, naming or eye tracking) used to examine the effects of word
length on visual word recognition offers rather inconsistent results ranging from
inhibitory (longer words are more difficult) to null effects. Frederiksen and Kroll
(1976), for instance, obtained reliable inhibitory length effects in naming but not
in lexical decision, while Hudson and Bergman (1985) found length effects in both
types of tasks. It needs to be noted that the factor under discussion is of crucial
importance also for the bilingual visual recognition studies where the effects of
word length on the recognition of foreign language words have been investigated
in a number of studies (New et al. 2006; Lemhöfer et al. 2008). Such a criterion
was also used in the experiments presented in the final chapter of the book.
4.3.3 Frequency Effect
Results of lexical decision tasks show that high frequency words are recognized
faster than low-frequency words, the relation known as the frequency effect (e.g.
Howes and Solomon 1951; Schilling et al. 1998). Participants respond faster to
high-frequency words than to low-frequency words in almost any lexical process-
ing task, including lexical decision, reading aloud, semantic categorization, and
picture naming. Not only response times are quicker but also error rates are lower
in the case of words that are used more frequently. In the bilingual domain, some
evidence suggests that the frequency effect might even be larger in the second as
compared with the first language (van Wijnendaele and Brysbaert 2002). In a lexi-
cal decision experiment conducted with Dutch-English bilinguals, Duyck et al.
(2008) compared the effect of word frequency on visual word recognition in the
first language with that in the second language and found that, even though corpus
frequency was matched across languages, bilinguals showed a considerably larger
frequency effect in their second language.
As the authors of the research suggest, the findings support models of lexical
access that incorporate the frequency effect as the result of an asymptotic learn-
ing process (Duyck et al. 2008, p. 852). However, a recent study by Peeters et al.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
110 4 Visual Word Recognition in Multilinguals
4.3.4 Neighbourhood Effect
Lexical neighbours, already mentioned in Sect. 4.2.2, are words with similar spell-
ing patterns. The role of a target’s orthographic neighbourhood in visual word rec-
ognition has been investigated in a number of studies both in the monolingual as
well as in the bilingual context. However, the role that activated neighbours play
in target recognition has so far not been determined unequivocally (cf. Pugh et al.
1994). As Warren points out, “a number of properties of the neighbourhood are
important for visual word recognition” (2013, p. 143). Above all, the neighbour-
hood size or density is strongly connected to the lexical frequency characteristics
of the neighbourhood and of the target itself. Generally, responses to low-fre-
quency words are affected by the neighbourhood effect, whereas in the case of
high-frequency words the effect is not likely to occur. To complicate things even
further, the nature of this effect depends on task demands. Depending on task type,
the neighbourhood density or word frequency, the effect yields facilitatory, inhibi-
tory or null results (cf. Pugh et al. 1994).
The neighbourhood effect has also been analysed within the bilingual context.
As already mentioned, in 1998 Van Heuven et al. (1998) conducted a series of pro-
gressive demasking and lexical decision experiments investigating how the rec-
ognition of target words exclusively belonging to one language is affected by the
existence of orthographic neighbours from the same or the other language of bilin-
guals. The results showed that increasing the number of orthographic neighbours
in Dutch systematically slowed response times to English target words in Dutch/
English bilinguals, whereas an increase in target language neighbours consistently
produced inhibitory effects for Dutch and facilitatory effects for English target
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.3 Factors Affecting Visual Word Recognition in Bilinguals 111
words (Van Heuven 1998, p. 458). The experiments were interpreted as providing
evidence for parallel activation of words in an integrated Dutch/English lexicon.
As stated above, cognates have been one of the most important sources of stimulus
materials in studies attempting to unravel the process of bilingual word recogni-
tion. The cognate facilitation effect (also CFE), according to which cognates are
processed faster than noncognates, has often been taken as evidence for a multi-
lingual lexicon that stores words of two or more languages in an integrated fash-
ion and/or for a lexical access procedure that activates word candidates in several
languages in parallel (the nonselective access hypothesis). Clearly, cognates have
been very useful as tools to investigate the language (non)specificity of lexical
access in bilinguals (cf. Friel and Kennison 2001 for an overview). A number of
experimental studies have demonstrated processing differences between cognates
and other words in bilinguals, thus indicating differences in the representations
depending on the word type (cognate/noncognate). In many tasks that involve
speech production or recognition, cognates have a processing advantage over
non-cognate translation equivalents and interlingual homophones or homographs.
Worth mentioning here is the fact that the cognate facilitation effect has also been
reported in recent ERP studies. Midgley et al. (2011) recorded the electroencepha-
logram (EEG) of English-French bilinguals while cognates and matched control
words were presented in the participants’ L1 (English) and L2 (French) language
blocks. The findings led the authors to the conclusion that the mapping from form
to meaning is facilitated for cognates, and that access to the bilingual lexicon is
language non-selective. To provide background for the experiments described in
Chap. 5, the subsequent paragraphs will expand on issues pertaining to different
aspects tackled in the cognate studies outlining both the research methods they
used as well as the obtained empirical findings.
4.3.5.1 Association Tasks
In his article published in 1976, Taylor tried to answer the question whether the
similarity between French and English words is a factor to be considered in bilin-
gual language behaviour. For the purpose of the research, conducted within a con-
tinued free-association task paradigm, he selected a number of French and English
translation equivalents which were also formally similar (i.e. in sound and mean-
ing). He noticed that
(…) in continued-word association to the two types of key words, French – English bilin-
guals produced different patterns of responses. More response words to the similar than to
the dissimilar key words tended to be translation equivalents between the two languages
(Taylor 1976, p. 85).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
112 4 Visual Word Recognition in Multilinguals
English and French cognate translations (e.g., carrot and carotte) elicited the same
primary associates (vegetable and légume, respectively) more often than non-
cognates did. Consequently, the response list for cognates had far more common
associates than the list of noncognates. Van Hell and De Groot (1998a) in their
study on Dutch-English bilinguals used a similar methodology and received simi-
lar results. Again, the participants produced associations to cognates faster than
to noncognates. Associations were also faster to concrete than to abstract words,
and faster to nouns than to verbs. The other finding was that within- and between-
language associations were more similar for cognates, concrete words and nouns.
Essentially, in both experiments (Taylor 1976 and Van Hell and De Groot 1998a),
cognates yielded far more associates than noncognates. For cognates, associ-
ates were more often translation equivalents than they were for noncognates.
Additionally, associates for cognates were far easier to generate than those for
noncognates. The presented findings were taken as evidence that the concep-
tual representation in bilingual memory depends, at least to some extent, on the
similarity of form. On the basis of their results, the authors suggested also that
those wordtypes for which association was faster (concrete words, nouns and cog-
nates) share more conceptual features. Nevertheless, Sherkina (2003, 2004) doubts
whether this has to be always true for cognates. She supports her claim with three
significant postulates. Firstly, there are cognates which share some of their mean-
ings between two languages, but not all. Secondly, there are cognates which have
more conceptual features in one language than in the other; that is, they have a
narrower meaning in one language than in the other. Moreover, there are non-
cognate translation equivalents which share most or even all conceptual features,
such as words for certain animals and plants, some scientific terms, and calques.
Accounting for the above, Sherkina posits that “the difference between cognates
and noncognates in the word association task is more likely to occur in the recog-
nition component than in the association part” (Sherkina, 003, p. 137).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.3 Factors Affecting Visual Word Recognition in Bilinguals 113
By the same token, some previous experiments proved that translation times for
cognates may be equal regardless of translation direction, whereas for noncognates
the results are different. In cued translation tasks, for instance, L2 to L1 transla-
tion is faster than L1 to L2 for noncognates (cf. De Groot et al. 1994; Costa et al.
2000). Also in translation recognition tasks (in which pairs of words from two lan-
guages are presented and participants need to decide whether the words form a
translation pair) bilinguals are found to make faster and more correct decisions for
cognate pairs than for noncognate pairs (cf. De Groot and Comijis 1995). More
recent translation experiments also lend support to the special status of cognates.
For example, Yudes et al., in their 2010 study on cognate effects in bilingual lan-
guage comprehension, found that in the translation decision task, faster and more
accurate responses were associated with cognates.
4.3.5.3 Priming Tasks
In many reaction time (RT) studies, which measure the CFE by comparing RTs for
cognates and matched language-specific control words, cognates were responded
to faster than control words that exist in only one language. This has since long
been established by studies on bilingual word recognition in the visual modal-
ity (cf. Caramazza and Brones 1979; Cristoffanini et al. 1986; De Groot and Nas
1991; Sánchez-Casas et al. 1992; Kroll and Stewart 1994; Dufour and Kroll 1995;
Dijkstra et al. 1998; Dijkstra et al. 1999; Schwartz and Kroll 2006; Schwartz et al.
2007; Voga and Grainger 2007; Lemhöfer et al. 2008). However, more recently,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
114 4 Visual Word Recognition in Multilinguals
the effect has also been observed in the auditory modality (cf. Marian and Spivey
2003) and in word production (cf. Costa et al. 2000; Costa et al. 2005).
Apparently, RT studies involve a variety of experimental paradigms. One of
the most typical is a lexical decision task during which participants are asked to
decide whether a letter string forms a word in either of the languages involved
in the experiment. A substantial body of evidence with monolingual subjects has
revealed that the time needed to recognize a word is influenced by the character-
istics of words that are formally similar to the stimulus (cf. Grainger and Dijkstra
1992; Rastle and Davis 2008). Similarly, in bilingual systems, both lexicons might
be initially activated while an extended cross-lexical search appears to be con-
ducted on the basis of the formal features of the input (Smith 1997). Since the
lexical decision task methodology has been used in a number of visual lexical rec-
ognition studies, it seems justified to take a closer look at some selected experi-
ments conducted within this research paradigm. Thus, the subsequent paragraphs
will centre around visual lexical decision experiments devised to corroborate the
facilitatory cognate effect.
Gerard and Scarborough (1989) conducted a visual lexical decision task with
proficient English-Spanish bilinguals whose dominant language was English. The
input set consisted of Spanish or English lexical items that were noncognates in
the two languages (e.g., dog—perro), or cognates (e.g., actual—actual, having
the same meaning), or homographic noncognates (e.g., red—red, meaning net in
English). The input stimuli (single words) were presented in two different condi-
tions; (1) as Spanish items, (2) as English items. The results of the test showed
that bilinguals generally processed L2 (Spanish) information more slowly than L1
(English) lexical items, but otherw there were no effects of formal similarity; the
bilinguals appeared to be functioning like monolinguals. However, a repetition of
the same orthographic pattern across languages produced facilitation for cognates
as well as for homographic noncognates. Gerard and Scarborough explained these
results claiming that “the cross-language repetition effect appears to be a general
effect of encoding the same orthographic pattern” (Gerard and Scarborough 1989,
p. 312).
A similar cognate advantage was identified in lexical decision tasks with bilin-
guals by Cristoffanini et al. (1986). They found a 100-ms advantage for cognates
in English-target lexical decisions with Spanish-dominant bilinguals. They argued
that a possible explanation of the cognate advantage is that both the English and
Spanish lexical entries of a cognate are activated. Given that lexical access is
faster in the bilingual’s dominant language, the lexical entry in that lexicon will be
accessed first and will prime the corresponding lexical entry in the other language
lexicon (cf. Solomyak and Marantz 2009).
A comparable facilitation effect was obtained in Dijkstra et al.’s (1999) visual
word recognition study. Their stimuli included cognates and interlingual homo-
phones/homographs with different degrees of phonological and orthographic
similarity. Therefore, there were 6 conditions (“S” stands for semantically simi-
lar, “O”—orthographically similar, and “P”—phonologically similar): SOP, SO,
SP (cognates); OP, O, and P (homophones or homographs). Each test word was
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.3 Factors Affecting Visual Word Recognition in Bilinguals 115
matched in target language frequency and in word length with a control word
which was completely dissimilar to its translation equivalent. In the first experi-
ment, the task was progressive demasking (subjects had to identify a word which
alternated with a mask); in the second, the task was lexical decision. Cognates
were recognized faster than noncognate control words. Reaction times for SOP,
SO and O were faster than for their control words, for SP—not different from, and
for P and OP—slower than control words. The results showed that semantic and
orthographic overlap led to faster RT, but phonological overlap slows them down.
Interestingly enough, a positive correlation between phonological similarity and
RT was found. This was taken to explain why, in a number of recognition studies,
cognates have a processing advantage, but interlingual homophones do not. Access
to the former is facilitated by at least semantic overlap, as well as, in the case of
homographic cognates, by orthographic overlap; the effect of phonological overlap
is not strong enough to neutralize the facilitation from two sources. Interlingual
homophones/homographs do not have semantic overlap, and, as Dijkstra et al.
(1999, p. 510) put it, “facilitation from orthography and inhibition from phonology
cancel each other”.
In the previous subchapter an eclectic look was taken at factors affecting visual
word recognition in bilinguals with the increased focus on the special status of
cognates resulting in their facilitatory effect. In the subsequent sections, it seems
reasonable to take a closer look at factors affecting cognate recognition which
include cross-language similarity, task-dependency, and word frequency.
Many linguists (cf. Lemhöfer and Dijkstra 2004; Lemhöfer et al. 2004; Solomyak
and Marantz 2009) point to the fact that cognates overlapping in form trigger
faster recognition. A question arises, however: how does the cognate facilita-
tion effect depend on cross-linguistic similarity? More precisely, what happens
if cognates are presented that are nonidentical in the language pair the bilingual
knows? For instance, are German-English bilinguals affected in their recognition
of the German word Nacht by its similarity (but nonidentity) to the English word
night? If form-identity is not required, how much form-overlap between the two
readings of a cognate will still induce a facilitation effect? The answer to these
questions can be found in the experiment conducted by Van Hell and Dijkstra
(2002) who had trilinguals with Dutch as their native language, English as their
L2 and French as their L3 perform a word association task or a lexical decision
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
116 4 Visual Word Recognition in Multilinguals
task in their L1. Stimulus words were (mostly) nonidentical cognates such as bak-
ker (English baker, French boulanger) and droom (English dream, French rêve).
Shorter association and lexical decision times were observed for Dutch-English
cognates than for noncognates. For trilinguals with a higher proficiency in French,
faster responses in lexical decision were found for both Dutch-English and Dutch-
French cognates. The conclusion was that even when their orthographic and pho-
nological overlap across languages is incomplete, cognates may be recognized
faster than noncognates.
Similar results have been obtained by Dijkstra et al. (2010) who examined the
influence of phonological and orthographic overlap of cognates in languages that
share the same alphabet. The results of the study were interpreted as suggesting
that the occurrence and the directionality of cross-language form similarity effects
may depend on task demands as well as cognate type. The study analyzed how
cross-linguistic similarity affects Dutch-English identical and nonidentical cognate
visual recognition using different types of lexical decision tasks. Performing one
of three tasks, Dutch-English bilinguals processed cognates with varying degrees
of form overlap between their English and Dutch counterparts (e.g., lamp-lamp
vs. flood-vloed vs. song-lied). In lexical decision, reaction times decreased going
from translation equivalents without any cross-linguistic orthographic overlap to
very similar but non-identical cognates. In turn, identical cognates showed a big
discontinuous processing advantage and were additionally subject to facilitation
from phonological similarity. In language decision, the effect of orthographic sim-
ilarity reversed: a cognate inhibition effect arose, the size of which increased with
orthographic similarity. Here identical cognates were markedly slower than other
cognates. In progressive demasking, no orthographic similarity effect was found
for non-identical cognates, but a semantic similarity effect arose. In addition, there
was a facilitation effect for identical cognates of low English frequency. More
importantly, for identical cognates, and only for identical cognates, a marked facil-
itatory effect of phonological similarity was present. This finding is in line with
the previous conclusions, according to which cognates with high orthographic
similarity across languages were named faster when they had a highly similar pho-
nological code than when they were more distinct. This contention was further
confirmed in yet another lexical decision task study. Schwartz et al. (2007) asked
English-Spanish bilinguals to name English and Spanish noncognate words as
well as cognate words that varied in their degree of phonological and orthographic
overlap. The results showed that cognates with high orthographic and phono-
logical overlap (O+P+) were named faster than cognates with high orthographic
overlap but low phonological similarity (O+P−). However, when the orthographic
form of cognates was different (O−P+ and O−P−), the effects of phonology
were not statistically reliable. The authors interpreted the results as evidence that
cross-language activation forwards from orthography to phonology. Similarly to
Dijkstra et al. (1999), Schwartz et al. showed that the facilitatory effects typically
related to lexical overlap across languages can be ‘reduced or turned into inhibi-
tion when there is not a consistent mapping across all codes’ (2007, p. 122).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.4 Factors Affecting Cognate Recognition 117
Both Schwartz et al.’s (2007) as well as Dijkstra et al.’s (2010) studies provided
evidence that is consistent with a localist connectionist view of bilingual mem-
ory. In view of this proposal, the cross-linguistic similarity of cognates leads to
a greater semantic activation, since the associated meaning receives activation
from two lexical representations rather than one, as it occurs with noncognates.
However, the degree of semantic activation depends on the orthographic and pho-
nological similarity of cognates due to the existence of inhibitory connections
between lexical representations. Consequently, the speed and accuracy of cognate
reading depends on their cross-linguistic similarity (identical and non-identical
cognates). Accordingly, it seems reasonable to believe that greater overlap causes
more facilitation. A number of experiments show that the degree of semantic and
phonological overlap positively correlates with the magnitude of effects caused by
sharing (cf. Dijkstra and Van Heuven 2002a, b; Lemhöfer et al. 2008; Solomyak
and Marantz 2009; Dijkstra et al. 2010).
Summing up, many language recognition studies confirm the assumption that
even not completely identical but only orthographically similar cognates yield
comparable effects. Based on the results of the studies presented above, it can be
concluded that only partial similarity is sufficient and that orthographic identity or
maximal phonological overlap are apparently not required for the cognate facili-
tation effect to arise. Importantly, recent work has shown that even for bilinguals
whose languages do not share script (e.g., Korean-English, Greek-French, Hebrew-
English, Japanese-English), cognate facilitation effects can be observed (cf. Gollan
et al. 1997; Voga and Grainger 2007; Kim and Davis 2003; Hoshino and Kroll
2008). The very recent research by Allen and Conklin (2013) more fully explores
what underpins the cognate ‘advantage’ in different script bilinguals (Japanese-
English). Instead of using the more traditional binary cognate/noncognate distinc-
tion, the authors used continuous measures of phonological and semantic overlap,
L2 proficiency and lexical variables (e.g., frequency). The results of the lexical
decision task demonstrated that increased phonological similarity (e.g., bus/basu/
vs. radio/rajio/) leads to faster response times.
On the surface, the word frequency effect (also WFE) resembles the cognate
facilitation effect to a significant extent. Both yield shorter naming latencies (for
high-frequency words and cognates), and, unlike in priming, this is not caused
by previous access to these or related words since both effects appear even after
repeated access to experimental items (cf. Jescheniak and Levelt 1994; De Groot
and Keijzer 2000; Duyck et al. 2008 on the WFE, Costa et al. 2000; Van Assche
et al. 2009; Dijkstra et al. 2010 on the CFE). Naturally, the question arises whether
these two effects might be related. Or, as Sherkina puts it: “might they be manifes-
tations of the same effect, within and across languages?” (2003, p. 142).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
118 4 Visual Word Recognition in Multilinguals
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.4 Factors Affecting Cognate Recognition 119
that cognate facilitation is larger for cognates with a low compared to cognates
with a high frequency in the target-language. To explain the differential effects of
both cognate readings Peeters et al. (2013) proposed that identical cognates would
be represented twice in the bilingual brain. These two representations can be char-
acterized by their own language-specific frequency and “can be linked to their own
language-specific plural markers, gender, and syntactic category (Peeters et al.
2013, p. 329).
Most researchers are now convinced that there is a common conceptual store for
all languages of a bilingual (or a multilingual), the contention which has been
demonstrated in many experimental studies reviewed in Chap. 3. However, as
Dijkstra et al. (2010) rightly observe, the biggest challenge for a model of the
multilingual lexicon in general is that it should advance a rational explanation for
both cross-language effects and the ability of multilinguals to use one language at
a time. As has been reviewed in the previous sections, cognates more often give
different results in a number of experiments than noncognates. Therefore, many
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
120 4 Visual Word Recognition in Multilinguals
psycholinguists propose that lexical representations for cognates are stored differ-
ently than those for other words. In order to be able to account for cognate effects,
a multilingual mental lexicon model should allow sharing both conceptual and
lexical elements between two lexicons at some level. Theoretically, non-sharing
alternatives are possible, but in this case there should be some special process-
ing mechanisms, and, as Sherkina (2003) points out, a major challenge for them
would be in accounting for the frequency effects in the processing of cognates.
Over the years, several theoretical accounts attempting to answer the question of
how models of bilingual lexicon accommodates the cognate effect have been pro-
posed. In the subsequent sections, four theoretical positions discerned by Dijkstra
et al. (2010) will be presented with a special focus on the aspects of both cross-
linguistic similarity and word frequency.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.5 Models of the Bilingual Lexicon… 121
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
122 4 Visual Word Recognition in Multilinguals
However, more difficult to explain by this account are the cognate facilitation
effects observed by Voga and Grainger (2007). In their Greek-French masked
priming task, overlap-dependent facilitation effects arose for Greek-French cog-
nate primes relative to unrelated control primes (cf. Voga and Grainger 2007;
Experiments 2 and 3). It is hard to see how L1-L2 word associations could be
directly affected by (phonological) form overlap when the two languages involved
are characterized by different scripts. In addition, as Dijkstra et al. (2010) rightly
argue, a task account is badly needed to explain why lexical decision, language
decision, and progressive demasking lead to different result patterns for associa-
tively linked cognate representations. In contrast, somewhat distinct predictions
are proposed with respect to the role of cross-linguistic similarity and word fre-
quency by distributed connectionist models such as proposed by Thomas and
Van Heuven (2005); French and Jacquet (2004), and Li and Farkaš (2000), and
by localist connectionist models such as BIA (cf. Dijkstra, van Jaarsveld and
ten Brinke 1998; Voga and Grainger 2007) and BIA+ (Dijkstra and Van Heuven
2002a, b). Consequently, in the light of the research, the model promoting the
associatively linked orthographic representations of cognates seems to be highly
unlikely.
A third theoretical proportion trying to account for the cognate facilitation effect
is that of distributed connectionist model (cf. De Groot 1992; Thomas 1997;
French and Jacquet 2004; Thomas and Van Heuven 2005). Distributed mod-
els hold that the two readings for identical and nearly identical cognates share
most of the connections. Therefore, representations of these cognates in two
languages “lie closely together in multidimensional space” (Thomas and Van
Heuven 2005). Consequently, responses to such cognates should be faster than
for control words, because, “being in the same region, the attractors of the two
representations of a cognate exert a joint force” (Dijkstra et al. 2010, p. 11).
Responses to cognates with decreasing levels of similarity should gradually
become slower because the joint force decreases. Additionally, a sharp rise in
RTs would not be expected when one compares form-identical to slightly non-
identical cognates.
According to Thomas and Van Heuven (2005), in distributed connectionist
models, word attractors lying in the same area of a multidimensional space set up
by stimulus dimensions exert a joint “pull” during the word recognition process.
Co-activated cognate readings lead therefore to faster RTs than language-specific
control words resulting in the cognate facilitation effect. Furthermore, because the
joint attractor force is larger, the closer the two representations are, the faster RTs
for these cognates. Finally, as observed by Thomas and Van Heuven (2005) and
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.5 Models of the Bilingual Lexicon… 123
Dijkstra et al. (2010), a higher frequency of use might strengthen the connections
between representations and allow faster movement through multidimensional
space. As confirmed by numerous experiments (cf. Thomas 2002; Thomas and Van
Heuven 2005; Dijkstra and Van Heuven 2002a, b and Dijkstra et al. 2010), the dis-
tributed connectionist model, as presented above, offers a convincing explanation
to the cognate facilitation effect. Nevertheless, it is only the localist connectionist
model (BIA and later BIA+) that managed to account for the cognate effect occur-
ring under different task demands.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
124 4 Visual Word Recognition in Multilinguals
Worth mentioning here is the fact that while early research on the represen-
tations of cognates (cf. Dijkstra et al. 2010) allowed for the possibility of the
existence of a single symbolic representation for identical cognates with two rep-
resentations for nonidentical cognates, in their recent study, Peeters et al. (2013)
basing on the behaviourally and electrophysiologically gathered data, developed
the model in which identical cognates have, at some level (e.g., morphology), two
form-related representations and a (partially) shared semantic representation. In
their proposal identical cognates have two representations at the morphological
level, in between a shared orthographic form and a shared semantic representation
concluding that,
Although prima facie it might not seem optimally efficient to have two form-related
representations for the same word, when we keep in mind that identical cognates can
have different plural markers, gender and/or relative frequencies across languages and
are generally learnt in a different context (L1 at home, L2 at school/abroad), this might
nonetheless be the way such identical words are represented. The language-specific plural
markers, gender, and relative frequencies could then be related to a language-specific
morpheme for both readings of the identical cognates (Peeters et al. 2013, p. 331).
As has been shown above, the majority of studies using interlingual homo-
graphs, interlingual neighbours or cognates in relation to visual lexical recogni-
tion focused on isolated words. They all show that lexical representations from
both languages are activated when reading in one language (language non-selec-
tive lexical access). However, more recently the three types of words have also
been used to analyse lexical recognition in sentential context (cf. Elston-Güttler
et al 2005; Van Assche 2012). In other words, a question has been posed of
whether language non-selective access generalizes to word recognition in sen-
tence contexts which provides not only a language cue but also semantic con-
straint for upcoming words.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.6 Bilingual Word Recognition in a Sentence Context 125
One of the early hypotheses assumes that the presentation of words in a sen-
tence context restricts lexical activation to words of the target language only. As
Van Assche et al. (2013) aptly state, this would actually be “quite an efficient strat-
egy to speed up word recognition, because it reduces the number of lexical can-
didates” (2013, p. 3). Moreover, evidence coming from the monolingual context
shows that semantic and syntactic restrictions imposed by a sentence are used to
speed up recognition of upcoming words. For instance, in their study with ambigu-
ous words Binder and Rayner (1998) proved that monolingual sentential context
affects lexical access of such words. Still, the results of many visual recognition
studies conducted within the bilingual domain suggest that the language of the
preceding words is insufficient to restrict lexical access to words of the target lan-
guage, even when reading in the native language (cf. Schwartz and Kroll 2006;
Duyck et al. 2008; van Hell and De Groot 2008; Titone et al. 2011; Van Assche
et al. 2009, 2011; Bultena et al. 2013). As Van Assche et al. (2013) conclude in
their comprehensive review of the studies on bilingual word recognition in a
sentence context,
mixed results have been obtained for semantically constraining sentences, but recent studies
using time-sensitive eye movement recordings suggest that even a strong semantic context
does not necessarily eliminate cross-lingual activation effects, at least for early interaction
effects reflected in early reading time measures (Van Assche et al. 2013, pp. 3–4).
Beyond doubt, one of the aims of second and subsequent language acquisition
studies should be exploring processes responsible for mastering the foreign
language lexical system with a view to identifying effective solutions which
are transferable to the language classroom. Thus, an important research area
that awaits a more thorough exploration is the role of the facilitatory cognate
effect in the process of foreign language acquisition. So far, findings from many
studies in line with evidence from psycholinguistic research show a cognate
facilitation effect in L2 word learning that makes cognates easier to learn and
remember than non-cognate translations. It has been confirmed that the formal
similarity of cross-language cognates provides means for language learners to
connect L2 and L1 lexical forms and thus to process the higher level represen-
tations such as concepts or mental images through the L1 word form and not
directly from the L2 lexical representation in word recognition. Analogously, it
has been proved that language learners can activate L2 lexical forms through its
L1 similarity neighbour in word production. Results of numerous studies indi-
cate that cognates provide learners with a number of significant advantages. By
way of illustration, Lotto and De Groot (1998) found that cognates and high-fre-
quency words were easier to learn than noncognates and low-frequency words,
whereas De Groot and Keijzer (2000) proved that cognates and concrete words
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
126 4 Visual Word Recognition in Multilinguals
were easier to learn and less susceptible to forgetting than noncognates and
abstract words. Also, Ellis and Beaton (1993) corroborated the assumption that
“native-to-foreign learning is shown to be easier the more the FL words conform
to the phonological and orthographic patterns of the native language” (1993, p.
559). Friel and Kennison in their state-of-the-art article on cognates in foreign
language learning offer a comprehensive summary of cognate-induced advan-
tages claiming that,
participants with no prior knowledge of the to-be-learned language recall a higher per-
centage of cognates than noncognates, achieve a high level of recall in fewer learning
sessions for cognates than noncognates, and yield faster response latencies in translating
cognates than they do for noncognates (Friel and Kennison 2001, p. 249).
In this connection, Lotto and De Groot (1998) and De Groot and Keijzer (2000)
suggested a possible source for the superior foreign language vocabulary learn-
ing performance for cognates, considering both the learning stage (storage) and
the testing stage (retrieval) as possible loci of the effect. Their explanation extends
the view of bilingual memory representation that assumes shared representations
for cognates, but language-specific representations for noncognates (cf. Sánchez-
Casas et al. 1992; cf.). If true, the learning of a FL word that shares a noncognate
relation with the corresponding L1 word involves creating a new entry in memory,
whereas learning a cognate word may only involve adding new information to, or
adapting, a representation already stored there prior to the learning episode. The
latter process may be less demanding than the former, causing the learning advan-
tage of cognates over noncognates. As is apparent from the above review, findings
of many studies actually lead to the conclusion that cognates can be a significant
source of positive transfer and can facilitate vocabulary acquisition in L2 as they
offer “an encouraging springboard” (Aronin and Hufeizen 2009, p. 9) into learn-
ing of a new language (cf. Nation 1990; Odlin 1989; Ringbom 2001, 2007; Aronin
and Hufeizen 2009).
4.8 Conclusion
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
4.8 Conclusion 127
The above discussion of the special status of cognates both in modelling visual
recognition and in foreign language learning and teaching has not included the
multilingual context. The following chapter will be devoted solely to the descrip-
tion of empirical experiments conducted in the multilingual, or more specifically
trilingual, domain. The chapter will address the question of whether the cognate
facilitation effect extends to three languages and whether cognates in three lan-
guages lead to stronger facilitation effects (cf. Lemhöfer et al. 2004; Dijkstra et al.
2010; Szubko-Sitarek 2012).
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Chapter 5
Exploring the Cognate Facilitation
Effect in Multilingual Word Recognition
5.1 Introduction
The foregoing chapters have been devoted to the discussion of various issues related
to multilingual processing. Theoretical in nature, they have concentrated on three
main areas: the presentation of various aspects of the multifaceted concept of multi-
lingualism, the discussion of various theoretical models developed to operationalize
the mechanisms responsible for mono- and multilingual lexical processing and the
delineation of different factors influencing the multilingual lexical access. Special
attention has been given to the account of the cognate facilitation effect occurring
in the bilingual mental lexicon. The present chapter, in turn, makes an attempt to
verify the assumptions concerning the storage and processing of cognates in the
context of the trilingual lexicon. More specifically, it will be devoted to the presen-
tation of the findings of a series of experiments aimed at verifying the existence of
the cognate facilitation effect in the trilingual lexical processing.
As transpires from the foregoing overview of different lexical units used in the
research on the multilingual storage, access and retrieval, cross-language cognates
are of special interest for designing a model of the multilingual lexicon since, as
many psycholinguists claim, there is a possibility that at least part of their repre-
sentations is shared between languages. Hence the present chapter will be entirely
devoted to the analysis of the special position of cognates in the lexical retrieval in
the multilingual mind. The first part of the chapter reports on the outcomes of the
off-line studies whose aim was the analysis of the perception of cognates from the
form similarity perspective. The outcome of the study is a list of German-English
cognates as identified by the speakers of Polish (cf. Appendix A and B). A selection
of items from the list created in these two tasks was used as stimulus material in a
set of four experiments conducted with a view to exploring the language-specific or
language-integrated nature of trilingual lexical processing by examining the issues
connected with the lexical organization of cognates. More specifically, the research
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
130 5 Exploring the Cognate Facilitation Effect …
The first step in the study, whose experiments are to be described in the present
chapter, was to create a reliable corpus of German-English cognate vocabulary. To
achieve this goal an empirical means of identifying cognate words from the psycho-
linguistic perspective had to be established. In one early research study of cognate
effects (Taylor 1976), the experimenter herself decided which words were “similar”,
“dissimilar”, and “misleading” cognates, and only used items that, in her judgment,
fell unambiguously into one of the first two categories. However, this subjective
procedure made the results of her experiments unreliable. Almost two decades later,
two more objective techniques for identifying cognates were tested. First, De Groot
and Nas (1991) proposed a similarity rating task, in which lists of translation pairs
were given to Dutch-English bilinguals, who rated the similarity of the word-pairs
on a 7-point scale. Mean ratings across a pool of participants were used to confirm
the experimenter’s own classification. A totally different procedure, widely known
as a translation-elicitation task, was introduced by Kroll and Stewart (1994) and
Dufour and Kroll (1995). The researchers gave English speakers a list of written
words in the foreign language (Dutch) asking them to guess each word’s English
translation. The idea was that a cognate word in a foreign language must formally
overlap with its native-language translation equivalent to trigger the meaning.
A contextual cue was provided, with items grouped together under their category
headings (e.g., vegetables, clothing). Those words which were correctly translated
by more than 50 % of the participants were designated as cognates.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.2 Methods for Establishing Cognate Status … 131
5.2.1 Similarity-Rating Study
The first study was modelled on the similarity-rating technique, initially created by
De Groot and Nas (1991) and later developed by Dijkstra et al. (2010). The aim
of the study was to construct a list of words on a continuum ranging from items
identical in form and meaning, via translation equivalents that moderately resem-
ble each other with respect to orthography and phonology, to words that have a
completely different form in two languages. The participants were instructed to
base their ratings on overlapping sound and appearance between translation pairs.
The outcomes of the experiment were juxtaposed with the list of L2–L3 cognates
pre-selected by the author herself. Mean ratings across a pool of participants were
used to confirm the experimenter’s own classification. Additionally, the study was
meant to check whether the correlation between word orthographic (O) and pho-
nological (P) aspects, confirmed for L1–L2 lexical relation was also valid for
L2–L3 connection. The research was carried out in two stages. The first stage
involved the selection of the German-English word-pairs and the creation of a ran-
domised list of word-pairs. The following stage involved the collection of data.
5.2.1.1 The Participants
Ninety-eight students took part in the first stage of selection. They all filled biodata
charts, where they were asked to declare what languages they had learnt. For the
next stage only those were selected who declared at least B1 proficiency in German.
Sixty two trilingual subjects (mean age 23.4 years, SD 2.1, 48 women 14 men) took
part in the second stage of the experiment. They were all 2nd and 3rd year students
of English Department at the University of Łódź and the Higher Vocational State
School in Włocławek. First, the participants completed a language history question-
naire in which they were asked to self-assess their proficiency in the foreign lan-
guages they knew. The results of the questionnaire are listed in Table 5.1. All the
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
132 5 Exploring the Cognate Facilitation Effect …
participants had Polish as their native language; they had had experience with the
English language for at least 7 years (mean 8.75, SD 2.97) and with German for at
least 3 years (mean 4.23, SD 2.43). They were not paid for participating, nor did
they receive extra course credits.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.2 Methods for Establishing Cognate Status … 133
bilingual context (Friel and Kennison 2001) where the subjects rated word-pairs con-
taining characters not found in their L1 to be less similar than those without these
characters.
5.2.1.3 Procedure
Lists from 5 subjects had to be rejected due to too high a number of not completed
answers. To achieve the same number of answers on each aspect (O and P), the data
for one more participant were not included in the statistical analysis. Additionally,
items that were inadvertently skipped by the participants, as well as their matched
partners, were eliminated from the analysis resulting in a loss of less than 5 % of the
data. Mean similarity ratings for O and P were calculated for each word pair showing a
similar rating pattern as reflected in the high correlation (r = 0.82, p < 0.001) between
them. These results seem to suggest that from a language learner perspective, an L2–
L3 word pair with a highly similar orthography is also generally perceived as the one
having highly similar phonology and vice versa. Interestingly enough, the obtained
pattern of results was similar to the one for L1–L2 pairs. Both orthographic and pho-
nological aspects turned out to be important in the process of rating similarity between
foreign language items. It needs to be noted, however, that the similarity rating for P
was calculated on the basis of the word’s written form, which might have influenced
the judgment. Clearly, the spoken version of the similarity-rating study is needed to
establish the factual role of phonology in the process of identifying cognate pairs.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
134 5 Exploring the Cognate Facilitation Effect …
As indicated by the data accrued in the course of this experiment, there was no
significant correlation between test items O- or P-rating scores and their English
frequency, as confirmed by correlations of 0.1 and −0.04, respectively. This sug-
gests that the participants were not influenced by the English frequency of the lexi-
cal items while they were judging the orthographic or phonological similarities of
the word pairs which seems to be justifiable considering the fact that English was
the participants’ second language and the proficiency level clearly did not allow for
this factor to influence the ratings as it happened to be the case in the research (cf.
De Groot and Nas 1991; Friel and Kennison 2001) where participants judged the
similarity of word pairs where one item came from their native language. Both the
stimuli and ratings for the list of words are tabulated in Appendix A.
As noted before, an additional aim of the presented experiment was to determine
whether words containing German-specific characters (i.e. umlauted vowels) were
responded to differently than other words. Table 5.2 displays mean similarity ratings
for the two conditions of this factor, hereafter referred to as the spelling factor.
The main effect of German-specific characters reached significance as con-
firmed by the analysis of variance (ANOVA). Similarity ratings for translation pairs
in which the German word contained a German-specific character were significantly
lower than for those in which the German word did not contain such characters
F(1, 98) = 908.11, p < 0.001. However, the spelling effect should be interpreted with
caution, as there were many more words without German-specific characters than
words with German-specific characters. Furthermore, the difference observed can
be due to some inherent distinctions between the two types of words. Namely, the
German words containing German-specific characters were lower in mean word fre-
quency than the words without such characters (39 vs. 96). It needs to be noted that
the words containing German-specific characters were not included in the following
multilingual visual word recognition experiments. However, a study employing such
words is needed to establish the role of diacritical marks both in rating the form simi-
larity and, more generally, in cognate processing.
5.2.2 Translation-Elicitation Task
To verify the validity of the similarity-rating task, another experiment was con-
ducted. As mentioned before, it employed the translation-elicitation paradigm, ini-
tially used by Kroll and Stewart (1994) and Dufour and Kroll (1995), described in
the introduction to this section. The difference again was the language combina-
tion. Unlike the original study where the participants basing on the formal resem-
blance translated foreign language words form a language they did not know into
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.2 Methods for Establishing Cognate Status … 135
their native tongue, in the experiment reported below the participants were asked
to translate unknown foreign language words into their L2 which they declared to
know only at the B1/B2 level. Consequently, to do the task they had to activate their
second language lexical store as the majority of German items did not bear any for-
mal resemblance to their Polish counterparts.
5.2.2.1 The Participants
The list of 100 German nouns, previously used in the similarity rating task,
was collated and grouped by category wherever possible (e.g., food; parts of
body). The items for which no clear category label could be given were sim-
ply grouped under the heading “Other Items”. All the words were presented in
the written form. The order of categories was random, as was the order of indi-
vidual words within each category. Additionally, mention must be made of the
fact that some of the selected German-English pairs bear some orthographic as
well as phonological similarity to their L1-Polish semantic counterparts. This
factor must have facilitated the participants’ choices, the outcomes of which will
be discussed in subsequent paragraphs. The subjects were asked to go through
the list of the German words one by one, and for each word, to write down what
they thought its English translation might be. They were encouraged to offer a
response for every word, even if they felt it was a complete guess, in preference
to leaving a blank. No guide to German spelling-sound relationships was offered.
Classifications were, therefore, based on orthographic rather than phonological
similarity. For practical reasons, the study was conducted in the paper-and-pencil
format. Omitted responses were scored as incorrect answers. Spelling errors
and plurals did not invalidate otherwise correct translations. The mean propor-
tion of correct responses was tallied for each item, and the results are tabulated in
Appendix B.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
136 5 Exploring the Cognate Facilitation Effect …
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.2 Methods for Establishing Cognate Status … 137
Another relevant factor was the choice of translation words. Admittedly, there is
more than one way to translate almost every word, especially when no context is
provided. In the present study, the responses were coded by the experimenter as cor-
rect or incorrect translations, based on whether the suggested translation would be
the first definition of a word in the English-German Duden-Oxford Dictionary.
Essentially, 47 word-pairs were deemed to be cognate, and 53 word-pairs were
identified as noncognate. As Table 5.4 shows, the cognate and noncognate sets of
words were comparable in their mean logarithmic frequencies2 and length in both
languages. As in the studies conducted for L1–L2 language combination, also in the
present experiment the cognate word-pairs were more similar in length across lan-
guages than the noncognate word-pairs which may be interpreted as reflecting the
importance of form similarity for the perception of cognates.
Basing on their research, Van Hell and De Groot’s (1998) indicated that cognate
word-pairs are closer than noncognate translation equivalents in meaning and there-
fore also in their context and frequency of use in the two languages. Interestingly,
this suggestion was also confirmed in the present study where the correlation
between word frequencies across languages was also somewhat higher for cognate
word-pairs. This finding, however, needs to be treated with due caution as it would
be difficult to prove whether the correlation was only dependent on L2 frequency.
Considering the selected items, all of them coming from basic vocabulary list, it
seems highly probable that L1 word frequency also played its role.
2 Sincethe scales of measurement differ for the German and English frequency norms, values
should not be compared directly. Therefore, logarithmic frequencies were introduced. All the
data concerning the logarithmic frequencies were taken from CELEX database.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
138 5 Exploring the Cognate Facilitation Effect …
selection process, two independent tests, briefly presented in the foregoing sections,
were conducted. It has to be noted that following this approach, it was not neces-
sary a priori to decide if a particular item was to be considered as a cognate or a
noncognate, thus avoiding artifacts in allocating particular items to one group or
the other. The results from the two experiments correlated highly (r = 0.85), with
cognates scoring a mean similarity rating of 4.27 on a 5-point scale, while noncog-
nates scored a mean of 1.62, and led to the compilation of a reliable list of German-
English cognates as perceived from the psycholinguistic perspective.
Last but not least, it seems reasonable to refer the present results to a project by
Friel and Kennison (2001) who conducted an extensive research study aimed at ver-
ifying the effectiveness of the two methods of identifying cognates, false cognates
and non-cognates. It needs to be stressed that as their project focused on L1–L2
cognate relations the level of similarity was rated exclusively from the native lan-
guage perspective. In turn, in the tests described above, respondents were asked to
rate and translate pairs of words deriving from foreign languages which primarily
necessitated the activation of the second language lexicon and interestingly did not
exclude the activation of the first language store. In the following paragraphs the
results of the present tests will be compared to the outcomes of Friel and Kennison
study.
Of the 35 stimuli common to the present sets and those used by Friel and
Kennison (2001), classifications agreed for 13 cognates and 22 noncognates. The
four discrepant items were: milk, frog, mouse and stone. Some points should be
made as to the source of the discrepancies. All of the items were clearly cog-
nate in the present study (with translation rates of at least 70 %), but noncognate
in Friel and Kennison’s study. All these words fall within the cognate range in
Friel and Kennison’s similarity ratings, confirming that they are quite similar in
for. Also, the correct translation was given for those items more often than any
other translation, but the rate simply failed to exceed 50 % which was the border
line. The possible reason may be connected with the fact that when no context
was provided, some participants were reminded of other orthographically similar
words instead. Providing a context—the category groupings in this study—was
apparently sufficient to favour the correct translation and consequently it facili-
tated the task. On the other hand, Friel and Kennison’s (2001) primary reason for
randomising, rather than categorising, their lists of stimuli was that many of their
items were abstract words which could not clearly be categorised. Undoubtedly,
the categorised items may in some sense have had an unfair advantage over the
“other” items in the presented study, not so much by a process of elimination as
Friel and Kennison suggest, because the lists were by no means exhaustive of cat-
egory items, but simply by favouring the potential translation which falls within
the category. Another difference between the studies is that all the items which
Friel and Kennison (2001) found to be cognates (i.e., items identified correctly by
more than 50 % of participants) did overlap in form. That is, there were no items
for which correct translations were often produced despite the lack of form simi-
larity, such as the Zitrone-lemon pair. The possible explanation for this discrep-
ancy was already discussed in the previous paragraphs. In sum, two previously
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.2 Methods for Establishing Cognate Status … 139
described methods for identifying cognates were tested in a new trilingual con-
text. They both were shown to yield similar results. The resulting database of
German-English cognates was further used in a set of experiments devised to ver-
ify the nonselective access hypothesis in the trilingual context. The results of the
experiments are reported in the following sections.
The previous chapters discussed problems related to models and hypotheses which
propose how lexical items are stored and processed by monolinguals, bilinguals
and, most importantly, multilinguals. As has been shown, the question of whether
words from different languages are stored in one or more containers has recently
been replaced by the issue of how words from different languages in a multilin-
gual person’s lexicon are accessed. The question of whether there is one single
integrated lexicon (cf. Cook 1996), two or more separate lexicons (cf. Smith
1997; Singleton 1999, 2002) or two or more subsystems under one big system
(cf. Paradis 2004) ceased to be in the centre of researchers’ attention. Instead, the
central question most of the multilingual research is concerned with nowadays is
language-selective versus language-nonselective access (cf. de Bot 2004, p. 17).
As has been documented in the theoretical chapters, the majority of empirical evi-
dence gathered in psycholinguistic and neurolinguistic studies seem to support
the contention that during bilingual lexical access, even if the two languages are
indeed represented differently, both are activated, although perhaps to different
degrees (cf. De Groot and Nas 1991; Dijkstra et al. 1999; De Groot et al. 2000;
Jared and Kroll 2001; De Groot 2002; de Bot 2004; Dijkstra 2007). According to
the BIA+ model (Dijkstra and Van Heuven 2002a), which constitutes the theo-
retical background for a number of lexical processing experiments conducted
within the bilingual domain, the visual presentation of a word to a bilingual per-
son leads to parallel activation of orthographic input representations both in the
native language (L1) and in the second language (L2). These representations then
activate associated semantic and phonological representations, leading to a com-
plex interaction (or resonance process; Dijkstra and Van Heuven 2002a, p. 183)
between codes from which the lexical candidate corresponding to the input word
emerges and is recognized. Analogously, in the case of trilingual lexical process-
ing research, many linguists (cf. Dijkstra 2003a, b; Li Wei 2003, 2006; Hall and
Ecke 2003; Dijkstra 2007) argue for a unified lexicon with language nonselective
access. Some of them, like Dijkstra (2003a, b, Experiment 1 in the present chapter)
focus on visual perception, whereas others, e.g., Li Wei (2003, 2006), deal with
production. Furthering the question of the nonselectivity of language access, it
needs to be stressed that it can be influenced and, as some linguists claim, lim-
ited by a number of factors. Indeed, previous investigations identified several vari-
ables influencing the lexical organization of non-native languages and the manner
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
140 5 Exploring the Cognate Facilitation Effect …
in which words in non-native languages (L2, L3, etc.) are linked to one another
and to their translation-equivalents in the native language (L1) (cf. De Groot and
Hoeks 1995; Grosjean 1997; Dijkstra et al. 1999; Grosjean 2008). In studying the
interaction of multilinguals’ languages at least three factors seem to influence the
strength of connections between orthographically and semantically overlapping
words. First, task demands and experimental stimulus materials. Second, the mul-
tilingual’s relative language proficiency. Third, the perceived similarity between
languages (psychotypological effect, cf. Kellermann 1983).
The question whether the three languages known to the learner simultane-
ously influence visual word recognition performance has been addressed in
Experiment 1. To test the influence of the above mentioned factors on the mul-
tilingual processing, three separate experiments (Experiments 2, 3 and 4) have
been conducted. As regards the first factor pertaining to the task demands and
the language of the stimulus material, there is clear evidence in the literature
that task demands can affect multilingual performance to a considerable extent.
Many researchers even claim that it may be inappropriate to talk about multilin-
gual word recognition in general without specifying the precise task and experi-
mental circumstances under which it takes place because performance is both task
and context dependent (cf. De Groot et al. 2000; Dijkstra et al. 2000; De Groot
et al. 2002; Dijkstra 2007; Lemhöfer and Radach 2009). Similarly, according to
Dijkstra’s BIA+ model, the result patterns for cognates will differ depending
on task demands. To verify this assumption for a trilingual lexicon, the results
of Experiment 2 will be contrasted with the data from Experiment 1 and will
undergo further analysis. Clearly, to achieve reliable results, in both experiments
the same group of stimulus words will be tested. The second factor in the dis-
cussion of whether knowledge of one language affects performance in the other
language concerns the relative language proficiency. According to Van Hell and
Dijkstra (2002, p. 783), “relative language proficiency will affect the bilingual’s
sensitivity to L1 interference when he/she is processing in L2, and the sensitiv-
ity to L2 interference when processing in L1”. The underlying rationale here
is that less activation is needed to recognize words that are used relatively fre-
quently, as are words in a language in which the speaker is relatively proficient.
Experiment 3 analyzes the role of the second foreign language proficiency on
the organization of cognates, which is studied in greater detail by manipulating
the subjects’ proficiency in the nontarget languages. In Experiments 1 and 2, tri-
linguals were most fluent in their L1 (Polish), less fluent in their L2 (English),
and least fluent in their L3 (German). In Experiment 3 trilinguals with a higher
level of proficiency in German were tested. These trilinguals’ proficiency levels
in L2 and L3, as declared by the participants themselves, were comparable. It was
expected that the level of proficiency may exert some influence on the strength
of interlingual connections. Additionally, numerous previous studies have shown
that processing in the weaker language, L2, can be influenced by knowledge of
the stronger language, L1 (e.g., De Groot et al. 2000; Dijkstra et al. 1998, 1999;
Van Heuven et al. 1998; Grosjean 2008). In a divergence from these studies,
Experiments 2 and 3 test the effects of L2 and L3 on L1 in an exclusively native
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.3 Investigating Cognate Effects in Trilingual Visual Word … 141
language context. The rationale behind these experiments is that a minimal level
of nontarget language proficiency seems to be required before any weaker lan-
guage effects become noticeable in L1 processing. Finally, the third factor that
influences the interaction of the multilingual’s languages to be focused upon in
this chapter, is the perceived similarity of the languages involved. Hall and Ecke
(2003) explained transfer-related mistakes by way of links which are formed
across languages at the word or concept level. They proposed that the links
are formed on the basis of similarity between items already well known to the
speaker (belonging to L1 and/or L2) and the newcomers from the L3. They also
suggested that more transfer between a speaker’s L3 and L2 than between his/her
L3 and L1 might be a reflection of a link between foreign languages in the mind.
Moreover, according to Singleton (2002, 2003), it seems plausible that in the case
when L3 is typologically closer to L2 than to L1, in the early stages of the pro-
cess of acquiring L3, L3 forms will be linked to L2 ones rather than to those of
L1. Singleton (2003) implies also that with growing L3 proficiency, strong lexical
connections develop between L2 and L3 items. The connections get even stronger
under the influence of psychotypology (cf. Kellermann 1983). Experiment 4 aims
to verify the role of perceived similarity supported by the L2 status factor in the
organization of the trilingual lexicon.
As regards the methodology, what is worth mentioning is the fact that the bulk
of the studies investigating transfer, which have been reported on in the TLA lit-
erature and used as a basis for models of multilingual mental lexicons, employed
off-line methodologies—pen-and-paper formula, such as e.g., questionnaires.
Such methodologies offer more time and thus allow for extensive use of strate-
gies by participants. In contrast, in the experiments reported in the present dis-
sertation, an on-line methodology was adopted since it enables the researcher to
investigate processing in real time. Unlike off-line studies, on-line experiments
do not allow for the same amount of conscious control and thus might be more
potent in terms of shedding light on the composition of languages in the mul-
tilingual mind. In this study, an experimental setup has been chosen for which
the most reliable and frequently replicated bilingual cognate effects have so far
been obtained, namely a lexical decision task. The logic in using this type of task
is that it requires individuals to search their lexicons for a lexical representation
that matches the letter string presented. The representation of a lexical item con-
tains information regarding the word’s orthography, phonology, and semantics,
and these aspects of the word are retrieved during the task. Finally, it needs to be
pointed out that Experiments 1–4 were conducted in the order of presentation and
each next experiment was designed on the basis of the conclusions drawn from
the previous one. In the subsequent paragraphs, a more detailed outline of the
rationale behind the design of the experiments as well as details concerning the
participants, materials, procedures and results will be provided, Next, a general
discussion will follow.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
142 5 Exploring the Cognate Facilitation Effect …
5.4.1 Research Hypotheses
First, it was expected that the participants would react faster and more accurately
to German-Polish cognates than to German control words. Such a finding would
replicate the standard cognate effect (cf. Djikstra and Van Heuven 2002a, b) for a
new language combination and provide additional evidence in support of language
nonselective access. Second, an even stronger view of language nonselective access
would be supported if three languages at a time can influence word recognition. In
that case, the cognate status of the stimuli with respect to English should have an
additional effect on top of the standard cognate effect. In other words, the recog-
nition performance for German-Polish-English cognates should be even faster and
more accurate than that for German-Polish cognates. If that turned out to be true,
the experiment would replicate the Lemhöfer et al. (2004) study conducted for a
different combination of languages and add further evidence to the discussion
on the nonselective visual word recognition in the multilingual mental lexicon.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.4 Study 1: Exploring the Cognate Effect in Trilingual … 143
Specifically, the study reported in the present section sought to address the follow-
ing hypotheses:
H1. If there is language nonspecific selection in lexical access, faster RTs will be
found for Polish-German cognates than for German control words.
H2. If language nonspecific selection in lexical access extends to three languages,
RTs found for “triple” Polish-English-German cognates will be even shorter
than those obtained for “double” Polish-German cognates.
5.4.2 The Participants
The experiment involved 27 trilinguals with Polish as their mother tongue (L1),
and English and German as their second (L2) and third language (L3), respec-
tively. All the participants were “unbalanced” trilinguals; i.e., they were not as
proficient in their second and third languages—English and German—as in their
mother tongue—Polish. All of the testees were students (second- and third-years)
in the Institute of English Studies at the University of Łódź. All of them had learned
English and German as a foreign language at school. English from about the age
of 9, German from about the age of 14. The participants’ L2 competence was not
tested before the study. It was assumed that their status of students of English
Philology ensured advanced L2 proficiency. Before the experiment, however, a lan-
guage background questionnaire was administered, in which, 31 prospective partici-
pants were asked to report on their competencies in all of their foreign languages.
4 students had to be excluded from the experiment due to their relatively advanced
knowledge of Spanish that could affect the results. Additionally, a few respond-
ents declared some competence in French, Italian or Russian, but those selected
for the experiment described their competence as very low. To ensure that the tes-
tees constituted a homogenous sample as far as their L3 (German) competence is
concerned, only those students were selected who described their L3 level as pre-
intermediate (A2) or intermediate (B1), as specified by the Common European
Framework for Languages descriptors (henceforth CEFR). A summary of the par-
ticipants’ learning history is given in Table 5.5.
The data from 3 participants had to be excluded due to their faulty performance,
which sets the mortality rate of the experiment at 11 %. The remaining 24 partici-
pants were between 20 and 24 years old with the mean of 21.66. 17 were female,
7 were male. All the participants were right-handed. They were not paid for their
participation, nor given any course credits.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
144 5 Exploring the Cognate Facilitation Effect …
5.4.3 Procedure
The experimental session took about 30 min during which the participants carried
out a German lexical decision task involving triple Polish-English-German cognates
(hereafter referred to as PGE cognates), double Polish-German cognates (PG cog-
nates) and German control words. Testing took place individually on a PC; a modi-
fied program based on the Reaction Time Instrument Builder3 was used. The
subjects were seated at the 17-in. computer screen, where stimuli were presented in
black 14 point uppercase TIMES NEW ROMAN on a white background. One but-
ton on the side of the dominant hand of the participants was assigned to the “yes”
response, the other button to the “no” reaction. At the beginning of each trial, a fixa-
tion point appeared in the middle of the screen for 1 s, then a blank interval of
500 ms followed. Next, the test word appeared. The item stayed in view until a
response had been provided or until a time-out of 5 s had passed. The next trial was
started 500 ms after the response was given. All items were presented in uppercase
letters, because in German the case of the first letter can be a cue for the syntactic
class of a word (nouns are written with a capital). The experiment consisted of two
blocks of 48 items each. The first two items of each block were warm-up items (a
nonword and a German filler word) which were not included in the analyzes. The
participants took a short break (5 min) between the blocks. The order of items
within the lists was pseudo-randomised with no more than four words or nonwords
in a row. In order to provide an exclusively L3 context (cf. Grosjean’s language
mode theory; Grosjean 1989), before the test all the participants were asked to read
a short German text. Additionally, the task instruction explaining that in this task the
testees were to decide as quickly and as accurately as possible whether a letter
string appearing on the screen was an existing German word or not was also written
in German. A 5-min-long training session preceded the experimental part.
5.4.4 Materials
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.4 Study 1: Exploring the Cognate Effect in Trilingual … 145
and (L2) English (e.g., PLAN) and the remaining words were noncognates—
German control words that were different from both their Polish and English trans-
lations (e.g., KOPF, meaning Eng.: head, Pl.: głowa). In addition to the test stimuli,
2 German words as well as 2 PGE cognates, 2 PG cognates and 2 nonwords, all
different from any of the test stimuli, were selected as practice items. In the word
materials only nouns were used because they are the only content words that pos-
sess the same lemma form in all three languages. Verbs and many adjectives are
morphologically marked by suffixes in both German and Polish (e.g., sing-en or
śpiew-ać, meaning to sing), while they are not marked in English, which usually
results in different lemma forms of these words in the three languages. All stimu-
lus items are listed in Appendix C and The word characteristics are presented in
Table 5.6. The two groups of stimuli: cognates and control words will be described
separately.
5.4.4.1 Cognates
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
146 5 Exploring the Cognate Facilitation Effect …
5.4.4.2 Nonwords
A number of nonwords that was equal to the number of words (48) was generated
by changing one or more letters in an existing German noun of 4–6 letters. All non-
words were orthographically legal in German and they did not exist as words in any
of the three languages. They were all created with the help of WordGen (cf. Duyck
et al. 2004), an online program that uses the CELEX and Lexique lexical data-
bases for word selection and nonword generation in Dutch, English, German, and
French. 14 items were adopted from the studies by Lemhöfer and Dijkstra (2004)
and Lemhöfer et al. (2004). The nonwords were matched to the word items in terms
of their mean length and number of syllables—the mean number of letters 4.68 and
the mean number of syllables 1.54.
For the RT analysis, only correct responses were considered. The overall error rate
amounted to 9.4 %, including 8.1 % on test words. Furthermore, reaction times
exceeding two times the standard deviation from the item mean counted as outli-
ers and were excluded from the set of valid responses. Outliers accounted for 1.7 %
of all the responses and they were approximately equally distributed across condi-
tions. Erroneous responses on test words (8.1 %) were excluded from the analysis,
so that 9.8 % of the data were discarded in total. The data on the erroneous response
matched item partners were not excluded since such a procedure could have led
to too high a percentage of the excluded data points. Moreover, in many studies
employing similar methodology (e.g., Lemhöfer et al. 2004; Van Hell and De Groot
2008) results obtained with and without the exclusion of the data on the outliers and
erroneous response matched item partners yielded comparable results.
Similarly to the Lemhöfer et al. (2008) study, error rates and RTs were analyzed
over participants only, because the selected cognates and controls were matched item-
by-item and can be seen as an almost exhaustive set of items with the given restric-
tions. The data gathered in the experiment were entered into the Statistica 9 program
for statistical analysis. The relevant data are summarised in Table 5.7.
Analysis of the obtained RTs indicates that all three languages became activated
and influenced the subject’s responses to the following targets. As can be seen from
the data illustrated in Figure 5.1, the participants responded much faster (743.17 ms)
Table 5.7 The mean RT SD ER SD
RTs and error rates for
PGE cognates 743.17 25.74 5.2 3.7
participants in Experiment 1
PG cognates 765.92 32.74 7.8 4.1
German controls 819.58 66.48 11.4 6.2
German fillers 823.17 63.15 15.0 8.1
nonwords 897.73 91.16 7.7 4.9
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.4 Study 1: Exploring the Cognate Effect in Trilingual … 147
780
765 3*(FRJQDWHV
ms
760 3*FRJQDWHV
743 FRQWUROV
740
720
700
Table 5.8 Paired samples t-test for the three types of stimuli words (PGE cognates, PG cognates
and German controls) in a German lexical decision task (Experiment 1)
Paired samples Mean SD SME t df Sig.
(2-tailed)
PGE versus control −76.42 67.26 19.42 −3.94 11 0.002
PG versus control −53.67 63.67 18.38 −2.92 11 0.014
PGE versus PG −22.75 33.45 9.65 −2.36 11 0.038
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
148 5 Exploring the Cognate Facilitation Effect …
Statistically significant cognate effects obtained for PGE and PG cognates con-
firmed by RTs and ERs analyzes conducted both with paired-samples t-test as well
as one-way ANOVAs point to language nonspecific selection which extends to three
languages. Thus, both hypotheses have been confirmed. First, the “standard” cog-
nate effect in lexical decision for a new language combination (Polish and German)
was replicated. Polish-German cognates were responded to faster than exclusively
German control words. Second, an additional cognate effect on top of the standard
cognate effect could be demonstrated for the trilingual population: words that had
the same form and meaning in all three languages (Polish, German, and English)
were recognized even faster and more accurately than the matched Polish-German
cognates with a dissimilar English translation. This indicates that during the recog-
nition of words in a given foreign language, not only the mother tongue, but also
another non-native language (English) exerts influence on recognition performance.
Admittedly, the obtained effects cannot be explained without the involvement of
all three languages: If the participants had selectively activated their German lex-
icon, there should have been no cognate effect whatsoever; if they had only acti-
vated the relevant lexicon (German) and their native language (Polish), there would
not have been any RT difference between PG and PGE cognates. Not only do the
present data reveal that the cognate effect can accumulate over languages but they
also show that co-activation of three lexicons occurs even within the same words.
It seems that even with a constraining experimental setting (German task, German
instruction) the trilingual participants did not manage to completely deactivate their
L1 and/or L2. Both of these languages became activated when the participants saw
the triple cognate word and both of these languages significantly affected lexical
decision times in reaction to the presented cognates.
Clearly, the presented interpretation of the obtained results could be challenged.
Firstly, it could be claimed that the difference between PGE and PG group might have
been caused by the fact that not all words in the PG group were identical cognates. It
needs to be noted here that non-identical cognates are reported to be recognized more
slowly (cf. Lemhöfer and Dijkstra 2004; Dijkstra 2007). However, an additional anal-
ysis conducted exclusively on the pairs of identical cognates repeated the results as
confirmed by one-way ANOVA F(1, 12) = 4.986, p < 0.05. Secondly, it could be
argued that the triple cognate effect might have been obtained due to the higher level
of proficiency in L2—English. Similarly, the lower level of L3 might have led to
stronger L1–L3 connections triggering faster RTs for PG cognates. Doubtless, a
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.4 Study 1: Exploring the Cognate Effect in Trilingual … 149
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
150 5 Exploring the Cognate Facilitation Effect …
for decisions on words that were cognates with respect to a third language. On the
whole, however, the typically stronger cognate effects from L1 on L2 than from L2
on L1 prove that L2 representations are generally activated less strongly or less rap-
idly than L1 representations, implying that they have less chance to affect the response
when L1 is the target. In other words, when cognates are processed in a second or
third language context, the first-language reading not only becomes active but it facil-
itates recognition as well (cf. Dijkstra 2007; Lemhöfer et al 2008; Study 1 above).
Consequently, it seems reasonable to claim that, in the previous experiment, before
the L3 (German) target reading of a cognate became active, the L1 (Polish) reading
had already affected target processing. In contrast, in the Polish language-specific lex-
ical decision task, multilinguals can be assumed to respond to the first reading of the
cognate they identify, which will often be their L1 reading. Thus, the cross-linguistic
effect measured relative to Polish controls is expected to be considerably smaller. In
fact, the question arises whether cognate effects are still present in the recognition of
L3 cognates in a first-language context and if that be the case, will the cumulative
influence of the stronger and the weaker foreign language (L2 and L3) lead to even
faster L1 recognition, thus adding evidence to the nonselective access extending to
three languages. Taken together, the issue under investigation is whether the cognate
effects found for PGE and PG cognates in the German task relative to German (L3)
control words (cf. Study 1) can also be demonstrated in comparison with Polish (L1)
control words in a purely L1 (Polish) task. If so, semantic and orthographic overlap of
cognates should cause facilitatory effects on word recognition. Alternatively, effects
caused by form similarity may be task dependent and nonsignificant relative to the
Polish control words. As Lemhöfer et al. (2008) rightly observe, this option would
indicate that whereas the simultaneous activation of an L1 code affects the recognition
of words in the second and/or third language, the reverse is not the case.
5.5.1 Research Hypotheses
The present study set out to investigate the generality of the language-nonselective
access hypothesis. More precisely, by changing the language of the task from L3 to
L1 the present study, in contrast to the experiment reported in the previous section,
tested the dependence of lexical retrieval on task demands, as well as the influence
of the foreign languages (L2 and L3) on the native language visual word recogni-
tion. Specifically, the following hypotheses were addressed:
H1. If the nonselective access hypothesis holds true in the exclusively native lan-
guage context and the weaker foreign language is potent enough to influence
the dominant language processing, performance will lead to different result pat-
terns for cognates than for noncognates. Faster RTs will be found for Polish-
German cognates than for Polish control words.
H2. If language nonspecific selection in lexical access in the exclusively
native language context extends to three languages and foreign languages
are potent enough to influence the dominant language processing, triple
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.5 Study 2: The Role of Task Demands in the Trilingual Processing … 151
5.5.2 The Participants
Nineteen participants (14 women and 5 men), drawn from the group involved in
Experiment 1, took part in the present study. Five students from the previous experi-
ment did not participate in the present task. The two sessions were conducted within
the interval of 6 weeks.
The apparatus and procedure of the lexical decision task in the present experi-
ment were similar to those used in Experiment 1. The stimulus materials consisted
of triple and double cognates from the previous study. However, since the lan-
guage of the task was L1, German control words were replaced with Polish con-
trols. Additionally, orthographically non-identical PG cognates were transformed
into their Polish equivalents. As in Experiment 1, the three groups of words (i.e.,
cognates with English and German, cognates with German, and control words)
were matched item-by-item for Polish frequency, length and number of syllables.
A group of Polish fillers (12) and a group of nonwords (48) were also included.
Again, in the word materials only nouns were used. All stimulus items are listed in
Appendix D and word characteristics are presented in Table 5.9.
For the present Polish language specific lexical decision experiment, only Polish
control words that were purely Polish nouns were included. They were noncog-
nates, and resembled neither their English, nor their German translation in either
orthography or phonology; examples are Polish NOGA (Ger.: Bein, Eng.: leg) or
PIES (Ger.: Hund; Eng.: dog). In order to keep the proportion of cognates in the
experiment at no more than 50 % of the words 12 additional pure Polish fillers were
included with similar characteristics as the Polish control words.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
152 5 Exploring the Cognate Facilitation Effect …
5.5.3.2 Nonwords
A number of nonwords that was equal to the number of words (48) was constructed
by changing one or more letters in an existing Polish noun of 4–6 letters. All non-
words were orthographically legal in Polish. They did not exist as words in any of
the three languages. The nonwords were matched to the word items in terms of their
mean length and number of syllables—the mean number of letters 4.53 and the
mean number of syllables 1.47.
For the analysis of RTs, only correct reactions were considered. The overall error
rate was 7.4 %. Furthermore, RTs that lay more than two standard deviations away
from the item mean were considered outliers. The percentage of outliers among the
correct trials was 2.3 %. Errors on test words accounted for 5.9 %. In total 8.2 %
of the data were excluded. The mean RTs, standard deviations, and error rates are
listed in Table 5.10.
As can be inferred from the data illustrated in Fig. 5.3, the participants
responded faster (728 ms) to PGE cognates than to PG cognates (746 ms) and con-
trol words (757 ms).
To investigate the influence of item type on RT, paired samples t-test was con-
ducted whose results are presented in Table 5.11. Planned comparison showed that
Table 5.10 The mean RT SD ER SD
RTs and error rates for the
PGE cognates 728.58 21.53 4.2 3.1
participants in Experiment 2
PG cognates 746.33 43.22 5.3 5.3
Polish controls 757.17 36.24 8.4 6.2
Polish fillers 760.03 39.33 9.1 7.7
Nonwords 792.15 47.15 10.2 11.1
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.5 Study 2: The Role of Task Demands in the Trilingual Processing … 153
Fig. 5.3 Mean RTs obtained for PGE cognates, PG cognates and control words
PGE cognates were recognized 29 ms faster than monolingual Polish control words.
This difference turned out to be statistically significant (t(11) = 4.11; p < 0.005;
F(1, 36) = 8.740, p < 0.005). PG cognates were recognized only 11 ms faster than
control words. This difference failed to reach statistical significance (t(11) = 0.79;
p = 0.44). Similarly, the difference between PGE cognates and PG cognates, turned
out to be not statistically significant (t(11) = 1.77; p = 0.10). Similarly, the analysis
of error rates revealed the same pattern. The fewest errors were made on PGE cog-
nates, more errors were made on PG cognates, and the most errors were made on
Polish control words. Nevertheless, only pairwise comparisons for PGE cognates
and Polish control words gave statistically significant results (F(1, 22) = 4.405;
p < 0.05).
Interestingly enough, only the group of PGE cognates compared with their
matched controls yielded statistically significant results both in RTs and ERs analy-
sis. Since the difference in processing PG cognates and controls did not reach the
significance level, nor did the difference between PGE and PG cognates, it may be
argued that the weaker foreign language does not affect visual word recognition in
the exclusively native language context. Consequently, Hypotheses 1 and 2 could
not be corroborated. No significant influence of the weaker foreign language on the
dominant language processing has been reported. Nor was any data found that could
Table 5.11 Paired samples t-test for the three types of stimuli words (PGE cognates, PG cog-
nates and Polish controls) in a Polish lexical decision task (Experiment 2)
Paired samples Mean SD SME t df Sig.
(2-tailed)
PGE versus control 28.58 24.03 6.94 4.11 11 0.001
PG versus control 10.83 47.36 13.67 0.79 11 0.44
PGE versus PG 17.75 34.62 9.99 1.77 11 0.10
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
154 5 Exploring the Cognate Facilitation Effect …
Table 5.12 Paired differences in RTs achieved for PGE and PG cognates in comparison to their
matched controls in German and Polish lexical decision tasks (Experiments 1 and 2)
German controls Polish controls
Mean RT SD Mean RT SD
PGE cognates versus controls 76.42 67.26 28.58 24.03
PG cognates versus controls 53.67 63.67 10.83 47.36
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.5 Study 2: The Role of Task Demands in the Trilingual Processing … 155
in the subjects’ L2 and L3 proficiency. In view of the above, some further research
using participants declaring comparable levels of L2 and L3 proficiency seemed
advisable.
5.6.1 Research Hypothesis
If, following the previous research in this area, the manipulation of the trilinguals’
proficiency in their third language may qualify the emergence of effects of a weaker
foreign language knowledge on native language processing, performance should
lead to result patterns different for cognates than for noncognates. The hypothesis
for the present experiment is as follows:
H: If the nonselective access hypothesis in the exclusively native language con-
text depends on foreign language proficiency, the participants with higher L3 level
will recognize PG cognates faster than those whose L3 proficiency is lower.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
156 5 Exploring the Cognate Facilitation Effect …
5.6.2 The Participants
The materials were identical to those used in Experiment 2. The procedure fol-
lowed in Experiment 2 was adopted in Experiment 3 to calculate mean RTs for
the cognates with English and German, for the cognates with German, and for the
noncognates.
The data of 2 participants with exceptionally high error rates (above 15 %) were
excluded from the analysis. For the remaining 11 participants, the overall error
rate amounted to 7.6 % (5.8 % on test words). In addition, RTs that lay more than
two standard deviations away from the mean (2.8 % of all data) were classified as
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.6 Study 3: The Role of Relative Proficiency in the Third Language … 157
Table 5.14 The mean RT SD ER SD
RTs and error rates for the
PGE cognates 730.92 23.58 4.6 5.6
participants in Experiment 3
PG cognates 745.00 43.91 5.1 6.2
Polish controls 758.75 53.58 7.9 7.9
Polish fillers 762.12 41.27 8.7 9.1
Nonwords 803.21 46.15 11.5 10.3
outliers and omitted from the analysis. In total 8.6 % of the data were excluded.
Mean lexical decision times and errors are presented in Table 5.14.
In Experiment 3, the same stimulus materials as in Experiment 2 were tested.
Now, a cognate advantage arose in words that were cognates with their L3 trans-
lations but still, by conventional criteria, this difference is considered to be not
quite statistically significant (t(11) = 2.03, p = 0.06). Again, the results for PGE
cognates reached significance (t(11) = 2.34, p < 0.5), probably due to the influ-
ence from L2 English. This assumed role of English appears to be interesting
especially if considered in relation to the subjects’ linguistic profile. Table 5.15
shows the results of the paired samples t-test for the three types of stimuli words.
With respect to the hypothesis set in the present experiment, to test the gener-
ality of the language-nonselective access and the dependence of lexical retrieval
on language proficiency, cross-task comparisons were conducted. The calculated
RT difference for PG cognates as compared with their matched controls from
Experiment 2 was juxtaposed with the parallel data obtained in Experiment 3 (cf.
Table 5.16). The level of L3 proficiency appeared to exert no effect on the visual
recognition of cognates. PG cognates were not faster to recognize when presented
to more fluent L3 users. Clearly, the fact that the comparison did not yield statisti-
cally significant results (F(1, 30) = 0.037, p = 0.84.) falsifies the hypothesis. On
the other hand, it can still be argued that the participants’ L3 level was still too
low to influence native language performance in an exclusively native language
context.
To sum up, the findings resulting from the nontarget, and weaker, language pro-
ficiency manipulation support the theoretical position that multilinguals’ process-
ing system is profoundly nonselective with respect to language only as far as the
stronger foreign language is concerned. Hence it seems legitimate to say that a cer-
tain level of weaker foreign language proficiency is required before any L3 effects
Table 5.15 Paired samples t-test for the three types of stimuli words (PGE cognates, PG cog-
nates and Polish controls) in a Polish lexical decision task in Experiment 3
Paired samples Mean SD SME t df Sig.
(2-tailed)
PGE versus control 27.83 23.17 11.85 2.34 11 0.038
PG versus control 13.75 29.45 6.75 2.03 11 0.066
PGE versus PG 14.08 31.25 8.49 1.65 11 0.125
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
158 5 Exploring the Cognate Facilitation Effect …
5.7.1 Research Hypothesis
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.7 Study 4: The Role of Psychotypology in Trilingual Processing … 159
reaction times to cognate words presented in the lexical decision task. The hypoth-
esis for the present experiment is as follows:
H: If stronger interaction between typologically related foreign languages (L2
and L3) known to the speaker than between his/her L1 and L3 is the manifestation
of strong connections between L2 and L3 representations and not between L1 and
L3 ones in the multilingual mental lexicon, the cumulative effect of psychotypology
and the L2 status will lead to faster RTs found for English-German cognates than
for Polish-German cognates.
5.7.2 The Participants
A group of 17 trilinguals with Polish as their mother tongue (L1), and English and
German as their second (L2) and third language (L3), respectively, participated in
the experiment. All of the participants were students (III year) of English Philology
at the Academy of International Relations in Łodź. All of them had learned English
and German as a foreign language at school. English from about the age of 13,
German from about the age of 16. The participants’ L2 competence was not tested
before the study. It was assumed that their status of students of English Philology
ensured advanced L2 proficiency. Additionally, only those students were selected
for whom the pass mark at the English practical exam was at least 4 (good). Before
the experiment a language background questionnaire was administered, in which 19
participants were asked to report on their competencies in all of their foreign lan-
guages and the frequency of use of these languages. A few respondents declared
some competence in foreign languages other than English and German, but those
selected for the experiment (17) described their competence as very low. The
reported languages included Spanish and Russian. Again, all the participants were
“unbalanced” trilinguals, i.e., they were not as proficient in their second and third
languages—English and German—as in their mother tongue Polish (cf. Table 5.17).
The data from 3 participants had to be excluded from analysis which sets the
mortality rate at the level of 17.6 %. The remaining 14 participants were between
21 and 36 years old with the mean of 28.5. 10 were female, 4 were male. The
participants were not paid for their participation, nor given any course credits.
To ensure that the testees constituted a homogenous sample as far as their
L3 German competence is concerned only those students were selected who
described their L3 level as pre-intermediate or intermediate (A2 and B1, respec-
tively). A summary of the participants’ learning history is given in Table 5.17.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
160 5 Exploring the Cognate Facilitation Effect …
The small scale of the study in terms of the number of the participants was
determined by low availability of subjects with German as their L3 declaring at
least A2 level proficiency.
Three critical groups of words were compared with respect to the latencies and
accuracy of their recognition: German control words that were different from both
their English and Polish translations (e.g., GELD; Eng.: money, Pl.: pienia˛dze),
Polish-German cognates that were not cognates to English (e.g., DACH, meaning
roof in both Polish and German), and English-German cognates that overlapped in
orthography and meaning in L2 and L3 but were different in L1 (e.g., FINGER;
Pl.: palec). The group of Polish-German cognates from Experiment 1 was used.
Additionally 12 English-German (EG) cognates were selected from the CELEX
database. They had identical or similar spelling and meaning in both languages,
like the word FINGER. They matched the Polish-German cognates with respect to
the length, number of syllables and frequency. All of them were singular forms of
nouns with a length of between 3 and 6 letters and no more than two syllables. Both
German fillers as well as nonwords from Experiment 1 were used in the present
experiment. All stimulus items are listed in Appendix E and word characteristics are
presented in Table 5.18.
As for the procedure, the language-specific lexical decision task for the weak-
est language was used. The procedure employed in the previous experiments was
adopted to calculate mean RTs for the cognates with English, for the cognates with
Polish, and for the noncognates.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.7 Study 4: The Role of Psychotypology in Trilingual Processing … 161
For the RT analysis, only correct responses were considered. Reaction times
exceeding two times the standard deviation from the subject means counted as outli-
ers and were excluded from the set of valid responses. Outliers accounted for 1.7 %
of all the responses and they were approximately equally distributed across con-
ditions. Erroneous responses (7.8 % for test words) were also excluded from the
analysis. The remaining data were entered into Statistica 9 program for statistical
analysis. Mean RTs were calculated for the cognates with English, for the cognates
with Polish, and for the noncognates; all of them are presented in Table 5.19.
In order to analyse the recorded RTs, a t-test for paired-samples whose results are
listed in Table 5.20 was conducted. The effect of word type turned out to be not sig-
nificant (F(2, 33) = 1.23; p = 0.34). Mean lexical decision times to the L3 words that
were cognates with English were only slightly shorter than those to the noncognates
and failed to reach significance. Interestingly enough, however, the difference between
the L3 words that were cognates with Polish and the noncognates turned out to be
significant: t(11) = 3.06; p < 0.05. So did the difference between L1–L3 cognates and
their L2–L3 matched counterparts: t(11) = 2.75; p < 0.05.
Despite the fact that cross-linguistic influence between typologically related for-
eign languages has been observed in numerous off-line TLA studies (cf. Hall and
Ecke 2003; Ringbom 2007), the present study, which investigated on-line trilingual
processing, did not prove stronger connections between L2 and L3 lexical items
than between L1 and L3 ones. The data obtained in the present experiment provide
no statistically significant evidence for the effect of psychotypology which would be
powerful enough to foster strong connections between typologically related foreign
languages. Hence the hypothesis set in the present experiment has to be rejected.
Contrary to the initial assumptions, faster RTs documented for PG cognates than for
Table 5.20 Paired samples t-test for the three types of stimuli words (EG cognates, PG cognates
and German controls) in a German lexical decision task in Experiment 4
Paired samples Mean SD SME t df Sig.
(2-tailed)
EG versus control 3.75 21.04 2.66 1.40 11 0.186
PG versus control 18.83 17.03 6.13 3.06 11 0.010
EG versus PG 15.08 21.89 5.46 2.75 11 0.018
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
162 5 Exploring the Cognate Facilitation Effect …
835
830
830
826
825
820 PG cognates
ms
EG cognates
811
810
805
800
Fig. 5.4 Mean RTs obtained for PG cognates, EG cognates and German control words
their EG matched counterparts (cf. Fig. 5.4) prove that there are strong connections
between L3 and L1, even though L1 (Polish) is typologically more distant from L3
(German) than L2 (English), and it lacks the special status of the second language.
To be sure, lexical links between L1 and L3 might have developed as a result
of learning habits. As appears from the post hoc questionnaire, the participants of
Experiment 4 learnt their L3 in an L1-based environment. Their teacher of German
was a Polish native speaker, who did not resort to English during their German
instruction. They also declared wide access to Polish-German and German-Polish
dictionaries in the course of their L3 learning process. Taking all these data into
consideration, it seems reasonable to claim that the structure of their mental lexi-
cons may have been affected by the nature of their learning experience. It seems
that this factor may be powerful enough to foster strong connections between lan-
guages extensively and concurrently used in education.
5.8 General Discussion
Recapitulating, the four experiments reported in the present chapter were performed
with two main goals in mind. Firstly, the aim was to take a deeper look at the nature
of the representation of cognates in the trilingual mental lexicon by including them in
a language specific lexical decision task in which all three languages were involved.
Secondly, the subsequent experiments were carried out in order to investigate any pos-
sible influence of a set of factors (such as task dependency, language proficiency and
psychotypology), identified in the course of research within TLA, on the representa-
tion of cognates in the multilingual mental lexicon. More precisely, in Experiment 2,
the BIA+ model’s predictions about task demands and multilingual lexicon were veri-
fied by changing the task (the weakest vs. the strongest language context) but keeping
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.8 General Discussion 163
the stimulus materials constant. In Experiment 3, relative language proficiency and its
impact on lexical recognition was tested. The final experiment aimed at checking the
cumulative effect of the L2 status factor and psychotypology on the multilingual stor-
age and retrieval of cognates. The experiments have led to several conclusions about
cognate processing and representation in trilinguals, to be discussed in the paragraphs
to come.
Consistent with previous findings in the literature, and as expected, the reported
experiments proved that trilinguals processed cognates more quickly and accurately
than they processed noncognate control words. Obviously, the overlap of form and
meaning across languages facilitated lexical access. Before some more detailed
analysis of the results of the experiments is offered, it seems justifiable to reiterate
the most plausible accounts of the cognate facilitation effect on the basis of the con-
ducted experiments. In the literature on the topic, mainly concentrated on bilinguals,
different accounts of the observed bilingual cognate effects are proposed, depend-
ing on the way cognates are thought to be represented within the bilingual language
system. First, orthographically identical cognates may share a single orthographic
representation across the two languages, as some researchers suggest (cf. De Groot
and Nas 1991; Sánchez Casas et al. 1992). In this view, the cognate effect is sim-
ply a consequence of the cumulative frequency of a cognate across all relevant lan-
guages. Because a multilingual encounters a cognate more frequently than a word
that exists in only one language, it is the standard word frequency effect that causes
the RT advantage of cognates. The results obtained for the group of PGE cognates
in Experiment 1 appear to be in line with this explanation. Interestingly enough, the
single orthographic representation account is related to a learning-based explanation
of the cognate effect, according to which language learners can make use of pre-
existing L1 memory representations during the acquisition of cognates (cf. De Groot
and Keijzer 2000; De Groot and Van Hell 2005). However, there is a growing body
of evidence that the obtained result patterns for cognates are not just a consequence
of cumulative frequency effects (cf. De Groot et al. 2000; Dijkstra et al. 1999;
Dijkstra 2007). In fact, many linguists point out that this account is not easily rec-
oncilable with the finding of facilitatory cognate effects for cognates that are ortho-
graphically similar, but not identical, such as the Polish-German pair cel versus Ziel,
which was used in the described experiments. Given that many cognates in the pre-
sented experiments had a different orthographic representation in the two languages
(almost 46 % in PG group), it seems unlikely that the facilitatory cognate effects can
be explained by assuming that a cognate’s effective frequency is simply that of the
sum of the frequencies of two words across languages. Indeed, the participants in all
the tasks were faster at recognizing identical cognates to similar ones. Nevertheless,
the latter group still triggered the facilitatory cognate effect as confirmed by planned
comparisons conducted on the group of non-identical cognates and their matched
controls. Therefore, a second possibility that may account for both orthographically
identical and nonidentical cognate effects involves semantic feedback. Even if cog-
nates do not share the same orthographic representations for all languages to which
they belong, it can be assumed that they activate the same conceptual representa-
tion, because the overlap of their meanings in the different languages is usually large
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
164 5 Exploring the Cognate Facilitation Effect …
(cf. De Groot and Nas 1991; Van Hell and De Groot 1998a; De Groot and Van Hell
2005). Cognates would then be recognized faster because their semantic representa-
tions (receiving input from two rather than one orthographic representation) are acti-
vated to a larger degree (cf. BIA+ model). What is significant is that this account of
the cognate effect is in line with recent studies on monolingual word recognition that
also argue in favor of semantic-to-orthographic feedback (cf. Pecher 2001; Reimer
et al. 2001). Also the data gathered in the described experiments can be interpreted
within this localist connectionist framework, in the light of which cognate translation
equivalents are “only ‘special’ relative to noncognate translations in that they share
more orthographic, semantic, and/or phonological features across languages” (Voga
and Grainger 2007, p. 943). According to this account, the cognate facilitation effect
in reading might in fact be an orthographic-semantic priming effect: Overlapping
orthographic and semantic representations of both languages become active upon
the presentation of one of the readings of the cognate, leading to a facilitated rec-
ognition of cognates relative to noncognates. In this model, the cognate facilitation
effect depends on both cross-linguistic similarity (the factor that in all the experi-
ments could additionally trigger faster recognition of PGE words, all of which were
identical cognates) and word frequency. It has been argued that both a larger similar-
ity of the two readings of a cognate and a higher frequency result in a more strongly
activated shared semantic representation. Since the factor of cross-linguistic similar-
ity might have influenced the reported results, its role will be further analyzed in the
section devoted to final conclusions where some future experiments aimed at verify-
ing the role of this very factor are to be discussed.
More importantly, it appears that all of the enumerated approaches are extend-
able to three languages and reconcilable with the current findings. Just as the bilin-
gual cognate effect has been interpreted in terms of coactivation of two languages
(cf. De Groot et al. 2002; Dijkstra et al. 1999; Dijkstra 2007), it seems reasonable
to believe that data gathered in Experiment 1 provide evidence for the simultane-
ous involvement of all three languages during the word recognition process in L3.
As evidenced in Experiment 1, the obtained results reveal that the cognate effect
can accumulate over languages: while cognate status in one language caused shorter
word recognition latencies, the additional cognate status in one additional language
speeded up responses even more. Thus, it can be claimed that the notion of nonse-
lective lexical access that has recently received growing support within the bilingual
domain, seems to generalise to trilinguals and three languages. Within experiments
2–4 each experiment focused on a different set of factors which were meant to influ-
ence lexical connections between the languages known to the participants. It was
predicted prior to experiments 2–4 that, if the influence of factors in question is
potent enough to foster stronger connections with only one of the languages, the
facilitation effect will be observed only for this pair of languages. The absence of
the facilitation effect is not to mean that there are no links with the other language
(EG relation in Experiment 4), but that the connections are too weak to induce this
effect.
As mentioned before, Experiment 1 proved that processing the weakest language
(L3) word entails automatic, parallel activation of candidate words in the dominant,
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.8 General Discussion 165
stronger languages; not only L1 but also L2. It was concluded that in accordance
with the nonselective access hypothesis the presentation of a word in one language
automatically activates words from both the target and the nontarget languages in
parallel. From this it logically follows that cross-linguistic effects may arise in both
directions, manifesting themselves not only in a nondominant but also in a domi-
nant target language. This assumption gave rise to the hypotheses set in the sec-
ond experiment, whose aim was twofold. First, the influence of foreign language
knowledge on native language performance in an exclusively native language con-
text was studied. The objective was to verify the assumption in the light of which
weaker language knowledge may influence performance in the dominant language.
Needless to say, such a finding would additionally support nonselective access in
multilingual lexicon. Second, using the same group of PG cognates in Experiments
1 and 2 allowed the researcher to verify the influence of task demands on the rec-
ognition of words and the hypothesis that cognate effects might be task dependent.
As regards the nonselective position, it has been both strengthened and extended.
Experiment 2 showed that processing words in the strongest language can be influ-
enced by weaker language knowledge. In other words, the finding that foreign
language knowledge (in this case L2) affects L1 target word processing in an exclu-
sively native language context provided additional support for the theoretical posi-
tion that the language processing system of multilinguals is profoundly nonselective
with respect to language. On the other hand, the nonnative language influence on
the mother tongue was clear only for L2 since only PGE cognates were recognized
faster than their matched controls; L1–L3 cognates did not generate statistically sig-
nificant results. Admittedly, it was L2 that supported the recognition. The compari-
son of the results for PG cognates from Experiment 2 with those from Experiment 1
showed that the same list of stimulus words, recognized by the same group of sub-
jects, generated different RTs. Clearly, language nonselective access in the case of
three languages seems to be task dependent. And although more research is needed
to find out the exact nature of the demands posed by various tasks, Experiment 2
showed that the different results they produce may be informative with respect to
the underlying representation of cognates. Summing up, Experiment 2 showed that
processing words in the strongest language can be influenced by weaker language
knowledge—in this case, L2 knowledge. However, since L3 did not affect L1, it
was hypothesized that a certain level of weaker language proficiency is required
before any weaker language effects become noticeable in L1 processing.
In order to strengthen the interpretation of the post hoc analysis in Experiment 2
and the conclusion that L1 performance can be influenced by weaker language
knowledge if the multilingual has reached a certain level of weaker language pro-
ficiency, another experiment was performed (cf. Experiment 3). This time, how-
ever, trilinguals with a higher level of proficiency in L3 were used. Now again, as in
Experiment 2, lexical decision times to the triple cognates were shorter than those
to the noncognates. Similarly, responses on the L1 words that were cognates with
German were faster, though not statistically significant, than on the noncognates.
Accordingly, it can be concluded that, the manipulation of the trilinguals’ profi-
ciency in L2 and L3 revealed that their proficiency in L3 was too weak to exert any
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
166 5 Exploring the Cognate Facilitation Effect …
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
5.8 General Discussion 167
Finally, it may also be the case that the languages involved in the experiment were
still too similar and thus the psychotypology effect, typical for languages coming
from different language families, could not have been activated.
All in all, the findings triggered many additional questions that could constitute a
starting point for another set of experiments and that will be partly addressed in the
final part of this volume, together with the implications of the outcomes discussed
in this section for the structure of the multilingual lexicon and the conceptualiza-
tion of a multilingual speaker. A few tentative recommendations stemming from the
present observations concerning the ways in which the conclusions based on the
outcomes of the described experiments could be adopted in the process of foreign
language learning and teaching will be offered in the subsequent section, which is
also a conclusion to the present volume.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
This final section provides a critical synthesis of the major findings presented
throughout this volume. It aims to show how research on multilingual processing
can add to the current body of knowledge on third or additional language acquisi-
tion. First, it once again outlines the goals and the structure of the work, offering a
brief summary of the general findings presented in the context of the dispute over
the conceptualization of a multilingual learner. Next, it refers to limitations of the
conducted experiments and enumerates some suggestions for future research. Last
but not least, the final paragraphs address the role of the obtained results for multi-
lingual didactics.
The primary goal of this volume was to offer an insight into the structure of the
multilingual mental lexicon which is a product of the interplay of a whole array of
cross-linguistic factors in the minds of multilingual speakers. More specifically,
the book focused on the empirical investigations of the storage and retrieval of
double and triple cognates in a multilingual mind. All the aspects, analyzed in the
experiments reported in the practical chapter, are part of a broader question of how
multilinguals make their lexical decisions and how they recognize words from dif-
ferent languages. In order to place the investigations in a larger context, the text
was divided into five chapters. The preliminary theoretical considerations referring
to a plethora of multilingual issues included in Chap. 1 provided a relevant back-
ground against which most important problems concerning the mental lexicon of
mono- and multilingual learners were presented in Chaps. 2 and 3, respectively.
In Chap. 4 the focus of attention was shifted to the special position of cognates
in visual word recognition in the multilingual mind. Chap. 5 offered a detailed
account of a series of experiments conducted in order to verify the assumptions
concerning the storage and processing of cognates in the context of the Polish-
English-German lexicon.
The underlying aim of the present work was to explore whether multilinguals
should be considered as learners and speakers in their own right as opposed to L2
learners and if so, what consequences this would bring to third or additional lan-
guage teaching. As mentioned in Chap. 1, on the one hand, there are scholars (cf.
Baetens Beardsmore 1986; Li Wei 2007; Mitchell and Myles 2004) who assume
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
170 Conclusions and Implications
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Conclusions and Implications 171
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
172 Conclusions and Implications
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Conclusions and Implications 173
Over the decades, three major language acquisition research lines were
developed, each of which was expressed in various models and hypotheses (cf.
Hufeisen 2004; Riemer 2002): the contrastive hypothesis (which focused on com-
paring and contrasting languages, and to which we owe the delineation of interfer-
ence); the nativistic hypothesis (which postulated an inherent language acquisition
sequence within each individual and identified development-related errors); and
the interlanguage hypothesis (which described language acquisition as a dynamic
and systematic process focused on the new language to be learned, and in par-
ticular described transfer processes). It needs stressing, however, that all of these
hypotheses assumed as self-evident the involvement of only two languages and
“implied, at best in passing, that the acquisition or learning of other (foreign) lan-
guages did not differ from the learning of the first foreign language” (Hufeisen
2004, p. 7). Needless to say, the situation has changed radically within the last two
decades. The attention directed towards the investigation of multilingual learners
has resulted in the intense research activity in the field of multilingualism which
in turn has helped raise a general awareness both among scholars and practitio-
ners about the distinctiveness of multilinguals` language learning processes and
how these may differ from those of monolingual learners of a second language.
The integrating of the multilingual research findings into the classroom has led
to the development of tertiary language didactics which is now seeking to apply
this qualitative difference to advantage, including in L3 teaching learners’ previ-
ous cognitive and emotional experiences. What is also worth noting is the fact that
if the languages to be learned are very similar, this recourse can also be closely
related to linguistic aspects; if they are rather different, then it is more the case
that learning strategies and cognitive factors come into play. Interestingly enough,
it is admitted that particularly in the latter case, it is not absolutely necessary for
teachers to be perfect speakers of L2 (cf. Jessner 2006, 2008a, b; Aronin and
Hufeisen 2009) since it is not primarily the languages that they activate, but rather
the learning potential established during the learning of preceding languages. And
thus, what tertiary language teaching in school can achieve is, above all, the devel-
opment of language learning awareness that would help to learn subsequent lan-
guages efficiently.
As evidenced in the conducted experiments, in third or additional language
acquisition, an important aspect is the role of prior linguistic knowledge, since if
the learners perceive enough similarity between the two (or more) languages, they
rely strongly on their mother tongue and/or non-native languages. Therefore, the
facilitative role of cognates in TLA seems to be of no minor importance. By way
of illustration, it has been proved that fewer learning sessions are needed to help
trilinguals recall cognates than noncognates (cf. Carroll 1992). Similarly, response
latencies in translating cognates have been reported to be faster than for noncog-
nates (cf. Ellis and Beaton 1995, De Groot and Keijzer 2000). In fact, in SLA,
vocabulary acquisition based on detection and exploitation of similarities between
novel lexical input and prior lexical knowledge is a well-documented concept,
known as the parasitic strategy (cf. Hall 2002; Hall and Ecke 2003). Undeniably,
this approach can be equally applicable to the field of TLA. Hall et al. (2009) go
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
174 Conclusions and Implications
The same is essentially true of TLA. Thus, without further investigations, the
provision of recommendations and the development of materials for class-
room use, it is difficult to imagine that third or additional language teach-
ing could find its way into the language classroom. Thus, is hoped that this
volume can help teachers, testers as well as curriculum and material designers
to have a better understanding of multilingual processing. Only by being famil-
iar with the mental processes involved in multilingual acquisition can teach-
ers understand the problems their learners have to face when learning a third
or additional foreign language, while course book writers can produce more
efficient teaching materials. All in all, languages being taught in the class-
room need to be linked in order to profit from the synergies and to exploit the
resources that many of the pupils already have available through their prior lan-
guage knowledge (cf. Cenoz and Gorter 2013). Therefore, one of the most
difficult aims of future work on language teacher education will be to make
sure that all language teachers are experts on multilingualism, even if they
teach only one language. This issue is addressed by Krumm (2005, p. 35),
who emphasizes that the most difficult correction of the educational system will
be not to train teachers in English or French but experts in multilingualism, who
teach a certain language but also accept at the same time the multilingualism of
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Conclusions and Implications 175
their learners (Cenoz and Gorter 2013; Szubko-Sitarek 2014; Komorowska 2013).
In fact, an approach to softening the boundaries between languages has been put
forward by Cenoz and Gorter (2013). This approach promotes using the resources
multilingual learners have at their disposal in order to improve efficiency in lan-
guage teaching. It explores the possibility of establishing bridges between second
and foreign language teaching. As Hufeisen and Marx (2007) rightly claim, aware-
ness raising and interacting with samples of cross-linguistic similarities indeed
result in creating a mental model for intercomprehension.
The current multilingual projects (cf. EuroCom) focus on the question of how
the teaching and learning of tertiary languages can be structured in such a way
as to consciously incorporate the learner's existing language knowledge and lan-
guage learning experience (mother tongue, first foreign language). More precisely,
the purpose of many researchers and foreign language teachers should be to teach
subsequent foreign languages in a manner different from that used for the first for-
eign language in order to tap the potential already developed through utilizing the
knowledge of the mother tongue and the first foreign language.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Appendix A
Ratings were collected for English and German translation pairs that were test or
control items
Test Test O-rating P-rating Mean Con Con O-rating P-rating Mean
English German OP English German con con OP
rating rating
con
Bank Bank 5.00 5.00 5 Road Weg 1.00 1.00 1.00
Film Film 5.00 5.00 5 Tree Baum 1.00 1.00 1.00
Finger Finger 5.00 5.00 5 Proof Beweis 1.15 1.12 1.13
Gold Gold 5.00 5.00 5 Bell Glocke 1.32 1.25 1.28
Lift Lift 5.00 5.00 5 Knife Messer 1.00 1.00 1.00
Moment Moment 5.00 5.00 5 Woman Frau 1.15 1.05 1.1
Park Park 5.00 5.00 5 Sign Zeichen 1.45 1.15 1.3
Plan Plan 5.00 5.00 5 Chair Stühl 1.00 1.00 1.00
Radio Radio 5.00 5.00 5 River Fluss 1.00 1.00 1.00
Ring Ring 5.00 5.00 5 Duty Pflicht 1.00 1.00 1.00
Ski Ski 5.00 5.00 5 Fork Gabel 1.00 1.00 1.00
Sport Sport 5.00 5.00 5 Autumn Herbst 1.00 1.00 1.00
Talent Talent 5.00 5.00 5 Song Lied 1.00 1.00 1.00
Test Test 5.00 5.00 5 Aim Ziel 1.13 1.25 1.19
Trend Trend 5.00 5.00 5 Potato Kartoffel 2.30 2.50 2.4
Winter Winter 5.00 5.00 5 Horse Pferd 1.23 1.15 1.19
Wolf Wolf 5.00 5.00 5 Pillow Kissen 1.15 1.20 1.17
Echo Echo 5.00 4.87 4.93 Lemon Zitrone 1.21 1.15 1.18
Hotel Hotel 5.00 4.80 4.9 Leg Bein 1.25 1.00 1.13
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
178 Appendix A
Test Test O-rating P-rating Mean Con Con O-rating P-rating Mean
English German OP English German con con OP
rating rating
con
Grass Gras 4.50 5.00 4.75 Church Kirche 1.35 1.27 1.31
Garden Garten 4.60 4.87 4.73 Window Fenster 1.15 1.21 1.18
Hand Hand 5.00 4.30 4.65 Head Kopf 1.00 1.00 1.00
Fish Fisch 4.00 5.00 4.5 Mirror Spiegel 1.00 1.00 1.00
Doctor Doktor 4.50 4.28 4.39 Fear Angst 1.00 1.00 1.00
Lamp Lampe 4.32 3.98 4.15 Skirt Rock 1.00 1.00 1.00
Kiss Kuss 3.89 3.98 3.93 Cheese Käse 1.35 1.06 1.20
Hunger Hunger 4.20 3.57 3.88 Juice Saft 1.00 1.00 1.00
Summer Sommer 4.23 3.54 3.88 Animal Tier 1.00 1.00 1.00
Nose Nase 3.70 3.90 3.8 Advice Ratschlag 1.00 1.00 1.00
Milk Milch 4.28 3.30 3.79 Watch Uhr 1.25 1.13 1.19
Mouse Maus 2.50 5.00 3.75 Cherry Kirsche 1.65 1.35 1.5
Baker Bäcker 4.10 3.30 3.7 Soap Seife 1.25 1.15 1.2
Angel Engel 4.12 3.17 3.64 Spoon Löffel 1.00 1.25 1.12
Chance Chance 5.00 2.21 3.6 Size Grösse 1.15 1.13 1.14
Thing Ding 3.10 3.90 3.5 Room Zimmer 1.50 1.45 1.47
Tomato Tomate 3.92 3,03 3.47 Pear Birne 1.24 1.30 1.27
Book Buch 3.54 3.25 3.39 View Aussicht 1.13 1.15 1.14
School Schule 3.78 2.86 3.32 Power Macht 1.00 1.15 1.07
Boat Boot 3.45 3.12 3.28 Bag Tasche 1.00 1.00 1.00
Frog Frosch 2.30 1.23 3.18 Corner Ecke 1.00 1.00 1.00
Soup Suppe 3.12 2.85 2.98 Shirt Hemd 1.13 1.25 1.19
Carrot Karotte 3.15 2.75 2.95 Monkey Affe 1.13 1.00 1.06
Water Wasser 2.70 3.10 2.9 Fruit Obst 1.00 1.00 1.00
Sugar Zucker 2.98 2.76 2.87 Skin Haut 1.00 1.00 1.00
Coffee Kaffe 2.96 2.76 2.86 Bottle Flasche 1.00 1.00 1.00
Salt Salz 2.30 2.90 2.6 Bird Vogel 1.00 1.00 1.00
Stone Stein 2.50 1.90 2.2 Shop Geschäft 1.25 1.14 1.19
Lion Löwe 1.90 1.70 1.8 Tooth Zahn 1.63 1.50 1.56
Rain Regen 1.78 1.45 1.61 Train Zug 1.00 1.00 1.00
King König 1.56 2.35 1.50 Forest Wald 1.00 1.00 1.00
Abbreviations for test words: Test English English orthographic word form, Test German German
orthographic word form, O-rating mean orthographic rating, P-rating mean phonological rating,
Mean OP rating mean of O and P ratings per stimulus word. The same organization holds for the
matched control items, referred to as “con”
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Appendix B
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
180 Appendix B
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Appendix B 181
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
182 Appendix B
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Appendix C
German Fillers:
Nonwords:
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
184 Appendix C
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Appendix D
Polish Fillers:
Nonwords:
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
186 Appendix D
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
Appendix E
German Fillers:
Nonwords:
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
188 Appendix E
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
190 References
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 191
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
192 References
Butterworth, B. (ed.). 1983. Language Production Volume 2: Development, Writing and Other
Language Processes. London: Academic Press.
Butterworth, B. 1983. “Lexical representation”, in: B. Butterworth (ed.), 257–294.
Caramazza, A., A. Laudanna and C. Romani. 1988. “Lexical access and inflectional morphol-
ogy”, Cognition 28: 297–332.
Caramazza, A., and I. Brones. 1979. “Lexical access in bilinguals”, Bulletin of the Psychonomic
Society 13: 212–214.
Carr, T. H. and D. Dagenbach. 1990. Semantic priming and repetition priming from masked
words: Evidence for a center- surround attentional mechanism in perceptual recognition.
Journal of Experimental Psychology: Learning, Memory, and Cognition 16: 341–350.
Carreiras, M., M. Perea, C. Gil-López, R. Abu Mallouh and E. Salillas. 2013. “Neural correlates
of visual versus abstract letter processing in Roman and Arabic scripts“, Journal of Cognitive
Neuroscience 25/11: 1975–1985.
Carringer, D. 1974. “Creative thinking abilities of a Mexican youth. The relationship of bilin-
gualism”, Journal of Cross–cultural Psychology 5: 492–504.
Carroll, S. 1992. “On cognates”, Second Language Research 8: 93–119.
Cattell, J.M. 1887. “Experiments on the association of ideas”, Mind 12: 68–74.
Cenoz J. 2001. “The effect of linguistic distance, L2 status, and age on cross–linguistic influence
in third language acquisition”, in J. Cenoz, B. Hufeisen and U. Jessner (eds.) 2001a, 8–20.
Cenoz, J. and F. Genesee (eds.). 1998. Beyond Bilingualism. Multilingualism and Multilingual
Education. Clevedon: Multilingual Matters.
Cenoz, J. 2000. “Research on multilingual acquisition”, in: J. Cenoz and U. Jessner (eds.),
39–53.
Cenoz, J. 2003. “The Role of Typology in the Organization of the Multilingual Lexicon”, in: J.
Cenoz, B. Hufeisen and U. Jessner (eds.), 103–116.
Cenoz, J. 2008. “Teaching through Basque: achievements and challenges”, Language, Culture
and Curriculum 21: 1–4.
Cenoz, J. 2013. “Discussion: towards an educational perspective in CLIL language policy and
pedagogical practice”, International Journal of Bilingual Education and Bilingualism 16:
389–394.
Cenoz, J. and D. Gorter. 2011a. “Focus on Multilingualism: a Study of Trilingual Writing”,
TheModern Language Journal 95: 356–369.
Cenoz, J. and D. Gorter. 2011b. “A Holistic Approach to Multilingual Education: Introduction”,
The Modern Language Journal 95: 339–343.
Cenoz, J. and D. Gorter. 2013. “Towards a Plurilingual Approach in English Language Teaching:
Softening the Boundaries Between Languages”, TESOL Quarterly 47: 591–599.
Cenoz, J. and D. Lindsay. 1994. “Teaching English in a primary school: a project to introduce a
third language to eight year olds”, Language and Education 8: 201–210.
Cenoz, J. and F. Genesee. (eds.). 1998. Beyond Bilingualism. Multilingualism and Multilingual
Education. Clevedon: Multilingual Matters.
Cenoz, J. and F. Genesee. 1998. “Psycholinguistic perspectives on multilingualism and multilin-
gual education”, in: J. Cenoz and F. Genesee (eds.), 16–34.
Cenoz, J. and U. Jessner. (eds.). 2000. English in Europe. The Acquisition of a Third Language.
Clevedon: Multilingual Matters.
Cenoz, J. and U. Jessner. 2009. “The study of multilingualism in educational contexts”, in: L.
Aronin and B. Hufeisen. (eds.), 121–138.
Cenoz, J., B. Hufeisen and U. Jessner. (eds.). 2001a. Cross–linguistic Influence in Third
Language Acquisition. Clevedon: Multilingual Matters.
Cenoz, J., B. Hufeisen and U. Jessner. (eds.). 2001b. Looking Beyond Second Language
Acquisition. Tübingen: Stauffenburg Verlag.
Cenoz, J., B. Hufeisen and U. Jessner. (eds.). 2003. Multilingual Lexicon. Kluwer, Dordrecht.
Cenoz, J., B. Hufeisen and U. Jessner. 2003. “Why investigate the multilingual lexicon”, in: J.
Cenoz, B. Hufeisen and U. Jessner (eds.), 1–11.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 193
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
194 References
Cook V.J. 2007b. “The goals of ELT: reproducing native–speakers or promoting multi–compe-
tence among second language users?”, in: J. Cummins and C. Davison (eds.), 237–248.
Cook, V.J. (ed.). 2002. Portraits of the L2 User. Clevedon, Multilingual Matters.
Cook, V.J. (ed.). 2003. Effects of the Second Language on the First. Clevedon: Multilingual
Matters.
Cook, V.J. 1991. “The poverty–of–the–stimulus argument and multi–competence”, Second
Language Research 7/2: 103–117.
Cook, V.J. 1992. “Evidence for multi–competence”, Language Learning 42/4: 557–591.
Cook, V.J. 1993. “Holistic multi–competence: jeu d’esprit or paradigm shift?”, in: B. Kettemann
and W.W. Wieden (eds.), 4–8.
Cook, V.J. 1995. “Multi–competence and the learning of many languages”, Language, Culture
and Curriculum 8/2: 93–98.
Cook, V.J. 1996. “Competence and multi–competence”, in: G. Brown, K. Malmkjaer and J.
Williams. (eds.), 57–69.
Cook, V.J. 2000. “Is transfer the right word? Paper presented at the 7th International
Pragmatics Conference, July, 2000. Budapest.” Retrieved: 12th January 2010 from
http://homepage.ntlworld.com/vivian.c/Writings/Papers/Transfer2000.htm
Cook, V.J. 2002. “Background to the L2 user”, in: V.J. Cook (ed.), 1–28.
Cook, V.J. 2008. Second Language Learning and Language Teaching. 4th ed. London: Hodder
Education.
Cook, V.J. and M. Newson. 1996. Chomsky’s Universal Grammar: An Introduction. 2nd ed.
Oxford: Blackwell.
Copeland, J.E. and P.W. Davis. (eds.). 1981. The Seventh LACUS Forum. Columbia, SC:
Hornbeam Press.
Costa, A., A. Caramazza, A. And N. Sebastian–Galles. 2000. “The cognate facilitation effect:
Implications for models of lexical access”, Journal of Experimental Psychology: Learning,
Memory, and Cognition 26: 1283–1296.
Costa, A., and A. Caramazza. 1999. “Is lexical selection in bilingual speech production lan-
guage–specific? Further evidence from Spanish–English and English–Spanish bilinguals”,
Bilingualism 2: 231–244.
Costa, A., M. Santesteban and A. Caño. 2005. “On the facilitatory effects of cognates in bilingual
speech production”, Brain and Language 94: 94–103.
Craik F.I., E. Bialystok and M. Freedman. 2010. “Delaying the onset of Alzheimer disease: bilin-
gualism as a form of cognitive reserve”, Neurology, 75 (19): 1726–1729.
Cristoffanini, P., K. Kirsner and D. Milech. 1986. “Bilingual lexical representation: The status of
Spanish–English cognates”, Quarterly Journal of Experimental Psychology 38A: 367–393.
Crystal, D. 1997. English as a Global Language. Cambridge: Cambridge University Press.
Crystal, D. 2003. English as a Global Language. Second Edition. Cambridge: Cambridge
University Press.
Cummins, J. 1978. “Bilingualism and the development of metalinguistic awareness”, Journal of
Cross-Cultural Psychology 9: 131–149.
Cummins, J. 1979. “Linguistic interdependence and the educational development of bilingual
children”, Review of Educational Research 49: 222–251.
Cummins, J. and C. Davison (eds.). 2007. Handbook on English Language Teaching. London;
New York: Springer.
Cummins, J. 1976. “The influence of bilingualism on cognitive growth: A synthesis of research
findings and explanatory hypotheses”, Working Papers on Bilingualism 9: 1–43.
Cutler, A. (ed.). 2005. Twenty-first Century Psycholinguistics: Four Cornerstones. London:
Erlbaum.
Davis, C. J. 2010. “The spatial coding model of visual word identification”, Psychological
Review 117: 713–758.
Davis, Ch., R. Sánchez–Casas, J.E.García–Albea, M. Guasch, M. Molero and P. Ferré. 2010.
“Masked translation priming: Varying language experience and word type with Spanish–
English bilinguals”, Bilingualism: Language and Cognition 13/2: 137–155.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 195
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
196 References
De Houwer, A. 1999. “Two or more languages in early childhood: Some general points and prac-
tical recommendations”, Retrieved: 4th March 2005, from www.cal.org/resources/digest/earl
ychild.html
De Houwer, A. 2004. “Trilingual input and children’s language use in trilingual families in
Flanders”, in: C.Hoffmann and J. Ytsma. (eds.), 118–139.
de Lacy, P. (ed.). 2007. Handbook of phonological theory. Cambridge, MA: Cambridge
University Press.
De Saussure, F. 1959. Course in General Linguistics. Translated and annotated by Roy Harris.
London: Duckworth.
Deese, J. 1962. “The structure of associative meaning”, Psychological Review 69: 161–175.
Deese, J. 1965. The Structure of Association in Language and Thought. Baltimore: Johns
Hopkins University Press.
Dell, G.S. 1988. “The retrieval of phonological forms in production: Tests of predictions from a
connectionist model”, Journal of Memory and Language 27: 124–142.
Dell, G.S., F. Chang and Z.M. Griffin. 1999. “Connectionist Models of Language Production:
Lexical Access and Grammatical Encoding”, Cognitive Science 23: 517–542.
Desmet, T. and W. Duyck. 2007. “Bilingual Language Processing”, Language and Linguistics
Compass 1/3: 168–194.
Diaz, R.M. 1985b. “The intellectual power of bilingualism” Quarterly Newsletter of the
Laboratory of Cornporative Human Cognition 7:16–22.
Diebold, A.R. 1964. “Incipient bilingualism”, in: D. Hymes (ed.), 495–511.
Dijkstra, A. 2003a. “Lexical processing in bilinguals and multilinguals: The word selection prob-
lem”, in: J. Cenoz, B. Hufeisen and U. Jessner (eds.), 11–26.
Dijkstra, A. 2003b. Lexical storage and retrieval in bilinguals, in: R. van Hout, A. Hulk, F.
Kuiken and R. Towell (eds.), 129–150.
Dijkstra, A. 2005. Bilingual visual word recognition and lexical access, in: J.F. Kroll and A. De
Groot (eds.), 178–201.
Dijkstra, A. 2007. The multilingual lexicon, in: M.G. Gaskell. (ed.), 251–265.
Dijkstra, A. and W. van Heuven. 1998. “The BIA–model and bilingual word recognition”, in: J.
Grainger and A. Jacobs (eds.), 189–225.
Dijkstra, A. and W.J.B. van Heuven. 2002a. “The architecture of the bilingual word recognition
system: From identification to decision”, Bilingualism: Language and Cognition 5: 175–197.
Dijkstra, A. and W.J.B. van Heuven. 2002b. “Modeling bilingual word recognition: Past, present
and future. Authors’ response”, Bilingualism: Language and Cognition 5: 219–224.
Dijkstra, A., H. Van Jaarsveld and S. Ten Brinke. 1998. “Interlingual homograph recognition:
Effects of task demands and language intermixing”, Bilingualism: Language and Cognition
1: 51–66.
Dijkstra, A., J. Grainger and W.J.B. van Heuven. 1999. “Recognition of cognates and interlingual
homographs: The neglected role of phonology”, Journal of Memory and Language 41/ 4: 496
–518.
Dijkstra, A., M. Timmermans and H. Schriefers. 2000. “On being blinded by your other lan-
guage: Effects of task demands on interlingual homograph recognition”, Journal of Memory
and Language 42: 445–464.
Dijkstra, T. and J.G. van Hell. 2003. Testing the Language Mode hypothesis using trilinguals.
InternationalJournal of Bilingual Education and Bilingualism, 6: 2–16.
Dijkstra, T., K. Miwa, B. Brummelhuis, M. Sappelli and H. Baayen. 2010. “How cross–language
similarity and task demands affect cognate recognition”, Journal of Memory and Language
62: 284–301.
Diller, K. C. 1978. The Language Teaching Controversy. Rowley, MA: Newbury House.
Dimitropoulou, M., J.A. Duñabeitia and M. Carreiras. 2010. “Masked translation priming effects
with low proficient bilinguals”,Experimental Psychology 57 98–107.
Dmochowska, H. (ed.) 2010. Mały Rocznik Statystyczny – Polska 2010. GUS. Zakład
Wydawnictw Statystycznych: Warszawa.
Dörnyei, Z. 2005. The psychology of the language learner: Individual differences in second lan-
guage acquisition. Mahwah, NJ: Lawrence Erlbaum.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 197
Doughty, C. and M. Long (eds.). 2003. Handbook of Second Language Acquisition. Cambridge,
MA: Blackwell Publishers.
Dufour, R. and J.F. Kroll. 1995. “Matching words to concepts in two languages: A test of the
concept mediation model of bilingual representation”, Memory and Cognition 23: 166–180.
Duyck, W. 2005. “Translation and associative priming with cross–lingual pseudohomophones:
evidence for nonselective phonological activation in bilinguals”, Journal of Experimental
Psychology: Learning Memory and Cognition 31: 1340–59.
Duyck, W. and M. Brysbaert. 2004. “Forward and backward number translation requires con-
ceptual mediation in both balanced and unbalanced bilinguals”, Journal of Experimental
Psychology: Human Perception and Performance 30: 889–906.
Duyck, W., D. Vanderelst, T. Desmet and R.J. Hartsuiker. 2008. “The frequency effect in second–
language visual word recognition”, Psychonomic Bulletin and Review 15/4: 850–855.
Dziubalska–Kołaczyk, K. (ed.). 2006. IFAtuation: A Life in IFA. A Festschrift for Professor Jacek
Fisiak on the Occasion of His 70th Birthday. Poznań: Wydawnictwo Naukowe UAM.
Ecke, P. 2009. “The tip–of–the–tongue phenomenon as a window on (bilingual) lexical retrieval”,
in: A. Pavlenko (ed.) 185–209.
Ecke, P., C.J. Hall. 2013. “Tracking tip-of-the-tongue states in a multilingual speaker: Evidence
of attrition or instability in lexical systems?”, International Journal of Bilingualism
17/6: 734–751.
Edwards, J. 1994. Multilingualism. London: Routledge.
Edwards, J.R. 2012. Multilingualism: Understanding Linguistic Diversity. Continuum.
Ellis, A.W. and A.W. Young. 1996. Human cognitive neuropsychology. Hove, UK: Erlbaum
Ellis, N. and A. Beaton. 1993. “Factors affecting foreign language vocabulary: imagery key-
word mediators and phonological short-term memory”, Quarterly Journal of Experimental
Psychology 46A: 533–558.
Ellis, N.C. and A. Beaton. 1995. “Psycholinguistic determinants of foreign language vocabulary
learning”, in: B. Harley (ed.), 107–165.
Elston-Güttler, K.E., T.C. Gunter and S.A. Kotz. 2005. “Zooming into L2: Global language con-
text and adjustment affect processing of interlin-gual homographs in sentences”, Cognitive
Brain Research 25/1: 57–70.
Elwert, W. 1973. “Das zweisprachige Individuum: ein Selbstzeugnis“, in: Studien zu
romanischen Sprachen. Bd. 4 1–81. Wisebaden: Franz Steiner Verlag.
Emmorey, K. and V. Fromkin. 1988. The mental lexicon, in: F. Newmeyer (ed.), 124–149.
Evens M. (ed.). 1988. Relational Models of the Lexicon. Cambridge: Cambridge University
Press.
Fay, D. and A. Cutler. 1977. Malapropisms and the structure of the mental lexicon. Linguistic
Inquiry 8: 505–520.
Feldman L. (ed.). 1995. Morphological Aspects of Language Processing: Cross-Linguistic
Perspectives. Hillsdale, NJ: Lawrence Erlbaum Associates. Clahsen, H. 2006. “Linguistic
perspectives on morphological processing”, in: D. Wunderlich (ed.), 355–388.
Ferguson C.A., L. Menn and C. Stoel–Gammon (eds.). 1992. Phonological Development:
Models, Research, Implication. Timonium, MD: York Press.
Fernández, E.M. and H. Smith Cairns. 2011. Fundamentals of Psycholinguistics. Hoboken, NJ:
Wiley–Blackwell.
Field, J. 2004. Psycholinguistics: The Key Concepts. London: Routledge.
Fikkert, P. 2005. “Getting sounds structures in mind. Acquisition bridging linguistics and psy-
chology?” in: A.E. Cutler (ed.), 43–56.
Fikkert, P. 2007. “Acquiring phonology”, in: P. de Lacy (ed.), 537–554.
Fishman, J.A. (ed.). 1968. Readings in the Sociology of Language. The Hague: Mouton.
Fitzpatrick, T. 2007. “Word association patterns: unpacking the assumptions”, International
Journal of Applied Linguistics 17/3: 319–331.
Fitzpatrick, T. and C. Izura. 2011. “Word association in L1 and L2”, Studies in Second Language
Acquisition 33: 373–398.
Fodor, J. 1983. The Modularity of Mind: an Essay on Faculty Psychology. Cambridge, MA: MIT
Press.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
198 References
Fodor, J. 1989. “Why should the mind be modular?”, in: A. George (ed.), 1–22.
Forster, K.I. 1976. “Accessing the mental lexicon”, in: R. J. Wales and E.W. Walker (eds.),
257–287.
Forster, K.I. 1989. Basic issues in lexical processing, in: W.D. Marslen–Wilson (ed.), 75–107.
Forte, R. 2009. “Transfer in L3 Acquisition: The Role of Typology”, in: Y.I. Leong (ed.), 89–15.
Franceschini R. 2009. “The genesis and development of research in multilingualism: perspectives
for future research”, in: L. Aronin and B. Hufeisen (eds.), 27–61.
Franceschini, R., D. Zappatore and C. Nitsch. 2003. “Lexicon in the Brain: What Neurobiology
has to Say about Languages”, in: J. Cenoz, B. Hufeisen and U. Jessner. (eds.), 153–166.
Frederiksen, J.R. and J.F. Kroll. 1976. Spelling and sound: Approaches to the internal lexicon.
Journal of Experimental Psychology: Human Perception and Performance 2/3: 361–379.
French, R.M. 1998. “A simple recurrent network model of bilingual memory”, in: M.A.
Gernsbacher, M.A. and S.J. Derry. (eds.), 368–373.
French, R.M. and M. Jacquet. 2004. “Understanding bilingual memory: models and data”,
Trends in Cognitive Sciences 8: 87–93.
Friel, B.M. and S.M. Kennison. 2001. “Identifying German–English cognates, false cognates,
and non–cognates: methodological issues and descriptive norms”, Bilingualism: Language
and Cognition 4: 249–274.
Fromkin, V.A. (ed.). 1973. Speech Errors as Linguistic Evidence. The Hague: Mouton.
Frost, R. and J. Ziegler. 2007. “Speech and Spelling Interaction: The Interdependence of Visual
and Auditory Word Recogntion”, in: G. Gaskell (ed.), 107–118
Gabryś–Barker, D. 2005. Aspects of Multilingual Storage, Processing and Retrieval. Katowice:
Wydawnictwo Uniwersytetu Śla˛skiego.
Garfinkel, A. and K.E.Tabor. 1991. “Elementary School Foreign Languages and English Reading
Achievement: A New View of the Relationship”. Foreign Language Annals 24: 375–382.
Garman, M. 1990. Psycholinguistics. Cambridge: Cambridge University Press.
Garrod, S. 2006. “Psycholinguistic Research Methods”, in Encyclopedia of Language &
Linguistics (2nd Edition), Vol. 10. S. 251–257.
Gaskell, G. (ed.). 2007. Oxford Handbook of Psycholinguistics. Oxford: Oxford University Press.
Gass, S, and L. Selinker (eds.). 1983. Language Transfer in Language Learning. Rowley, MA:
Newbury House.
Gass, S. M. and L. Selinker. 1994. Second Language Acquisition. Hillsdale, NJ: Lawrence
Erlbaum.
Genesee, F. 2002. “Portrait of the bilingual child”, in: V.J. Cook (ed.), 167–197.
George, A. (ed.), 1989. Reflections on Chomsky. Oxford: Blackwell.
Gerard, L.D. and D.L. Scarborough, 1989. “Language–specific access of homographs by bilin-
guals”, Journal of Experimental Psychology: Learning, Memory, and Cognition 15: 305–315.
Gernsbacher, M.A. and S.J. Derry. (eds.), Proceedings of the Twentieth Annual Cognitive Science
Society Conference. NJ: LEA.
Goddard, H.H. 1917. “Mental tests and the immigrant”, Journal of Delinquency 2: 243–277.
Goldrick, M., J. Folk and B. Rapp. 2010. “Mrs. Malaprop's neighborhood: Using word errors to
reveal neighborhood structure”, Journal of Memory and Language 62: 113–134.
Gollan, T.H., K.I. Forster, and R. Frost. 1997. “Translation priming with different scripts:
Masked priming with cognates and noncognates in Hebrew–English bilinguals”, Journal of
Experimental Psychology: Learning, Memory, and Cognition 23: 1222–1139.
Gomez, P. et al. 2013. “A diffusion model account of masked versus unmasked priming: are
they qualitatively different?”, Journal of Experimental Psychology: Human Perception and
Performace, 39/6: 1731–1740.
Graddol, D. 1997. “The Future of English? A guide to forecasting the popularity of the English
language in the 21st century”, The British Council. Retrieved: 21st November 2009 from
www.britishcouncil.org/learning–elt–future.pdf.
Graddol, D. 2006. “English Next”, The British Council. Retrieved: 4th January 2011 from
www.britishcouncil.org/learning–research–english–next.pdf.
Grainger, J. and A. Jacobs (eds.), Localist Connectionist Approaches to Human Cognition.
Hillsdale, NJ: Lawrence Erlbaum.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 199
Grainger, J. and C. Beauvillain. 1987. “Language blocking and lexical access in bilinguals”,
Quarterly Journal of Experimental Psychology 39A: 295–319.
Grainger, J. and Ch. Frenck-Mestre. 1998. “Masked Priming by Translation Equivalents in
Proficient Bilinguals”, Language and Cognitive Processes 13/ 6: 601–623.
Grainger, J. and T. Dijkstra. 1992. “On the representation and use of language information in
bilinguals”, in: R.J. Harris (ed.), 207–220.
Grauberg, W. 1971. “An error analysis in German of first–year university students”, in: G. Perren
and J. Trim (eds.), 11–26.
Green, D.W. 1986. “Control, activation and resource”, Brain and Language27: 210–223.
Green, D.W. 1998. “Mental Control of the Bilingual Lexico–Semantic System. Bilingualism”,
Language and Cognition 1: 67–81.
Greenberg, J.H. (ed.). 1978. Universals of Language: Method and Theory, vol. l., Stanford, CA:
Stanford University Press.
Grosjean F. and P. Li (eds.). 2013. The Psycholinguistics of bilingualism. Malden, MA & Oxford:
Wiley-Blackwell.
Grosjean, F. (ed.). 2008. Studying bilinguals. Oxford: Oxford University Press.
Grosjean, F. 1982. Life with Two Languages: An Introduction to Bilingualism. Cambridge, MA:
Harvard University Press.
Grosjean, F. 1985. “The bilingual as a competent but specific speaker–hearer”, Journal of
Multilingual and Multicultural Development 6/6: 467–477.
Grosjean, F. 1989. “Neurolinguists, beware! The bilingual is not two monolinguals in one per-
son”, Brain and Language 36: 3–15.
Grosjean, F. 1992. Another view of bilingualism, in: R.J. Harris (ed.), 51–62.
Grosjean, F. 1997. “Processing mixed languages: Issues, findings and models” in: A.M.B. De
Groot and J.F. Kroll (eds.), 225–255.
Grosjean, F. 2001. “The bilingual’s language modes”, in: J.L. Nicol (ed.), 1–22.
Grosjean, F. 2008. Studying Bilinguals. Oxford–NewYork: Oxford University Press.
Grosjean, F. 2010. Bilingual: Life and Reality. Cambridge, MA: Harvard University Press.
Gruneberg, M., P. Morris and R. Sykes (eds.). 1988. Practical Aspects of Memory, Vol. 2.
London: John Wiley and Sons.
Guasch, M., R. Sánchez-Casas, P. Ferré and J.E. García-Albea. 2011. “Effects of the degree
of meaning similarity on cross-language semantic priming in highly proficient bilin-
guals”, Journal of Cognitive Psychology 23/8: 942–961.
Gubbins, P. and M. Holt (eds.). 2002. Beyond Boundaries: Language and Identity in
Contemporary Europe. Clevedon: Multilingual Matters.
Gullifer, J.W., J.F. Kroll and P.E. Dussias. 2013. “When Language Switching has No Apparent
Cost: Lexical Access in Sentence Context“, Frontiers in psychology 4: 1–13.
Haigh, C. A. and D. Jared. 2007. “The activation of phonological representations by bilinguals
while reading silently: Evidence from interlingual homophones”, Journal of Experimental
Psychology: Learning, Memory & Cognition 33:623–644.
Hakuta, K. 1986. Mirror of Language. The Debate on Multilingualism. New York: Basic Books.
Hall, C.J. 1992. “Making the right connections: Vocabulary learning and the mental lexi-
con”, Unpublished manuscript, Universidad de las Américas, Puebla: ERIC Document
Reproduction Service No. ED 363 128.
Hall, C.J. 2002. “The automatic cognate form assumption: Evidence for the Parasitic Model of
vocabulary development”, International Review of Applied Linguistics 40: 69–87.
Hall, C.J. and P. Ecke. 2003. “Parasitism as a default mechanism in L3 vocabulary acquisition”,
in: J. Cenoz, B. Hufeisen and U. Jessner (eds.), 71–85.
Hall, C.J., D. Newbrand, P. Ecke, U. Sperr, V. Marchand and L. Hayes. 2009. “Learners’ implicit
assumptions about syntactic frames in new L3 words: The role of cognates, typological prox-
imity and L2 status”, Language Learning 59/1: 153–202.
Hamers, J. and M. Blanc. 1989. Bilinguality and Bilingualism. Cambridge: Cambridge
University Press.
Han, Z.–H. (ed.), 2007. Understanding Second Language Process. Clevedon: Multilingual
Matters.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
200 References
Harding, E. and P. Riley. 1986. The Bilingual Family: a Handbook for Parents. Cambridge:
Cambridge University Press.
Harley, B. (ed.). 1995. Lexical Issues in Language Learning. Amsterdam: John Benjamins.
Harley, T.A. 2004. “Does cognitive neuropsychology have a future?”, Cognitive
Neuropsychology 21: 3–16.
Harley, T.A. 2010. Talking the Talk: Language, Psychology and Science. Hove: Psychology
Press.
Harley, T.A. 2008. The psychology of language: from data theory. London: Psychology Press.
Harris, R.J. (ed.). 1992. Cognitive Processing in Bilinguals. Amsterdam: Elsevier Science
Publishers.
Hartsuiker, R.J., M.J. Pickering and E. Veltkamp. 2004. “Is syntax separate or shared between
languages?”, Psychological Science 15: 409–414.
Hartsuiker, R.J., M.J. Pickering. 2008. “Language integration in bilingual sentence production”,
Acta Psychologica 128: 479–489.
Haugen, E. 1956. Bilingualism in the Americas. Alabama: American Dialect Society.
Haverkort, M. 2005. “Linguistic representation and language use in aphasia”, in: A. Cutler. (ed.),
57–68.
Henderson, L., J. Wallis and K. Knight. 1994. “Morphemic structure and lexical access”, in: H.
Bouma and D. Bouwhuis (eds.), 211–226.
Herdina, P. and U. Jessner. 2000. The dynamics of third language acquisition, in: J. Cenoz and U.
Jessner (eds.), 84–98.
Herdina, P. and U. Jessner. 2002. A Dynamic Model of Multilingualism. Clevedon: Multilingual
Matters.
Heredia, R. and J. Altarriba (eds.). 2002. Bilingual Sentence Processing. Amsterdam: Elsevier.
Heredia, R.R. 2008. “Mental models of bilingual memory”, in: J. Altarriba and R.R. Heredia.
(eds.), 39–67.
Herwig, A. 2001. “Plurilingual lexical organisation: Evidence from lexical processing in L1–L2–
L3–L4 translation”, in: J. Cenoz, B. Hufeisen and U. Jessner (eds.). 2001a, 115–137.
Hinger, B. 2001. “An intensive multilingual approach to language acquisition in school context”,
in: J. Cenoz, B. Hufeisen and U. Jessner (eds). 2001b, 121–136.
Hockett, C.F. 1958. A Course in Modern Linguistics. New York: Macmillan.
Hoffmann, C. and A. Stavans. 2007. “The evolution of trilingual code–switching from infancy
to school age: the shaping of trilingual competence through dynamic language dominance”,
InternationalJournal of Bilingualism 11/1: 55–72.
Hoffmann, C. and J. Ytsma. (eds.). 2004. Trilingualism in Family, School and Community.
Clevedon: Multilingual Matters.
Hoffmann, C. 2000. “The spread of English and the growth of multilingualism with English in
Europe”, in: J. Cenoz and U. Jessner (eds.), 1–21.
Hoffmann, C. 2001a. “Towards a description of trilingual competence”, InternationalJournal of
Bilingualism 5/1: 1–17.
Hoffmann, C. 2001b. “The status of trilingualism in bilingualism studies”, in: J. Cenoz, B.
Hufeisen and U. Jessner (eds.), 2001a, 13–18.
Hoffmann, C. and J. Ytsma. (eds.) 2004. Trilingualism in Family, School and Community.
Clevedon: Multilingual Matters.
Holquist, M. (ed.). 1981. The dialogic imagination: Four essays by M. M. Bakhtin. Austin, TX:
University of Texas Press.
Hoshino, N. and J.F. Kroll. 2008. “Cognate effects in picture naming: Does cross–language acti-
vation survive a change of script?”, Cognition 106: 501–511.
House, J. 2003. “English as a lingua franca: A threat to multilingualism?”, Journal of
Sociolinguistics 7/4: 556–578.
Howes, D.H. and R.L. Solomon. 1951. Visual duration threshold as a function of word probabil-
ity”, Journal of Erperimental Psychology 41: 401–410.
Hu, G. and N. Jiang. 2011. “Semantic integration in listening comprehension in a second language
Evidence from cross-modal priming”, in: P. Trofimovich and K. McDonough (eds.), 199–218.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 201
Hudson, P.T.W. and M.W. Bergman. 1985. “Lexical knowledge in word recognition: Word length
in naming and lexical decision tasks”, Journal of Memory & Language 24: 46–58.
Hufeisen, B. 1991. Englisch als erste und Deutsch als zweite Fremdsprache. Frankfurt am Main:
Peter Lang Verlag.
Hufeisen, B. 2004. “A brief introduction to the linguistic foundations”, in: B. Hufeisen and G.
Neuner (eds.), 7–13.
Hufeisen, B. and Aronin, L. 2009. “On the genesis and development of L3 research, multilingual-
ism and multiple language acquisition”, in: L. Aronin and B. Hufeisen (eds.), 1–11.
Hufeisen, B., and G. Neuner, (eds.). 2004. The Plurilingualism Project: Tertiary language learn-
ing – German after English. Graz: European Centre for Modern Languages.
Hufeisen, B. and N. Marx. 2007. “How can DaFnE and EuroComGerm contribute to the concept
of receptive multilingualism?”, in: J. ten Thije and L. Zeevaert (eds.), 307–321.
Husén, T. and S. Opper (eds.). 1983. Multicultural and Multilingual Education in Immigrant
Countries. Wenner Gren Symposium Series (38), Oxford: Pergamon Press.
Hymes, D. 1964. Language in Culture and Society. New York: Harper and Row.
Iatcu, T. 2000. “Teaching English as a third language to Hungarian–Romanian bilinguals”, in: J.
Cenoz and U. Jessner (eds.), 157–178.
Ibrahim, R. 2009. “The cognitive basis of diglossia in Arabic: Evidence from a repetition priming
study within and between languages”, Psychology research and behavior management 2: 93.
Jared, D. and C. Szucs. 2002. “Phonological activation in bilinguals: Evidence from interlingual
homograph naming”, Bilingualism: Language and Cognition 5/3: 225–239.
Jared, D. and J.F. Kroll. 2001. “Do bilinguals activate phonological representations in one or
both of their languages when naming words?”, Journal of Memory and Language 44: 2–31.
Jarvis, S. and A. Pavlenko. 2008. Cross-linguistic influence in language and
cognition. Routledge.
Jenkins, J. 2012. “English as a Lingua Franca from the classroom to the classroom” ELT Journal
66, (4): 486–494.
Jescheniak, J.D. and W.J.M. Levelt. 1994. “Word frequency effects in speech production:
retrieval of syntactic information and of phonological form”, Journal of Experimental
Psychology: Learning, Memory, and Cognition 20: 824–843.
Jespersen, O. 1922. Language. London: Allen and Unwin.
Jessner, U. 1996. “Towards a dynamic view of multilingualism”, in: M. Pütz. (ed.), 17–30.
Jessner, U. 2006. Linguistic Awareness in Multilinguals: English as a Third Language.
Edinburgh: Edinburgh University Press.
Jessner, U. 2008a. “Teaching third languages: Findings, trends and challenges”, Language
Teaching 41: 15–56.
Jessner, U. 2008b. “A DST model of multilingualism and the role of metalinguistic awareness”,
The Modern Language Journalm92/2: 270–283.
Jessner, U. 2013a. “Dynamics of multilingualism”, in: C.A. Chapelle (ed.), 1798–1805.
Jessner, U. 2013b. “Teaching a Third Language”, in: C.A. Chapelle (ed.), 5662–5665.
Jiang N. and K.I. Forster. 2001. “Cross–language priming asymmetries in lexical decision and
episodic recognition”, Journal of Memory and Language 44: 32–51.
Joaristi, L., L. Lizasoain, J.F. Lukas, and K. Santiago. 2009. “Trilingualism (Spanish, English
and Basque) in the educational system of the Basque country”, InternationalJournal of
Multilingualism 6/1:105–126.
Kaplan, R. (ed.). 2002. Handbook of Applied Linguistics. Oxford: Oxford University Press.
Karpińska–Szaj, K. (ed.), Nauka je˛zyków obcych w dobie integracji europejskiej. Łask: Oficyna
Wydawnicza LEKSEM.
Kecskes, I. 2010. “Situation-bound utterances as pragmatic acts”, Journal of Pragmatics 42, 11:
2889–2897.
Kecskes, I. and L. Albertazzi (eds.). 2007. Cognitive Aspects of Bilingualism. New York:
Springer.
Kellerman, E. 1983. “Now you see it, now you don’t”, in: S. Gass and L. Selinker (eds.),
112–134.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
202 References
Kellerman, E. 1978. “Giving Learners a Break: Native Language Intuitions as a Source of
Predictions about Transferability”, Working Papers on Bilingualism 15: 59–92.
Kemp, Ch. 2009. “Defining multilingualism”, in: L. Aronin and B. Hufeisen. (eds.), 11–27.
Kempen, G. and P. Hujbers. 1983. “The lexicalization process in sentence production and nam-
ing: Indirect election of words”, Cognition 14: 185–209.
Kerkhofs, R., A. Dijkstra, D. Chwilla and E. De Bruijn. 2006. “Testing a model for bilingual
semantic priming with interlingual homographs: RT and ERP effects”, Brain Research 1068:
170–183.
Kettemann, B. and W.W. Wieden (eds.). 1993. Current Issues in European Second Language
Research. Tübingen: Gunter Narr.
Kim, A. and V.T. Lai. 2012. “Rapid interactions between lexical semantic and word form
analysis during word recognition in context: Evidence from ERPs”, Journal of Cognitive
Neuroscience 24: 1104–1112.
Kim, J. and C. Davis. 2003. Task effects in masked cross–script translation and phonological
priming”, Journal of Memory and Language 49: 484–499.
Kirk–Greene, C.W.E. 1981. French False Friends. London: Routledge and Kegan Paul.
Kirsner, K., E. Lalor, and K. Hird. 1993. “The bilingual lexicon: Exercise, meaning and morphol-
ogy”, in: R. Schreuder and B. Weltens (eds.), 215–248.
Klein, E.C. 1995. “Second versus third language acquisition”, Language Learning 45: 419–465.
Koessler, M. 1975. Les faux amis des vocabulaires anglais et americain. Paris: Libraire Vuibert.
Koessler, M. and J. Derocquigny. 1946. Les faux amis. Paris: Librairie Vuibert.
Kolers, P.A. 1963. “Interlingual word associations”, Journal of Verbal Learning and Verbal
Behaviour 2: 291–300.
Komorowska, H. 2013. “Multilingualism: Its open and hidden agendas” SSLLT 3, 4: 463–482.
Kovelman, I., S.A. Baker and L.A. Petitto. 2008. “Bilingual and monolingual brains compared:
a functional magnetic resonance imaging investigation of syntactic processing and a possible
“neural signature” of bilingualism”, Journal of cognitive neuroscience 20/1: 153–169.
Kramsch, C. 2006. “The multilingual subject”, InternationalJournal of Applied Linguistics 16/1:
97–110.
Kramsch, C. 2014. “Teaching Foreign Languages in an Era of Globalization: Introduction”, The
Modern Language Journal 98/ 1: 296–311.
Kroll, J.F. 1993. “Assessing conceptual representations for words in a second language”, in: R.
Schreuder and B. Weltens (eds.), 53–81.
Kroll, J.F. and A. Dijkstra. 2002. The bilingual lexicon, in: R. Kaplan (ed.), 301–321.
Kroll, J.F. and A.M.B. De Groot (eds.). 2005. Handbook of Bilingualism: Psycholinguistic
Approaches. Oxford: Oxford University Press.
Kroll, J.F. and A.M.B. De Groot. 1997. “Lexical and conceptual memory in the bilingual:
Mapping form to meaning in two languages”, in: A.M.B. De Groot and J.F. Kroll (eds.),
169–199.
Kroll, J.F. and E. Stewart. 1994. “Category interference in translation and picture naming:
Evidence for asymmetric connections between bilingual memory representations”, Journal of
Memory and Language 33: 149–174.
Kroll, J.F. and G. Sunderman. 2003. “Cognitive processes in second language acquisition: The
development of lexical and conceptual representations”, in: C. Doughty and M. Long (eds.),
104–129.
Kroll, J.F. and J. Curley. 1988. “Lexical memory in novice bilinguals: The role of concepts
in retrieving second language words”, in: M. Gruneberg, P. Morris and R. Sykes (eds.),
389–395.
Kroll, J.F. and N. Tokowicz. 2005. Models of bilingual representation and processing, in: J.F.
Kroll and A.M.B. De Groot (eds.), 531–553.
Kroll, J.F. and P. Dussias. 2004. “The comprehension of words and sentences in two languages”,
in: T. Bhatia and W. Ritchie. (eds.), 169–200.
Labov, W. 1973. The boundaries of words and their meanings, in: C. N. Bailey and R. W. Shuy
(eds.), 340–373.
Lado, R. 1957. Linguistics Across Cultures. Ann Arbor: University of Michigan Press.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 203
Lakoff, G. 1982. “Categories”, in: Linguistic Society of Korea, eds., 1–88. Linguistics in the
Morning Calm. Seoul: Hanshin.
Lakoff, G. 1987. Women, Fire and Dangerous Things: What Categories Reveal about the Mind.
Chicago: University of Chicago Press.
Lalor, E. and K. Kirsner. 2000. “Cross–lingual transfer effects between English and Italian cog-
nates and noncognates”, InternationalJournal of Bilingualism 4: 385–398.
Langacker R.W. 1987. Foundations of Cognitive Grammar, Volume I, Theoretical
Prerequisites. Stanford, California: Stanford University Press.
Lasagabaster, D. 2000. “Three languages and three linguistic models in the Basque educational
system”, in: J. Cenoz and U. Jessner (eds.), 157–178.
Lasagabaster, D. 2001. “University students` attitudes towards English as an L3”, in: J. Cenoz, B.
Hufeisen and U. Jessner. (eds.). 2001a, 43–52.
Lemhöfer, K. and A. Dijkstra. 2004. “Recognizing cognates and interlingual homographs:
Effects of code similarity in generalized lexical decision”, Memory and Cognition 32: 4,
533–550.
Lemhöfer, K. and R. Radach. 2009. “Task context effects in bilingual nonword processing.
Experimental Psychology”, 561: 41–47.
Lemhöfer, K., A. Dijkstra and M. Michel. 2004. “Three languages, one ECHO: Cognate effects
in trilingual word recognition”, Language and Cognitive Processes 19: 585–611.
Lemhöfer, K., A. Dijkstra, H. Schriefers, R.H. Baayen, J. Grainger and P. Zwitserlood. 2008.
“Native language influences on word recognition in a second language: A mega–study”,
Journal of Experimental Psychology: Learning, Memory, and Cognition 34: 12–31.
Leong,Y.I. (ed.). 2009. Third Language Acquisition and Universal Grammar. Clevedon:
Multilingual Matters.
Leopold, W. (Volume 1–1939, Volume 2– 1947, Volume 3 and 4– 1949. Speech Development of a
Bilingual Child. A Linguist`s Record. Evanston. Ill.: Northwestern University Press.
Levelt, W.J.M. 1989. Speaking: from Intention to Articulation. Cambridge, MA: Bradford
Books/MIT Press.
Levelt, W.J.M. 1993. Lexical Access in Speech Production. Oxford: Blackwell.
Levelt, W.J.M., A. Roelofs and A.S. Meyer. 1999. “A theory of lexical access in speech produc-
tion”, Behavioral and Brain Sciences 22: 1–38.
Lewis, M.P., G.F. Simons and C.D. Fennig (eds.). 2014. Ethnologue: Languages of the World,
Seventeenth edition. Dallas, Texas: SIL International. Online version: http://www.ethnologu
e.com.
Léwy, N. and F. Grosjean. 2008. The Léwy and Grosjean BIMOLA model, in: F. Grosjean. (ed.),
201–210.
Li Wei. 2003. “Activation of lemmas in the multilingual mental lexicon and transfer in third lan-
guage learning”, in: J. Cenoz, B. Hufeisen and U. Jessner (eds.), 57–70.
Li Wei. 2006. “The multilingual mental lexicon and lemma transfer in third language learning”,
International Journal of Multilingualism 32: 88–104.
Li Wei (ed.). 2007. The Bilingualism Reader. London: Routledge.
Li Wei. 2008. “Research perspectives on bilingualism and multilingualism”, in: L. Wei and M.G.
Moyer (eds.), 1–17.
Li Wei and M.G. Moyer (eds.). 2008. The Blackwell Guide To Research Methods In Bilingualism
And Multilingualism. Oxford: Blackwell.
Li, P. and I. Farkaš. 2002. “A self–organizing connectionist model of bilingual processing”, in: R.
Heredia and J. Altarriba (eds.), 59–85.
Li, P., X. Zhao and B. MacWhinney. 2007. “Dynamic self-organization and early lexical develop-
ment in children”, Cognitive Science 31: 581–612.
Liberman, A.M., F.S. Cooper, D.P. Shankweiler and M. Studdert–Kennedy. 1967. “Perception of
the speech code”, Psychological Review 74: 431–461.
Lotto, L. and A.M.B. De Groot. 1998. “Effects of learning method and word type on acquiring
vocabulary in a new language”, Language Learning 48: 31–69.
Luo L., G. Luk Bial and E. Bialystok. 2010. “Effect of Language Proficiency and Executive
Control on Verbal Fluency Performance in Bilinguals”, Cognition 114: 29–41.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
204 References
MacKay, D.G. 1978. “Derivational rules and the internal lexicon”, Journal of Verbal Learning
and Verbal Behaviour 17: 61–71.
Mackey, W.F. 1968. “The description of bilingualism”, in: J.A. Fishman (ed.). 554–584.
MacLeod, C.M. 1991. “Half a century of research on the Stroop effect: An integrative review”,
Psychological Bulletin 109: 163–203.
Macnamara, J. 1967. “The bilinguals linguistic performance”, Journal of Social Issues 232:
58–77.
Malakoff, M.E. 1992. “Translation ability: A natural bilingual and metalinguistic
skill”, Advances in psychology, 83: 515–529.
Manchón, R,M. 2013. “Writing“, in: F. Grosjean and P.Li. (eds.), 100–116.
Marcus, S. and U. Frauenfelder. 1985. “Word recognition – uniqueness or deviation? A theoreti-
cal note”, Language and Cognitive Processes 1: 163–169.
Marian, V. and M. Spivey. 2003. “Competing activation in bilingual language processing:
Within– and between–language competition”, Bilingualism: Language and Cognition 6:
97–115.
Marian, V., H.K. Blumenfeld, E. Mizrahi, U. Kania, and A-K Cordes. 2013.“Multilingual Stroop
performance: Effects of trilingualism and proficiency on inhibitory control”, International
Journal of Multilingualism 10/1: 82–104.
Marslen–Wilson, W. and P. Warren. 1994. “Levels of perceptual representation and processes in
lexical access: Words, phonemes, and features”, Psychological Review 101: 653–675.
Marslen–Wilson, W. and A. Welsh, A. 1978. “Processing interactions and lexical access during
word recognition in continuous speech”, Cognitive Psychology 10: 29–63.
Marslen–Wilson, W. and L.K. Tyler. 1980. “The temporal structure of spoken language under-
standing”, Cognition 8: 1–71.
Marslen–Wilson, W.D. (ed.). 1989. Lexical Representation and Process. Cambridge, MA: MIT
Press.
Marslen-Wilson, W.D. 2007. “Morphological processes in language comprehension”, in G,
Gaskell. (ed.), 175–193.
Marslen-Wilson, W.D. and L.K. Tyler. 2007. “Morphology, language and the brain: the
decompositional substrate for language comprehension”, Philos Trans R Soc Lond B Biol
Sci 362(1481): 823–36.
Marx, N. and B. Hufeisen. 2004. “Critical overview of research on third language acquisition and
multilingualism published in the German language”, InternationalJournal of Multilingualism
1/ 2: 141–154.
McClelland, J.L. and D.E. Rumelhart. 1981. “An interactive activation model of context effect in
letter perception: Part 1. An account of basic findings”, Psychological Review 88: 375–407.
McCulloch, W.S. and W. Pitts. 1943. “A logical calculus of ideas immanent in nervous activity”,
Bulletin of Mathematical Biophysics 5: 115–133.
McLaughlin, B. and R. Nation. 1986. “Experts and novices: An information–processing
approach to the ‘good language learner’ problem”, Applied Psycholinguistics 7: 41–56.
Meara, P. 1982. “Word associations in a foreign language:, Nottingham Linguistics Circular 11:
29–38.
Melčuk, I. and A. Zholkovsky. 1988. “The Explanatory Combinatorial Dictionary”, in: M. Evens
(ed.), 41–74.
Midgley, K., P.J. Holcomb and J. Grainger. 2011. “Effects of cognate status on word comprehen-
sion in second language learners: an ERP investigation”, Journal of Cognitive Neuroscience
23: 1634–1647.
Miozzo, M., A, Costa, M, Hernández and B. Rapp. 2010. “Lexical processing in the bilingual
brain: Evidence from grammatical/morphological deficits”, Aphasiology 24: 262–287.
Mirman, D. 2011. “Effects of near and distant semantic neighbors on word produc-
tion”, Cognitive, Affective, and Behavioral Neuroscience 11/1: 32–43.
Mitchell, R. and F. Myles. 2004. Second language learning theories. Second edition. London:
Hodder Arnold.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 205
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
206 References
Pavlenko, A. (ed.). 2009. The Bilingual Mental Lexicon: Interdisciplinary Approaches. Clevedon:
Multilingual Matters.
Pavlenko, A. 2011. Thinking and speaking in two languages. Clevedon, UK: Multilingual
Matters.
Pawlak, M. 2007. “Questions from the language classroom revisited: What does SLA
research have to tell us about effective foreign language pedagogy?” in: J. Arabski (ed.),
57–72.
Peal, E. and W. Lambert. 1962. “The relation of bilingualism to intelligence”, Psychological
Monographs 76: 1–23.
Pecher, D. 2001. “Perception is a two–way junction: Feedback semantics in word recognition”,
Psychonomic Bulletin and Review 8: 545–551.
Peeters, D., T. Dijkstra and J. Grainger. 2013. “The representation and processing of identical
cognates by late bilinguals: RT and ERP effects”, Journal of Memory and Language 68:
315–332.
Perren G. and J. Trim (eds.). 1971. Applications in Linguistics. London: Cambridge University
Press.
Pickering, M.J. and S. Garrod. 2005. “Establishing and using routines during dialogue:
Implications for psychology and linguistics”, in: A. Cutler (ed.), 85–101.
Pickering, M.J. and S. Garrod. 2013. “An integrated theory of language production and compre-
hension”, Behavioral and Brain Sciences 36/4: 329–347.
Potter, M.C. 1979. “Mundane symbolism: The relations among objects, names, and ideas”, in:
N.R. Smith and M.B. Franklin (eds.), 281–294.
Potter, M.C., K.F. So, B. Von Eckardt and L. Feldman. 1984. “Lexical and conceptual representa-
tion in beginning and proficient bilinguals”, Journal of Verbal Learning and Verbal Behavior
23/ 1: 23–38.
Pugh, K.R., K. Rexer and L. Katz. 1994. “Evidence of flexible coding in visual word recog-
nition”, Journal of Experimental Psychology: Human Perception and Performance 20:
807–825.
Pütz M. and L. Sicola (eds.). 2010. Cognitive Processing in Second Language Acquisition: Inside
the learner's mind. Amsterdam / Philadelphia: John Benjamins.
Pütz, M. (ed.). 1996. Language Choices: Conditions, Constraints and Consequences.
Amsterdam: John Benjamins.
Quetz, J. and G. von der Handt (eds.). 2002. Neue Sprachen lehren und lernen.
Fremdsprachenunterricht in der Weiterbildung. Bielefeld: Bertelsmann.
Randall, M. 2007. Memory, psychology and second language learning. Amsterdam: John
Benjamins.
Rastle, K. and M.H. Davis. 2008. “Morphological decomposition based on the analysis of
orthography”, Language and Cognitive Processes 23: 942–971.
Ratcliff, R. 1978. “A theory of memory retrieval”, Psychological Review 85: 59–109.
Ratcliff, R., A. Thapar, P. Gomez and G. McKoon. 2004. “A diffusion model analysis of the
effects of aging in the lexical-decision task”, Psychology and Aging 19: 278–289.
Reichle, E.D. and C.A. Perfetti. 2003. “Morphology in word identification: A word–experience
model that accounts for morpheme frequency effects”, Scientific Studies of Reading, 71:
219–238.
Reid, A.A. and W.D. Marslen-Wilson. 2003. “Lexical representation of morphologically complex
words: Evidence from Polish”, in: R.H. Baayen and R. Schreuder.(eds.), 287–336.
Reimer, J.F., J.S. Brown and T.C. Lorsbach. 2001. “Orthographically mediated inhibition effects:
Evidence of activational feedback during visual word recognition”, Psychonomic Bulletin and
Review 8: 102–110.
Ricciardelli L.A. 1992. “Bilingualism and cognitive development in relation to threshold theory”,
Journal of Psycholinguistic Research 21, 4: 301–316.
Riehl, C.M. 2010. “The mental representation of bilingualism”, Wiley Interdisciplinary Reviews:
Cognitive Science 1/5: 750–758.
Riemer, C. 2002. “Wie lernt man Sprachen?”, in: J. Quetz and G. von der Handt. (eds.), 49–82.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 207
Ringbom, H. 1987. The Role of First Language in Foreign Language Learning. Clevedon:
Multilingual Matters.
Ringbom, H. 2001. “Lexical transfer in L3 production”, in: J. Cenoz, B. Hufeisen and U. Jessner
(eds.), 59–68.
Ringbom, H. 2007. Crosslinguistic Similarity in Foreign Language Learning. Clevedon:
Multilingual Matters.
Robinson, P. (ed.). 2012. Routledge Encyclopedia of Second Language Acquisition. London:
Routledge.
Roelofs, A. 1992. “A spreading activation theory of lemma retrieval in speaking”, Cognition 42:
107–142.
Roelofs, A. 1999. “Phonological segments and features as planning units in speech production”,
Language and cognitive processes 14: 173–200.
Romaine, S. 1995. Bilingualism. Second Edition. Oxford: Blackwell.
Rosch, E. 1975. “Cognitive representations of semantic categories”, Journal of Experimental
Psychology: General 104: 192–233.
Rosch, E. and C.B. Mervis. 1975. “Family resemblances: Studies in the internal structure of cat-
egories”, Cognitive Psychology 7: 442–460.
Rosenman, M. A. and F. Sudweeks. 1995. “Categorization and prototypes in design”, in: P.
Slezak, T. Caelli and R. Clark. (eds.), 189–212.
Rossell S.L., C.J. Price and A.C. Nobre. 2003.“The anatomy and time course of semantic prim-
ing investigated by fMRI and ERPs”, Neuropsychologia 41: 550–564.
Roux, P.W. 2013. “Words in the Mind: Exploring the relationship between word association and
lexical development”, Polyglossia 24: 80–91.
Rubin, G. S., C.A. Becker and R.H. Freeman. 1979. “Morphological structure and its effect on
word recognition”, Journal of Verbal Learning and Verbal Behaviour 18: 757–767.
Rumelhart, D. and McClelland, J. 1986. Parallel Distributed Processing. Cambridge, MA: MIT
Press.
Rumelhart, D.E. and J.L. McClelland. 1982. “An interactive activation model of context effects
in letter perception: II. The contextual enhancement effect and some tests and extensions of
the model”, Psychological Review 89: 60–94.
Safont Jordà, M.P. 2005. Third Language Learners: Pragmatic Production and Awareness.
Clevedon: Multilingual Matters.
Sánchez Casas, R.M., C.W. Davis and J.E. García Albea. 1992. “Bilingual lexical processing:
Exploring the cognate/noncognate distinction”, European Journal of Cognitive Psychology 4:
293–310.
Sánchez–Casas, R. and J.E. García–Albea. 2005. “The representation of cognate and noncognate
words in bilingual memory: Can cognate status be characterized as a special kind of morpho-
logical relation?”, in: J.F. Kroll and A.M.B. De Groot (eds.), 226–250.
Sandra, D. and M. Taft (eds.). 1994. Morphological Structure, Lexical Representation and
Lexical Access. Hove: Lawrence Erlbaum.
Schilling H.H., K. Rayner and J.I. Chumbley. 1998. “Comparing naming, lexical decision, and
eye fixation times: word frequency effects and individual differen-ces”, Mem Cognition
26/6:1270–81.
Schreuder, R. and B. Weltens (eds.). 1993. The Bilingual Lexicon. Amsterdam/Philadelphia: John
Benjamins.
Schreuder, R. and G.B. Flores d’Arcais. 1989. “Psycholinguistic issues in the lexical representa-
tion of meaning”, in: W.D. Marslen–Wilson (ed.), 409–436.
Schwartz, A.I. and J.F. Kroll. 2006. “Bilingual lexical activation in sentence context”, Journal of
Memory and Language 55: 197–212.
Schwartz, A.I., J.F. Kroll and M. Diaz. 2007. “Reading words in Spanish and English: Mapping
orthography to phonology in two languages”, Language and Cognitive Processes 22:
106–129.
Seashore, R. H. and L.D. Eckerson. 1940. “The measurement of individual differences in general
English vocabularies”, Journal of Educational Psychology 31: 14–38.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
208 References
Seidenberg, M. and J.L. McClelland. 1989. “A distributed developmental model of word recogni-
tion and naming”, Psychological Review 96: 523–568.
Seidlhofer, B. 2004. “Research Perspectives on Teaching English as a Lingua Franca”, Annual
Review of Applied Linguistics 24: 209–239.
Seidlhofer, B. 2011. Understanding English as a Lingua Franca. Oxford:Oxford University
Press.
Sharwood Smith, M.A. 1994. Second Language Acquisition: Theoretical Foundations. London:
Longman.
Sherkina, M. 2003. “The cognate facilitation effect in bilingual speech processing”, Toronto
Working Papers in Linguistics 21: 135–151.
Sherkina–Lieber, M. 2004. “The cognate facilitation effect in bilingual speech processing: The
case of Russian–English bilingualism”, Cahiers linguistiques d'Ottawa 32: 108–121.
Siek–Piskozub T., A. Wach and A. Raulinajtys. 2008. “Research on foreign language teaching in
Poland 2000–2006”, Language Teaching 41/1: 51–86.
Singleton, D. 1999. Exploring the Second Language Mental Lexicon. Cambridge: Cambridge
University Press.
Singleton, D. 2000. Language and the Lexicon. London: Edward Arnold.
Singleton, D. 2002. “Cross-linguistic interactions in the multilingual mental lexicon”, in
Proceeding of the Second International Conference on Third Language Acquisition
(CD-ROM) ed. Hooghiemstra,M., Leeuwarden, Fryske Akademy,
Singleton, D. 2003. “Perspectives on the multilingual lexicon: a critical synthesis”, in: J. Cenoz,
B. Hufeisen and U. Jessner (eds.), 167–176.
Singleton, D., J. Fishman, L. Aronin, L. and M. O Laoire. 2013. Current multilingualism: a new
linguistic dispensation. Berlin, Mouton de Gruyter.
Skutnabb–Kangas, T. 1984. Bilingualism or Not. Clevedon: Multilingual Matters.
Slezak, P., T. Caelli and R. Clark. (eds.). 1995. Perspectives on cognitive science: Theories,
experiments and foundations. Norwood, NJ: Ablex.
Smith, E.E., E.J. Shoben and L.J. Rips. 1974. “Structure and process in semantic memory: A fea-
tural model for semantic decisions”, Psychological Review 81: 214–241.
Smith, J.D. and J.P. Minda, 2002. “Distinguishing prototype-based and exemplar-based pro-
cesses in`` category learning”, Journal of Experimental Psychology: Learning, Memory, &
Cognition 28: 800–811.
Smith, M.C. 1997. “How do bilinguals access lexical information?”, in: A.M.B. De Groot and
J.F. Kroll (eds.), 145–168.
Smith, M.C. and K. Kirsner. 1982. “Language and orthography as irrelevant features in colour–
word and picture–word Stoop interference”, Quarterly Journal of Experimental Psychology
34: 153–170.
Smith, N.R. and M.B. Franklin (eds.). 1979. Symbolic functioning in childhood. Hillsdale: N.J.:
Erlbaum.
Smits, E., H. Martensen, A. Dijkstra and D. Sandra. 2006. “Naming interlingual homographs:
Variable competition and the role of the decision system”, Bilingualism: Language and
Cognition 9: 281–297.
Soares, C. and F. Grosjean. 1984. “Bilinguals in a monolingual and a bilingual speech mode: The
effect on lexical access”, Memory and Cognition 12/4: 380–386.
Solomyak, O. and A. Marantz. 2009. “Lexical access in early stages of visual word processing: A
single–trial correlational MEG study of heteronym recognition”, Brain and Language 108/3:
191–196.
Spöttl, C. 2001. “Expanding students` mental lexicons. A multilingual student perspective”, in: J.
Cenoz, B. Hufeisen and U. Jessner. (eds.). 2001b, 161–178.
Stanners, R.F., J.J. Neiser, W.P. Hernon and R. Hall. 1979. “Memory representation for morpho-
logically related words”, Journal of Verbal Learning and Verbal Behavior 18: 399–412.
Stavans, A. and M. Muchnik. 2007. “Language production in trilingual children: insights on code
switching and code mixing”, Sociolinguistic Studies 1/3: 483–511.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 209
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
210 References
Tzeglov, J., R. Henik, R. Sneg and O. Baruch. 1996. “Unintentional word reading via the pho-
nological route: The Stroop effect with cross–script homophones”, Journal of Experimental
Psychology; Learning, Memory, and Cognition 22: 336–349.
Ullman, M.T. 2001. “The neural basis of lexicon and grammar in first and second language: the
declarative/procedural model”, Bilingualism: Language and Cognition 4: 105–122.
Ungerer, F. and H. Schmid. 1996. An Introduction to Cognitive Linguistics. London: Longman.
Valencia, J.F. and J. Cenoz. 1992. “The role of bilingualism in foreign language acquisition”,
Journal of Multilingual and Multicultural Development 13/5: 433–449.
Van Assche E, W. Duyck and R.J. Hartsuiker. 2012. “Bilingual word recognition in a sentence
context”, Front. Psychology 3:174. doi: 10.3389/fpsyg.2012.00174.
Van Assche, E., D. Drieghe, W. Duyck, M. Welvaert and R.J. Hartsuiker. 2011. “The influence
of semantic constraints on bilingual word recognition during sentence reading”, Journal of
Memory and Language 64: 88–107.
Van Assche, E., W. Duyck, R.J. Hartsuiker and K. Diependaele. 2009. “Does bilingualism change
native-language reading? Cognate effects in sentence context”, Psychological Science 20:
923–927.
Van Assche, E., W. Duyck, W. and R.J. Hartsuiker. 2013. Phonological Recoding in Error detec-
tion: A Cross-sectional Study in Beginning Readers of Dutch. PLoS ONE 8/12: 1–11.
Van Hell, J.G. and A.M.B. De Groot. 1998a. “Conceptual representation in bilingual memory:
Effects of concreteness and cognate status in word association”, Bilingualism: Language and
Cognition 1: 193–211.
Van Hell, J.G. and A.M.B. De Groot. 1998b. “Disentangling context availability and concreteness
in lexical decision and word translation”, Quarterly Journal of Experimental Psychology,
Section A: Human Experimental Psychology 51: 41–63.
Van Hell, J.G. and A.M.B. De Groot. 2008. “Sentence context modulates visual word recognition
and translation in bilinguals”, Acta Psychologica 128: 431–451.
Van Hell, J.G. and T. Dijkstra. 2002. “Foreign language knowledge can influence native language
performance in exclusively native contexts”, Psychonomic Bulletin and Review 9: 780–789.
Van Heuven, W.J.B. and A. Dijkstra. 2001. The Semantic, Orthographic, and Phonological
Interactive Activation Model. Poster presented at the 12th Conference of the European
Society for Cognitive Psychology, Edinburgh, Scotland.
Van Heuven, W.J.B. and T. Dijkstra. 2010. “Language processing in the bilingual brain: fMRI
and ERP support for the BIA + model”, Brain Research Reviews 64: 104–122.
Van Heuven, W.J.B., A. Dijkstra and J. Grainger. 1998. “Orthographic neighborhood effects in
bilingual word recognition”, Journal of Memory and Language 39: 458–483.
Van Heuven, W.J.B., H. Schriefers, T. Dijkstra and P. Hagoort. 2008. “Language Conflict in the
Bilingual Brain” Cereb. Cortex 18, 11: 2706–2716.
Van Heuven, W.J.B., K. Conklin, E.L. Coderre, T. Guo and T. Dijkstra. 2011. “The influence
of cross-language similarity on within- and between-language Stroop effects in trilinguals”,
Frontiers in Psychology 2: 1–15.
Van Hout, R., A. Hulk, F. Kuiken and R. Towell (eds.). 2003. The Interface between Syntax and
the Lexicon in Second Language Acquisition. Amsterdam: John Benjamins.
van Wijnendaele, I. and M. Brysbaert. 2002. “Visual word recognition in bilingu-als: phono-
logical priming from the second to the first language”, Journal of Experimental Psychology:
Human Perception and Performance 28/3: 616–27.
Vigliocco and Hartsuiker. 2005.“Maximal input and feedback in production and comprehension”,
in: A. Cutler. (ed.), 209–228.
Vitevitch M.S. 1997. “The neighborhood characteristics of malapropisms”, Lang Speech 40:
211–28.
Voga, M. and J. Grainger. 2007. “Cognate status and cross–script translation priming”, Memory
and Cognition,35: 938–952.
Voorwinde, S. 1981. “A lexical and grammatical study in Dutch–English–German trilingualism”,
ITL Review of Applied Linguistics 52: 3–30.
WWW.EBOOK777.COM
free ebooks ==> www.ebook777.com
References 211
Votaw, M.C. 1992. “A functional view of bilingual lexico-semantic organization”, in: R.J. Harris.
(ed.), . 299–321.
Wales, R.J. and E.W. Walker (eds.). 1976. New Approaches to Language Mechanisms.
Amsterdam: North–Holland.
Wandruszka, M. 1979. Die Mehrsprachigkeit des Menschen. München, Zürich: Piper.
Warren, P. 2013. Introducing Psycholinguistics. Cambridge: Cambridge University Press.
Wartenburger, I., H.R. Heekeren, J. Abutalebi, S.F. Cappa, A. Villringer. 2003. “Early setting of
grammatical processing in the bilingual brain”, Neuron 37: 159–170.
Wattendorf, E., B. Westermann, D. Zappatore, R. Franceschini, G. Lüdi, E-W Radü and C.
Nitsch. 2001. “Different languages activate different subfields in Broca's area”, Neuroimage
13: 624.
Weinreich, U. 1953. Language in Contact: Findings and Problems. New York: Linguistic Circle
of New York.
Weir, C. 1988. Communicative Language Testing with Special Reference to English as a Foreign
Language. Exeter: University of Exeter.
Weisgerber, L. 1966. “Vorurteile und Gefahren der Zweisprachigkeit”, Wirkendes Wort 16:
73–89.
Wełna, J. 1977. “Deceptive words: A study in the contrastive lexicon of Polish and English”,
Papers and studies in contrastive linguistics 7: 73–84.
Whitney, C. 2008. “Comparison of the SERIOL and SOLAR theories of letter-position encod-
ing”, Brain and Language 107: 170–178.
Whitney, C. 2011. “Location, location, location: how it affects the neighborhood (effect)”, Brain
and Language 118: 90–104.
Whitney, C. and P. Cornelissen. 2008. “SERIOL Reading”, Language and Cognitive Processes
23: 143–164.
Widła, H. 2005. “Sposób akwizycji kolejnych je˛zyków: Problemy metodologiczne i implikacje
dydaktyczne“, in: K. Karpińska–Szaj (ed.), 37–48.
Wierzbicka, A. 1985. Lexicography and Conceptual Analysis. Ann Arbor: Karoma.
Williams, S. and B. Hammarberg. 1998. “Language switches in L3 production: Implications for a
polyglot speaking model”, Applied Linguistics 19/3: 295–333.
Wittgenstein, L. 1953. Philosophical Investigations. Translated by G. E. M. Anscombe, Oxford:
Basil Blackwell.
Woutersen, M. 1997. The Organisation of the Bilectal Lexicon. Nijmegen: University of
Nijmegen.
Wunderlich D. (ed.). 2006. Advances in the theory of the lexicon. Mouton de Gruyter: Berlin.
Yip, V. and S. Matthews. 2007. “Relative Clauses in Cantonese-English Bilingual Children:
Typological Challenges and Processing Motivations”, Studies in Second Language
Acquisition 29: 277–300.
Yip, V. and S. Matthews. 2010. “The acquisition of Chinese in bilingual and multilingual con-
texts”, International Journal of Bilingualism 14, 1: 127–146.
Ytsma, J. 2000. “Trilingual primary education in Friesland”, in: J. Cenoz and U. Jessner (eds.),
222–236.
Yudes, C., Macizo, P. and Bajo, T. 2010. “Cognate effects in bilingual language comprehension
tasks”, Cognitive Neuroscience and Neuropsychology 21/7: 507–512.
WWW.EBOOK777.COM