You are on page 1of 21

380 FALL 2012

Is Thinking Aloud Reactive When


Writing in the Heritage Language?
Iñigo Yanguas
University of San Diego

Beatriz Lado
Lehman College of the City University of New York

Abstract: Critics argue that requiring subjects to verbalize their thoughts while com-
pleting certain language tasks increases the participants’ cognitive load and impairs their
final performance (e.g., Jourdenais, 2001). Despite the importance of this claim for lan-
guage instructors, few studies have produced contradicting evidence after an empirical
study (e.g., Goo, 2010).
This study investigates whether thinking aloud in either English or Spanish hindered
performance during a semi-guided writing task completed by 37 college students whose
heritage language (HL) is Spanish. This study also explores the validity of implementing
think-alouds to investigate HL writing, an area which requires more research. Results
indicate that thinking aloud while writing in the HL benefits fluency and accuracy. These
results support studies that found positive reactivity and those concluding that concurrent
verbalizations should be employed with caution.

Key words: Spanish heritage language learners, concurrent verbalizations, reactivity,


think-aloud protocols, writing tasks

Introduction
Understanding the cognitive processes that underlie second language (L2) develop-
ment is one of the key issues in the instructed second language acquisition (SLA)
field in the 21st century. A better insight into these processes would certainly have
positive consequences for this field. Theoretically, it could help explain differences
in rate and level of attainment among learners, which have been so evasive to schol-
ars. Pragmatically, instructors and curriculum developers could potentially use this
knowledge to improve L2 teaching and maximize students’ efforts to become pro-
ficient in another language.
Two main options are available to researchers interested in providing insight
into learners’ minds as they perform tasks in the L2: concurrent and retrospective
verbalization procedures (Ericsson & Simon, 1993). The former requires asking
learners to speak aloud their thoughts, which are recorded, while carrying out a

Iñigo Yanguas (PhD, Georgetown University) is Associate Professor of Spanish at


the University of San Diego, California.
Beatriz Lado (PhD, Georgetown University) is Assistant Professor of Applied Lin-
guistics at Lehman College of the City University of New York.
Foreign Language Annals, Vol. 45, Iss. 3, pp. 380–399. © 2012 by American Council on the Teaching of Foreign Languages.
DOI: 10.111/j.1944-9720.2012.01198.x
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 381

learning task. The latter involves asking (Bowles, 2010, p. 106). Among the variables
learners about their perceptions of their per- considered, Bowles stressed type of verbali-
formance after the task; stimulated recall is zation (cognitive vs. metacognitive), type of
widely used. A short time after the task, sub- task (e.g., reading, writing, grammar learn-
jects are exposed to the recording of their ing), and level of L2 proficiency, and she
task performance so that they can recall claimed that research on writing and reac-
their thought processes when prompted by tivity with advanced levels of L2 proficiency
the researcher (Gass & Mackey, 2000). is required.
There are advocates and opponents The present study, therefore, adds to
for both methods among SLA scholars, as the reactivity literature by investigating
both present limitations. On the one hand, heritage language (HL) learners’ writing,
concurrent verbal protocols have been an area that has gone mainly unexplored in
argued to be reactive; several researchers this strand of research. Although HL speak-
have contended that demanding subjects to ers usually have a language of preference for
think aloud while carrying out certain tasks particular domains (i.e., education, home,
places an additional burden on participants’ friends, etc.), they are considered bilinguals
cognitive load so that final performance with a high level of oral proficiency in both
is most certainly affected (Ellis, 2001; the language they learn at home and the
Jourdenais, 2001). On the other hand, language they learn in school. In the former,
retrospective verbalizations are obviously however, they usually display very low levels
“subject to confounds such as memory of literacy. For this reason, those HL speak-
decay” (Bowles, 2010, p. 11). It is therefore ers who take Spanish at the university level
possible to question the validity and accu- are often placed in courses for HL speak-
racy of the information gathered through ers designed to help them improve their
the use of this technique. formal command of the language. Hence,
The argument continues, and the field is this study expands on previous findings on
still debating the validity of concurrent pro- reactivity to determine whether thinking
tocols to investigate L2 learners’ cognitive aloud affects performance during a writing
processes. The issue is key to SLA, as critics task completed by Spanish HL speakers. By
of the use of think-aloud procedures have doing so, this study also explores the valid-
argued that this technique imposes a double ity of implementing think-alouds to investi-
burden on participants’ cognitive process- gate writing by HL speakers, an area where
ing capacity, which could affect their ability the number of studies has been very limited
to process language. Due to the important (e.g., Schwartz, 2003, 2005). Participants
consequences of this claim for SLA, a grow- in this study could use their language of
ing body of researchers has empirically choice to perform the think-alouds.
addressed the reactivity of verbal proto-
cols (Bowles, 2008; Bowles & Leow, 2005;
Goo, 2010; Leow & Morgan-Short, 2004; Review of the Literature
Rossomondo, 2007; Sachs & Polio, 2007; Think-aloud protocols have been widely
Sachs & Suh, 2007; Sanz, Lin, Lado, used to investigate cognitive processes in sev-
Bowden, & Stafford, 2009; Yoshida, 2008). eral fields since the 1980s (for a summary,
It is not possible to draw general conclu- see Bowles, 2010). It was not until the early
sions from these studies because they have 1990s, however, that the distinction between
produced contradicting evidence. Bowles concurrent and retrospective verbalizations
conducted a meta-analysis that included was made (Ericsson & Simon, 1993). These
most of these studies and concluded that researchers classified verbalizations as either
“the answer to the question of reactivity and concurrent or retrospective, categorizing
think-alouds is not a simple ‘yes’ or ‘no’ but the temporal frame in which verbal proto-
rather is dependent on a host of variables” cols are collected. The former requires that
382 FALL 2012

subjects verbalize their thoughts while they evidence. Some investigations have not
carry out the task, whereas the latter makes found any supportive evidence for reactiv-
participants recall their thoughts shortly ity. For example, Leow and Morgan-Short
after completing the task. In addition, these (2004) compared a control group with no
authors distinguished three different levels of verbalizations (–TA) to a nonmetalinguistic
verbalizations: Type 1—vocalization of verbal verbalization group (⫹TA) in 77 Spanish
contents, Type 2—verbalizations of nonver- language beginners. As dependent meas-
bal contents that are put into verbal form, and ures, this study used text comprehension,
Type 3—those protocols that have subjects intake, and written production of the for-
explaining their thought processes. In the mal imperative in Spanish, and participants
context of SLA research, Bowles and Leow were instructed to think aloud during the
(2005) and Bowles (2008) referred to ver- treatment and the tests in either their L1
balizations of thoughts per se as nonmetalin- or L2, as they preferred. Results showed no
guistic and verbalizations of explanations and significant differences between the groups
justifications as metalinguistic. More recently, on either text comprehension or postassess-
Bowles (2010) used the terms metacognitive ment task performance. No supportive evi-
and nonmetacognitive, because those terms dence for reactivity was found. In a similar
could be used to refer to both verbal and non- vein, Bowles and Leow (2005) compared
verbal tasks. The present study adheres to the a control group to two different verbaliza-
former terminology, which has traditionally tion groups, metalinguistic and nonmeta-
been used in SLA to refer to verbalizations linguistic, in 45 Spanish advanced learners.
produced during verbal tasks. As in Leow and Morgan-Short (2004),
As mentioned above, reactivity (either participants were asked to read a text with
positive, improving task performance or neg- enhanced forms of the target form, which
ative, deteriorating learners’ performance) was the Spanish pluperfect subjunctive.
seems to be one of the most controversial Assessment tools included one comprehen-
issues concerning the use of concurrent ver- sion test and two fill-in-the-blank produc-
bal elicitation techniques, and, as such, it tion tests with both old and new items.
has been investigated in studies across differ- Results showed no reactivity effect for the
ent fields in cognitive psychology (Bowles, nonmetalinguistic group in comprehen-
2010). In general, results of these studies, sion and production tasks, whereas a slight
most of which utilized nonverbal tasks, decrease was shown in comprehension for
have supported Ericsson and Simon’s claim the metalinguistic group; furthermore, no
that verbalizations that require participants difference was shown in production for the
to explain their thoughts have more poten- metalinguistic group when compared to
tial for reactivity. Bowles’s review, including either group. In addition, both verbaliza-
studies that appeared after Ericsson and tion groups took significantly more time to
Simon (1993), also supported Ericsson and complete the tasks when compared to the
Simon’s claim that the type of cognitive task control group.
conducted may influence the presence or Along the same lines, Yoshida (2008)
absence of reactivity. Overall, it seems that did not find any reactivity. This study inves-
studies with tasks that pose greater cogni- tigated reading and reactivity in 64 Japa-
tive demands are more susceptible to reac- nese ESL learners distributed in six groups
tivity when compared to those with less formed from a combination of ⫹/–TA and
cognitively demanding tasks. type of reading condition (control text, text
with questions embedded, and text with
instructions to prepare an outline). After
The Exploration of Reactivity in SLA reading the passage and completing the
As mentioned above, studies in this strand corresponding task, participants completed
of research have produced contradicting a written recall task to measure reading
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 383

comprehension. Participants in the think- one of six groups in a computer-assisted


aloud group produced metalinguistic ver- language learning (CALL) environment;
balizations as they were asked to explain the groups were formed from a combina-
their decisions in the language of their tion of type of think-aloud (metalinguistic,
choice (L1 or L2). The results showed nonmetalinguistic, and silent) and type of
that the ⫹TA group needed more time to feedback (⫹/– explicit). All groups com-
complete the task. However, no difference pleted a Pac-Man game designed to teach
between groups was found for performance. the correct use of the verb gustar (to like).
Among the studies that found sup- Participants in the think-aloud group could
portive evidence for reactivity, Sachs and verbalize in their L1 or L2, as desired, while
Polio (2007) is the only study that has completing the treatment and production
investigated reactivity with a writing task. tests. The results revealed reactivity in the
They found negative reactivity when inves- metalinguistic group for latency and for
tigating a writing task and different types production of items seen during the treat-
of feedback in 15 high-intermediate ESL ment. Bowles found no reactivity, however,
students. These authors reported on two for the nonmetalinguistic group in any of
studies conducted to investigate any pos- the measurements or for the metalinguistic
sible differences in the linguistic accuracy group in production of new items.
of writing revisions when different feed- Rossomondo (2007) investigated reac-
back was provided as well as how reports tivity in a study with 161 first-semester
of awareness were related to the changes Spanish learners. The study explored the
they made on those revisions. Think- role of temporal indicators (e.g., temporal
aloud procedures were utilized to investi- adverbs) in the incidental acquisition of
gate reactivity and to assess awareness and the Spanish future tense. Participants were
linguistic accuracy of revisions. Results instructed to read a text for comprehension
of the first study (a three-week repeated- and were later asked to complete compre-
measures design) showed that learners in hension, recognition, and production tests.
the –TA group produced significantly more The results revealed no difference between
accurate revisions than those in the ⫹TA groups (⫹/–TA) in the comprehension test.
group. However, the second study (a non- However, in the recognition and production
repeated-measures design) results showed tests, the ⫹TA outperformed the –TA group.
no reactivity. The authors concluded that More recently, Sanz et al. (2009)
“although [concurrent] verbal protocols reported on two CALL experiments carried
can shed light on learner-internal processes out to assess the impact of concurrent ver-
in relation to written feedback, they should balizations. In these studies, a think-aloud
be interpreted with care […] keep in mind group and a control group interacted with
that the requirement to verbalize concur- the computer to receive a Latin grammar
rently might affect their results” (p. 86). lesson, task-essential practice, and explicit
When interpreting these results, it should feedback. In a pre-post design, both groups
be borne in mind that, as opposed to the were assessed on accuracy and reaction
majority of the studies on reactivity where time for three different tests: aural interpre-
participants were asked to verbalize in their tation, grammatical judgment, and written
language of preference, Sachs and Polio production. These two studies only differed
asked their participants to verbalize in the in the treatment: The second study did not
L2. include a grammar lesson. Results showed
Research where reactivity was found that for the first study, verbalizations did
also includes Bowles (2008). In this study, not affect accuracy scores but considerably
Bowles found reactivity only for some slowed down performance on grammatical-
variables. She conducted a study with 194 ity judgments. For the second study, how-
first-semester Spanish learners assigned to ever, results showed that concurrent verbal
384 FALL 2012

protocols positively affected performance time and group (⫹/–TA). A main effect for
on all accuracy tests. These results have group was only found on text completion,
important implications for the field, given which indicated that both groups performed
that they seem to suggest that “reactiv- differently regardless of time.
ity does not only depend on the nature of These results can also be interpreted in
the treatment or on the tool to measure the the context of the meta-analysis conducted
effects but also on the nature of the depend- by Bowles (2010), in which all these stud-
ent variable” (Sanz et al., 2009, p. 2). These ies (except Goo, 2010) were included. In
authors concluded that concurrent verbali- this meta-analysis, the researcher concluded
zations should be used with caution. that the presence or absence of reactivity
Goo (2010) investigated think-alouds was related to the variables present in each
and working memory capacity in a study study. Factors such as type of verbal report
with 42 first-semester Spanish learn- (metacognitive vs. nonmetacognitive), type
ers completing a reading task and two L1 of task (reading, writing, and grammar
working memory (WM) tests (listening learning), or L2 proficiency were included
and operation span test). Participants were in the analyses and revealed contradicting
divided into two groups (⫹TA and –TA) results. Thinking aloud had a small, facilita-
and were instructed to read a text contain- tive effect on text comprehension, a small to
ing 20 instances of the target form (i.e., medium-sized facilitative effect on receptive
immediate future in Spanish in the first form learning, and a medium-sized effect on
person plural vamos a ⫹ infinitivo [we are measures of productive form learning, rang-
going to ⫹ infinitive]). Assessment tools ing from being detrimental to being facili-
included a reading comprehension test tative. Bowles (2010) concluded that more
and a fill-in-the-blank production test. The research is needed on reactivity in relation
results showed negative reactivity for read- to several factors, such as language of verbal
ing comprehension but not for production report, writing, and level of L2 learners. The
of the target form. In addition, this study present study contributes to filling some of
also revealed an interesting finding with these gaps by investigating reactivity in the
regard to the relationship between thinking writing of Spanish HL speakers, English/
aloud and WM capacity: Reactivity seemed Spanish bilinguals enrolled in a Spanish-for-
to appear among high WM capacity learn- native-speakers college course. HL speakers
ers during the production of new items. are considered advanced learners of their
Some studies, such as Sachs and Suh HL, although their needs are thought to dif-
(2007), have not been conclusive in their fer from those of the regular advanced L2
results. These authors explored reactivity in learner. Their needs typically include devel-
a study on recast effects in a computer-medi- oping a formal variety of the language as
ated communication (CMC) environment. well as improving their writing skills. Thus,
Thirty Korean EFL learners read a story in in addition to contributing to the reactivity
their native language and were later asked to issue with a population of high-proficiency
retell it in English. The think-aloud group learners, this study may be informative to
was asked to verbalize their thoughts in the the field of HL writing, where concurrent
language in which they felt most comfort- verbal protocols have rarely been used. The
able. The target form was the sequencing of next section summarizes the studies that
tenses in English. Assessment tasks included have investigated writing in HL speakers
a text completion multiple-choice test and an from different perspectives.
interactive computer-mediated test designed
to measure learners’ ability to produce the
target form. As mentioned by the research- Spanish HL Speakers’ Writing
ers, the results were not conclusive for reac- In broad terms, two types of studies abound
tivity due to the lack of interaction between in the field of HL writing: investigations that
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 385

explore HL speakers’ writing strategies (e.g., strengthening their overall grasp of the her-
Schwartz, 2003, 2005) and studies that itage language” (p. 10).
offer pedagogical models to improve this As noted by Jourdenais (2001), verbal
population’s writing skills (e.g., Chevalier, protocols have been employed extensively
2004; Martínez, 2007). Recently, attitudes in writing research to inform general mod-
toward and experiences of writing among els of composition (e.g., Hayes & Flower,
HL speakers outside the academic context 1980) and to investigate the cognitive pro-
have also been investigated (e.g., Callahan, cesses of L1 and L2 writers (e.g., Swain &
2010). Lapkin, 1995). The only two studies in the
Martínez (2007) claimed that inves- Spanish-as-HL literature that have included
tigations have recently revealed two dif- think-aloud protocols in their research
ferent tendencies in the study of writing designs did so to explore writing strategies.
strategies in Spanish as HL: backward and Schwartz’s (2003) case study, for example,
forward biliteracy. In the former, writing utilized different data-gathering techniques
follows the rhetorical traditions in the in order to investigate Spanish HL speakers’
dominant language, whereas in the lat- writing strategies. First, her three partici-
ter, HL writers transgress those barriers pants filled in the self-efficacy questionnaire
and create “their own transcultural paths that assessed their level of confidence in
of expression” (p. 31). In particular, his the four skills. Participants also made indi-
study focused on the variation between vidual appointments with the researcher to
overt and null subject pronouns in Spanish complete the writing assignment (a descrip-
in the context of those two different writ- tive writing task) in her office. Subjects
ing tendencies. After analyzing the writing were then instructed on how to think aloud
of two different types of text by 13 first- and were given the topic for the composi-
year HL speakers, Martínez concluded that tion essay that they would complete while
his data suggested that transfer of literacy thinking aloud. These concurrent verbali-
skills is “discursively situated in different zations were tape-recorded, and so were the
contexts of situation” (p. 40). In this fash- interviews that took place right after par-
ion, he challenged the notion that linguis- ticipants had finished their compositions.
tic and rhetoric transfer is unidimensional Schwartz claimed that the most interesting
and proposed a model of multidimensional findings concerning the think-alouds were
transfer. This model takes into account that all three participants thought mainly
and validates the multiple literacies that in English, followed similar writing strat-
students bring with them to the classroom egies (even if they were not equally suc-
in both the HL and the dominant language. cessful), and based their writing on what
Chevalier (2004) took the opposite they thought sounded right to their
stance and proposed a pedagogical model own Spanish-native ear. Schwartz (2003)
based on the view that literacy should be reported that the “student dealt with search-
developed by expanding HL speakers’ famil- ing for the ‘right’ way to express her ideas
iarity with genres of written discourse. The in Spanish so that the language she used
model has six writing modes grouped in satisfied her inner L1/HL model” (p. 247).
four stages: stage I (conversation), stage II Interestingly, she acknowledged that these
(description and narrative), stage III (evalu- verbal protocols had actually intervened in
ation and explanation), and stage IV (argu- all three participants’ writing processes, as
ment). In each one of these modes, writers revealed in the interview. In particular, she
focus on the processes, discourse types, and mentioned that all three participants’ writ-
linguistic resources needed in every genre. ing routines were disrupted by the think-
Chevalier concluded that this model “is aloud procedures.
designed to give students tools for express- In her follow-up study, Schwartz (2005)
ing themselves effectively in writing and explored differences and similarities
386 FALL 2012

between two different types of bilingual “assumptions cannot be made about writing
students according to Valdés’s (1997) well- proficiency based on perceived oral profi-
known definition. Valdés defined Type A ciency or on prior preparation when defined
speakers as having basic academic skills as attendance or participation in a course or
in Spanish and being fluent in a rural con- program” (Schwartz, 2005, p. 333).
tact variety of Spanish. Type B students do In summary, it appears that given the
not have any academic skills in Spanish clear imbalance between HL speakers’ oral
and are limited speakers of a rural contact and written abilities, the field has focused
variety of Spanish. Three of Schwartz’s par- on fostering these speakers’ literacy skills.
ticipants were considered Type B, whereas As seen above, the exploration of writing
the remaining two participants were clas- strategies has been a very fruitful line of
sified as Type A bilinguals. Schwartz not research in which think-alouds have been
only compared both groups of students but used as a tool to gain insight into the writ-
also analyzed the individual writing perfor- ers’ mental processes. However, the reactiv-
mance and writing strategies used by each ity of think-aloud procedures is still under
participant in three different writing phases debate, and more research is needed to
(pre-writing, composing, and editing). ascertain what tasks are amenable to con-
Schwartz assessed six linguistic measures: current verbal protocols. Bowles (2010)
number of words, number of T-units, num- put it clearly: “[T]he answer to the ques-
ber of words per T-unit (W/T), total num- tion is not one-size-fits-all” (p. 137). There
ber of clauses divided by total number of are many variables involved and different
T-units (C/T ratio), total number of errors contexts that make the use of think-alouds
divided by total number of T-units (E/T hard to compare across the board. For these
ratio), and number of error-free T-units. reasons, there is a need to increase the num-
The definition of a T-unit used in Schwartz’s ber of studies that address reactivity so that
study is a main clause plus any subordinate the field can have a better insight into the
clauses containing a subject and a finite use of concurrent verbal protocols with dif-
verb. ferent learning tasks. In particular, there is
Think-aloud protocols were used to a shortage of research that investigates the
assess writing strategies along with a strat- reactivity of concurrent verbal protocols
egy questionnaire. Several similarities were in writing tasks and with more proficient
found within and between the groups. All participants.
participants composed strategies (e.g., Thus, the main goal of this study was to
repeating words, phrases, or parts of a sen- investigate the effects of concurrent verbali-
tence) and planned and revised as they zations on the performance of Spanish HL
wrote rather than before or after writing a speakers in a semi-guided writing task. The
first version. As for differences, the most research question is as follows:
distinct one seemed to be the total depend- Are there any significant differences in
ence on translation as a composing strategy the writing performance of two different
of Type B participants. Type A students, think-aloud groups (⫹TA/–TA) on measures
however, did not use this strategy. Given of:
the reduced number of participants, no sta-
1. Fluency (W, W/T);
tistical analyses were carried out to explore
2. Accuracy (EFT);
possible differences between groups in the
3. Lexical complexity (LV)?
fluency, complexity, and accuracy measures.
Descriptive analyses, however, showed
very small differences between groups in Method
all measures. These results made Schwartz Participants
conclude that many factors affect the writ- Two intact sections of two classes (N ⫽ 24,
ing level of Spanish HL speakers and that 20) designed for Spanish HL speakers at
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 387

a major public research university on the the study took place, 20 and 17 respec-
Eastern Seaboard took part in this study. tively in each class.
These classes were called “Review of Oral
and Written Spanish for Native Speakers
Educated in the United States” and were Materials—Semi-Guided Writing Task
offered by the Spanish Department as part On this writing task, participants were
of a sequence of three courses for native instructed to create their own story based
speakers of Spanish. The first course in this on three comic strips (Appendix B shows
sequence (Spanish I for Native Speakers) the original instructions in Spanish). In
was designed for those students who spoke these instructions, they were asked to write
Spanish at home but who had never for- a story in the past based on the comic strips.
mally studied the language. This course pro- They had to narrate what they thought was
vided a review of oral and written Spanish going on in the story, including charac-
in the content area of Latinos in the United ter descriptions and dialogue. They were
States. The program was developed around also reminded that they should use con-
the topics present in the textbook assigned nectors and any other elements necessary
for the class: Nuevos Mundos (Roca, 1999). to improve the flow of the story. “Baldo”
The second course (Spanish II for Native comic strips (by Hector D. Cantu and
Speakers) was developed for those students Carlos Castellanos) were used for this
who had successfully completed the previ- activity; they feature Latino characters and
ous course focusing on complex grammati- themes. On the strips used in the present
cal structures through the reading of texts study (see Appendix B), we can see Baldo
and writing of compositions and essays in and his father interacting in different ways.
the content area of American Latinos/as. In This free-writing activity was used to
addition, Spanish HL learners could take practice and reinforce the formal features,
the third course in the sequence (Commu- both linguistic and textual, of narrative writ-
nication Skills for Native Speakers of Span- ing that were being discussed in class. To com-
ish), which exposed students to rhetorical plete it successfully, learners had to perform
devices and provided practice in formal at different levels. First, this task asked par-
public speaking in Spanish. ticipants to narrate a story in the past, which
The participants (see Appendix A) would require them to use both the pret-
were 20 female and 24 male students erit and the imperfect tenses of regular and
enrolled in two sections of the first course irregular verbs. Second, they were required to
in the sequence for native speakers (Span- include dialogues in which they would have
ish I for Native Speakers). Their origins to use the present tense of both regular and
were diverse; 11 participants were born irregular verbs and form questions. Finally,
outside the United States, but only six had learners had to compose a story with all
stayed in their home country long enough the necessary cohesive devices discussed in
to attend school. Only three participants class for this type of writing. Therefore, this
reported a parent or grandparent who was task posed a double burden on participants,
born in the United States, and all the rest because they were required to draw on their
reported that their parents and grandpar- explicit knowledge of certain linguistic and
ents were born in their country of origin. In textual structures in order to complete it suc-
all, more than 56% of the participants were cessfully. As claimed by Montrul (2010), HL
to some degree of Central American origin. learners are primarily naturalistic learners,
To be included in the study sample, partici- and as such, they are better than L2 learners
pants had to be enrolled in these classes. at processing the language aurally. They are
Extra credit or any other type of incentive worse, however, in written tasks that maxi-
was not offered for participating. A total of mize the use of metalinguistic and explicit
37 participants attended class on the days knowledge of the language.
388 FALL 2012

The level of difficulty that this task vated they were to carry out the task and
posed for the participants (i.e., first-year how they felt about it.
Spanish HL students with limited formal Specific instructions on how to think
training in Spanish) was a key argument in aloud then followed. Participants were told
selecting it for the present study, given that that they should verbalize their thoughts
the main goal was to explore think-aloud as they wrote their story (i.e., say out loud
reactivity. In other words, the task had to anything they would say to themselves
be difficult enough for these researchers to while they thought). As this is the main
assess the facilitative or negative effects of difference between linguistic and meta-
concurrent verbal protocols. linguistic concurrent verbal protocols, it
was emphasized that they should not try
to explain or justify their thoughts. The
Procedure decision about which of these types of ver-
Both groups of participants, those who had balizations to use was based on their ability
to think aloud while doing the task (⫹TA, to answer the research questions (Bowles,
N ⫽ 20) and those who did not (–TA, 2010); in the present case, linguistic proto-
N ⫽ 17), met in the language lab in the cols, as proven in the pilot study, provided
third week of November at their regular enough detail for the researchers to inves-
class time. Because these were two different tigate motivational fluctuation along the
sections of the same class, these two groups task.
met on consecutive days. Both the instruc- Finally, as a warmup for the ⫹TA
tor of the class and one of the researchers group, one of the researchers played part
were present at this time; participants were of a think-aloud protocol using a sam-
familiar with the researcher, given that this ple from the pilot study so that learners
study was part of a larger longitudinal pro- would be familiar with the type of verbali-
ject that investigated general and specific zation required. Following the SLA tradi-
motivation throughout the semester. tion, participants were allowed to think
On the day in question, participants aloud in their L1, their L2, or a combi-
in the ⫹TA group were instructed on how nation of the two, as they desired. When
to think aloud and what the process of this procedure was clear to participants,
recording their thoughts was going to be. they were provided with the instructions
Previously, they had agreed on the con- to complete the writing task and the task
sent form to having their thoughts digi- itself (see Appendix B). The researcher
tally recorded. This form explicitly stated went through the instructions with the
that they needed to speak aloud whatever learners, making sure they were clear. On
came to their mind while they were carry- the blank page they were given to write
ing out the task. Before starting the writ- on, participants were reminded to always
ing task, the researcher clearly explained think aloud. In addition, the researcher
the procedure to participants. As suggested reminded participants to think aloud
by Bowles (2010), these instructions reit- while circulating in the language lab. After
erated the reason participants were being 25 minutes, the researcher gathered all
asked to think aloud, and they also speci- the writing samples—this time limit was
fied how they should think aloud as well as established after the pilot study in which
a warm-up activity. Participants were thus all participants finished the task within
reminded that these verbal protocols were this time frame.
being recorded to analyze and assess task The same procedure took place in the
motivational processes during their perfor- language laboratory with the –TA group,
mance. The message was also conveyed in the only difference being the think-aloud
plain and simple language: The focus of the explanations, instructions, and warmup
investigation was on exploring how moti- that this group did not undergo.
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 389

Analysis and Scoring Procedures procedures reacted with task performance,


In order to find an answer to this study’s four one-way ANOVAs were performed on
only research question, a series of one-way the scores of every writing measure assessed
analysis of variance (ANOVA) was per- in this study (W, LV, W/T, EFT). The scores
formed on the scores of the four different for each one of these measures were entered
linguistic variables measured in the writ- as dependent variables on the ANOVAs
ing task (see below) for think-aloud groups conducted, whereas group (⫹TA/–TA) was
(⫹TA/–TA). entered as the independent variable.
Four linguistic variables were explored The results presented in Table 1 show
in this semi-guided writing task: number a significant difference between TA groups
of words (W), lexical variety (LV), number in one of the writing measures analyzed:
of words per T-unit (W/T), and error-free EFT (F(1,36) ⫽ 8.896, p ⫽ 0.005). In
T-units (EFT). W and LV (fluency and lexi- addition, a trend toward significance
cal complexity measures, respectively) were was found for LV (F(1,36) ⫽ 4.057, p ⫽
used by Kormos and Dörnyei (2004) in their 0.052). Following Cohen (1988), medium
investigation of oral argumentative tasks effect sizes were found (d ⬎ 0.50) for both
and served to compare results across tasks. analyses. The means for the TA groups
W is the only frequency fluency measure, (see Table 2) help further interpret these
according to Wolfe-Quintero, Inagaki, and results: The concurrent verbal protocols
Kim (1998), for which any significant dif- utilized were positively reactive for this
ference across groups has been found. The accuracy measure. In other words, those
Uber Index was utilized to determine LV. participants who thought aloud while
Jarvis (2002) proposed the use of this for- writing this task made fewer errors. In
mula to compensate for the type-token ratio addition, their lexicon tended to be more
problem (i.e., this ratio falls significantly as varied than that of participants who did
the speech sample increases). Kormos and not speak their thoughts. Finally, no sig-
Dörnyei (2004) used this formula, which is nificant difference was found between TA
an arithmetical transformation of the type- groups for W or W/T.
token ratio: Uber Index: (log tokens) 2 / (log
tokens – log types). Regarding W/T, Wolfe-
Quintero et al. (1998) claimed that it is one Discussion
of the best fluency measures of development This study was conducted to investigate the
in the L2 writing research field because it is effects of thinking aloud on the writing of
“consistently linear and significantly related HL speakers in Spanish-for-native-speak-
to program or school levels” (p. 119). Finally, ers college courses. This line of research
EFT is the accuracy measure for which most has been suggested in the most recent lit-
noteworthy/significant results have been erature on think-alouds, because clearly
shown, according to the meta-analysis car- there is a lack of studies that address the
ried out by Wolfe-Quintero et al. (1998). reactivity of concurrent verbalizations
While there are several definitions involving advanced learners and writ-
of the term T-unit in the field of L2 writ- ing tasks (Bowles, 2010). In this fashion,
ing, the present study operationalizes this the researchers set out to assess how non-
term as a minimal, independent, termina- metacognitive concurrent verbalizations
ble clause, which has all modifying phrases affect fluency (W, W/T), accuracy (EFT),
attached to it (Larsen-Freeman, 1991). and lexical complexity (LV) measures in
writing. Specifically, the research question
investigated possible group differences
Results (⫹/–TA) on these variables’ scores. The
In order to find an answer to the research results warrant different responses depend-
question and assess whether think-aloud ing on the variable under investigation.
390 FALL 2012

TABLE 1
One-Way ANOVA Analyses

ANOVA
Sum of Mean
Squares df Square F Sig.
W Between 7,880.910 1 7,880.910 2.692 .110
groups
Within 102,476.009 35 2,927.886
groups
Total 110,356.919 36
EFT Between 116.576 1 116.576 8.896 .005
groups
Within 458.668 35 13.105
groups
Total 575.243 36
W/T Between 8.669 1 8.669 .752 .392
groups
Within 403.701 35 11.534
groups
Total 412.370 36
LV Between 25,265.280 1 25,265.280 4.057 .052
groups
Within 217,946.978 35 6227.057
groups
Total 243,212.259 36

In other words, there were no significant has a small, detrimental effect on produc-
differences between verbalization groups tive form learning compared to a group that
on either fluency measure (W/WT). The writes silently” (p. 104). The contradictory
present results, however, do indeed indi- evidence provided by this study could be
cate significant differences between groups explained by looking at the type of design,
on the accuracy measure utilized (EFT). task demands, level and language (L1 and
The group of students that thought aloud L2) of the participants, and language of ver-
wrote significantly more EFTs. Finally, balization. Whereas Sachs and Polio (2007)
regarding the lexical complexity measure and Sachs and Suh (2007) investigated ver-
assessed here (LV), results showed a trend balizing in conjunction with writing and
toward significance that should be further feedback, this study did not include any
investigated. type of feedback; it was designed to meas-
These results run contrary to the SLA ure the effects of thinking aloud on HL
studies that included writing in their designs learners’ overall fluency, accuracy, and com-
(Sachs & Polio, 2007; Sachs & Suh, 2007). plexity in writing. Sachs and Polio asked
Bowles (2010) concluded in her meta- participants to verbalize while comparing
analysis that “thinking aloud while writing an original writing piece to a reformulated
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 391

TABLE 2
Descriptive Statistics for One-Way ANOVA

N Mean Std. Deviation


W +TA 20 184.05 56.43
–TA 17 154.76 51.21
Total 37 170.59 55.36

W/T +TA 20 11.65 3.50


–TA 17 12.62 3.25
Total 37 12.10 3.38

EFT +TA 20 11.15 3.68


–TA 17 7.58 3.53
Total 37 9.51 3.99

LV +TA 20 309.41 80.51


–TA 17 256.97 76.96
Total 37 285.32 82.19

version. In addition, participants had to Um, what to do here? Like they’re argu-
verbalize in their L2, which may have posed ing; then he’s looking for a car, pulls out
a greater cognitive burden on participants. a book about being a teacher. Let
On the contrary, participants in our study, me see … ¿dónde empezamos? … en Espa-
as in Sachs and Suh, were able to verbalize ñol … [pause] [appears to be thinking]
in either their L1 or their L2, and no nega-
tive reactivity was revealed. These results [OK, good, let’s start, let’s go! Alright,
appear to provide evidence to support the so we got Baldo and his dad … . What
argument that verbalizing in the language to do here? Like they’re arguing; then
of choice while writing does not hinder he’s looking for a car, pulls out a book
the overall writing process. Finally, partici- about being a teacher. Let me see …
pants in Sachs and Polio and Sachs and Suh Where should we start? In Spanish…]
were ESL learners, while participants in our In line with Bowles’s (2010) meta-analysis,
study, as previously mentioned, were Span- the present results appear to indicate that
ish HL learners. reactivity depends on the type of measure
The following excerpt shows how par- investigated. Verbalizing did not have an
ticipants in this study did indeed make effect on the number of words produced
use of both English and Spanish in their overall (W) or on the number of words pro-
verbalizations. For example, John’s verbal duced per T-unit (W/T), which supports
protocol (participants’ names have been studies where no reactivity was found for
changed to protect their identity) shows nonmetalinguistic verbalizations (Bowles,
how he spoke his mind in both languages 2008; Bowles & Leow, 2005; Goo, 2010;
as he wished. Leow & Morgan-Short, 2004; Sachs &
[unintelligible, reading the instruc- Suh, 2007). In other words, fluency was
tions] … OK, good, comenzamos, let’s go! not affected by current verbalizations. Nev-
Alright, so we got Baldo y su padre … ertheless, contrary to the research above,
392 FALL 2012

these results show reactivity for EFTs, which [Baldo was getting home from
seems to indicate that thinking aloud could school … “Wassup papa?” Baldo said
have made participants write more accurate to his dad. Baldo was getting home
T-units. In addition, marginal significance from school, and his dad was reading
between TA groups was found for LV. the paper. His dad answered, “Hi, son;
These findings support research that how are you?”]
found that verbalization enhances task
performance in both non-SLA (Ahlum- This excerpt shows Tomás’s think-aloud dur-
Heath & Di Vesta, 1986; Baumann, ing the first part of the composition. There
Seifert-Kessell, & Jones, 1992; Gagné & was no explicit mention on the part of the
Smith, 1962; Meissner, Brigham, & Kel- learner that might indicate that the process
ley, 2001; Williams & Davids, 1997) and of thinking aloud was helping him to write
SLA studies (Rossomondo, 2007; Sanz sentences or compose paragraphs. One can
et al., 2009). The study herein extends see, however, Tomás’s writing process at this
these results to include HL learners; that stage and how he seemed to struggle com-
is, verbalizing is not only beneficial for posing a longer unit such as a sentence. By
naïve (Sanz et al., 2009) or first-semester thinking aloud and repeating what he was
learners (Rossomondo, 2007), but also writing, this learner appeared to be able to
for bilinguals who grew up using two complete the sentence, not without effort. It
languages. In addition, the results herein could be argued that the process of think-
show that verbalizing might improve ing aloud itself seemed to have contributed
accuracy in L2 writing, thus expanding to the successful completion of the sentence
positive reactivity beyond text compre- and also to write more accurate sentences.
hension (Rossomondo, 2007) or language There were several other cases like the
processing during practice with feedback one shown above that implicitly pointed
(Sanz et al., 2009). to learners realizing their deficiencies in
There are two types of evidence in the composing the text, but the discussion now
transcriptions of the verbal procedures turns to the second type of evidence: those
that seem to provide valid arguments to cases in which participants clearly stated
support the above assumption—that the that there was some kind of problem with
think-alouds themselves made learners their writing structure. Andrés’s verbal pro-
write more accurate T-units. First, there tocol displayed such an example:
are implicit indications that participants Uhm el … hi jo le di … ce a su pa … dre
may have been made aware of the syntactic que no quiii … e re ir a la es … cueee …
structure of their writing by simply speak- la, que ya no quieee re iiiiir a la es cue
ing aloud. For instance, an excerpt from la … el pa dre cree … que el hi … jo no
Tomás’s think-aloud protocol could be used es … ta se…rio, no está hablando en serio,
to illustrate this point: en se … rio. El hi … joooo le di … ceee
que sí essss … taaa seee … riiiio. ¿Tiene
Baldo estaba ehhh llegando … es … sentido eso? creo que me falta algo aquí
cu … e … a … [reviews the last sentence] mmmm.
wassup pop le dijo Baldo a su papá, Baldo
estaba llegando de su escuela y su papá [Uh, the son tells his father that he
estaba … eh … estaba, estaba, estaba … doesn’t want to go to school, that he
eh … eh … leyendo un periódico, periód- doesn’t want to go anymore. The father
ico, pe … rio … di … co, y su papá estaba believes that … that the son … is not
leyendo un periódico, su … eh … eh papá serious, he is not being serious. The
le con … tes … to hola mijo mijo, hola son says that he is. The son says that he
mijo, hola mijo, co mo es tas? eheh huh is serious. Am I making sense? There’s
[pauses] [appears to be thinking] something missing here, I believe.]
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 393

Here was an explicit statement in which en … ton … ces entonces le dijo el papa
Andrés claimed that there was something que ya andaba, andaba bien [writing].
missing in his writing. There are other pos-
[Then they went to see the car and try
sible interpretations, but this process of
it, and then his dad said that it was
awareness could also be attributed to the
OK.]
verbal protocol itself because it was mak-
ing this learner say what he was writing out Besides reviewing for meaning, one could
loud. argue that these reviewing strategies helped
Along the same lines, Christine’s ver- participants in this group improve the qual-
bal protocol also contained an explicit ity of the sentences they wrote. Thinking
comment that signaled her perception of aloud would then help learners monitor
a problem in her writing. This participant their own writing processes and acquire
explicitly stated that part of her writing did helpful strategies, just as it helped com-
not sound right. It seems that what she had prehension monitoring in Baumann et al.
just said out loud made her aware of some (1992). However, to compare the two
structural problems with her writing: groups and validate this argument, one
would have to analyze the writing strategies
En el último comic strip … a ver … Baaal
of participants in the silent group, which
do en … tra la co … ci … na le … yen …
the present study did not set out to do.
do un li bro su pa … pa le pree gun … ta
Further investigations of HL learners’
es … tas leyendo, estás leyendo … bal do
writing should be carried out to confirm
le mu … es traaa el li … bro y di … ce
the above interpretations; nevertheless,
how to be a teacher. No, this doesn’t sound
they seem to be very plausible explanations
right … Let me see … .
for the quantitative results achieved in this
[In the last comic strip, let’s see, Baldo study.
enters the kitchen reading a book. His Combining all measures, the research-
dad asks him, “What are you reading?” ers can conclude that thinking aloud did not
Baldo shows him the book and says, make participants write more words overall
“How to be a teacher.” No, this doesn’t or per clause, but it made them write more
sound right … Let me see … .] accurate clauses than the ones written by par-
ticipants in the silent group. This interpreta-
A further argument to support the fact
tion of the results is in line with the claim
that participants in the think-aloud
of Sanz et al. (2009) that “requiring learn-
group wrote more accurate T-units was
ers to perform think-aloud protocols has
found in some of their writing strategies.
the potential to alter the very processes they
Eighty percent of the participants in this
are meant to reflect” (p. 63). As Sanz et al.
group generated and implemented writ-
also noted, an explanation for the enhanced
ing subtasks, such as rereading instruc-
performance can be found in Swain’s output
tions or reviewing their writing after they
hypothesis (Swain, 1985, 1995, 2000), given
had finished one comic strip. More than
that verbalizing allows the learner to get
80 percent of these participants reread
additional input, thus enhancing the oppor-
some of the sentences they had written
tunities to notice certain forms that might
at some point during the composition of
then become part of the learner’s internal
their story. The following is an example in
grammar (Schmidt, 1993). For HL learn-
Rico’s verbal protocol:
ers, verbalizing may help them test or notice
En … ton … ces fue … ron entonces certain expressions that sound native-like to
fueron a ver … fueron a ver … uhm … them (Schwartz, 2003).
el carro, el ca … rro y … uhm … y a The following excerpt from Anne’s
pro … pro … bar … lo y a probarlo … think-aloud protocol illustrates this last
[pauses] [reviews the last sentence] point well:
394 FALL 2012

Ser algo en la vida, en … la … vi … da, si study, most of these participants made edits
que … ría ser al … go en la vida te … nía at the surface level while they were writing
que … If he wanted to be someone, he had rather than making a final revision after they
to get a job and show some responsibility were finished writing. In addition, very few
… Que si quería ser algo en la vida tenía writers utilized any pre-writing strategies.
que … agarrar un trabajo … good idea, As far as writing strategies are concerned,
un trabajo, a … ga … rrar … un … tra … the researchers have to agree with Schwartz’s
ba … jo, uhm uhm … fueron para cami- (2005) conclusions:
nar … fue … ron … pa … ra … ca … mi
For these students, as for all of us,
… nar … y vieron un carro for sale … de
the most active phase of writing is the
venta? en venta? en venta … uhm uhm …
composing phase. They read their work
en venta, vie … ron … un ca … rro … en
critically for meaning while they write,
… mmm ven … ta … en … ven … ta?
constantly evaluating how their text
[Be someone in life, if he wanted to be sounds […] they have to resort to their
someone, he had to get a job and show own oral models to guide their writing.
some responsibility … Went walking, (p. 332)
and they saw a car for sale … On sale?
However, this does not imply that all Span-
For sale?]
ish HL speakers’ writing strategies should
Supporting Schwartz (2003), it is evident be considered equal. On the contrary, these
how Anne tests aloud two different ways of researchers argue, as have other scholars in
saying “on sale” (de venta and en venta) and the field (e.g., Martínez, 2007; Schwartz,
decides to use en venta, which is the most 2003, 2005), that many factors and vari-
common way of saying “for sale” in Span- ables affect the skill level of these learners
ish. It would seem that testing aloud her and that the approaches utilized to explore
options allowed Anne to come up with the them should be kept open to try to match
right choice, confirming that concurrent this variability. Think-aloud procedures are
verbal protocols might aid HL learners in one of the methods that can allow research-
noticing native-like expressions. ers to take such an open stance, and as
The main purpose of this study is to ana- such, they should be considered. Neverthe-
lyze think-aloud reactivity in HL speakers’ less, given that verbalizations may have an
writing, not their writing strategies. How- effect on task performance, it is advisable to
ever, some writing strategies were indeed include a control group.
revealed in the analysis of the verbalization
protocols. As mentioned above, 80 percent
of the participants in the ⫹TA group gener- Conclusions and Future Research
ated some writing subtasks, such as reread- This study has provided some evidence
ing instructions or summarizing the story to support the notion that thinking aloud
they were going to write before actually may have a positive effect on the writing
starting. An even larger percentage of learn- process of HL learners. In addition, it has
ers reread some of the sentences they had shown that verbalization does not play
written at some point during the composi- a role in overall writing fluency, but it
tion of their story while they were writing. seems to have an effect on some accuracy
Schwartz (2005) found these strategies to be measures. Furthermore, thinking aloud
common in the two population samples that while writing seems to help HL learners
she analyzed in her case study. She classified notice both their weaknesses and strengths,
strategies such as commenting on the draft; which leads to better monitoring of their
repeating words, phrases, or sentences; or own writing process and to self-correc-
rereading as composing strategies (versus tion. Finally, although verbalizing provides
pre-writing and editing strategies). As in her valuable information about the types of
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 395

strategies used by learners, these results Bowles, M. A. (2010). The think aloud contro-
suggest that caution is needed when imple- versy in second language research. New York:
menting think-alouds to investigate writ- Routledge.
ing. As suggested by previous researchers Bowles, M. A., & Leow, R. P. (2005). Reactiv-
(Sachs & Polio, 2007; Sanz et al., 2009), the ity and type of verbal report in SLA research
methodology: Expanding the scope of investi-
inclusion of a silent control group is recom- gation. Studies in Second Language Acquisition,
mended. 27, 415–440.
Further studies should include a larger
Callahan, L. (2010). U.S. Latinos’ use of writ-
sample to replicate these results. In addi- ten Spanish: Realities and aspirations. Herit-
tion, it would be advisable to include ran- age Language Journal, 7. Retrieved July 16,
domization of the groups in the research 2012, from http://www.heritagelanguages.org
design. Moreover, it would be interesting to Chevalier, J. F. (2004). Heritage language lit-
include different populations in an attempt eracy: Theory and practice. Heritage Language
to find out if the results for HL learn- Journal, 2. Retrieved July 16, 2012, from
ers are observed, for example, in non-HL http://www.heritagelanguages.org
high-proficient L2 learners. The pedagogi- Cohen, J. (1988). Statistical power analysis for
cal implications of finding a positive effect the behavioral sciences (2nd ed.). Hillsdale, NJ:
of verbalizing on the writing process of Lawrence Erlbaum Associates.
these two populations would be obvious Ellis, R. (2001). Introduction: Investigating
and should be taken into account in any form-focused instruction. Language Learning,
51(Suppl. 1), 1–46.
writing class. Future research should also
look into the effect of HL verbalizations on Ericsson, K. A., & Simon, H. A. (1993). Proto-
writing with feedback, as in Sachs and Polio col analysis: Verbal reports as data. Cambridge,
MA: MIT Press.
(2007), in order to investigate if the positive
evidence provided by the think-alouds on a Gagné, R. H., & Smith, E. C. (1962). A study
of the effects of verbalization on problem-
final writing product is maintained when solving. Journal of Experimental Psychology,
different revisions come into play. Finally, 63, 12–18.
it could be argued that the demands of the
Gass, S., & Mackey, A. (2000). Stimulated
task in the present study (a semi-open and recall methodology in second language research.
relatively untimed writing task) allowed Mahwah, NJ: Lawrence Erlbaum Associates.
participants to reflect on their writing pro- Goo, J. (2010). Working memory and reactiv-
cess more and to use their verbalizations as ity. Language Learning, 43, 245–254.
a successful strategy. It remains to be seen
Hayes, J. R., & Flower, L. S. (1980). Writing as
whether these same results will appear in problem solving. Visible Language, 14, 388–399.
more controlled or more demanding tasks.
Jarvis, S. (2002). Short texts, best-fitting
curves, and new measures of lexical diversity.
Language Testing, 19, 57–84.
References
Jourdenais, R. (2001). Cognition, instruction,
Ahlum-Heath, M. E., & Di Vesta, F. J. (1986). and protocol analysis. In P. Robinson (Ed.),
The effect of conscious controlled verbalization Cognition and second language instruction (pp.
of a cognitive strategy on transfer in problem- 354–375). Cambridge, MA: Cambridge Uni-
solving. Memory and Cognition, 14, 281–285. versity Press.
Baumann, J. F., Seifert-Kessell, N., & Jones, L. Kormos, J., & Dörnyei, Z. (2004). The interac-
A. (1992). Effect of think-aloud instruction tion of linguistic and motivational variables in
on elementary students’ comprehension mon- second language task performance. Zeitschrift
itoring abilities. Journal of Reading Behavior, für Interkulturellen Fremdsprachenunter-
24, 143–172. richt. Retrieved April 10, 2011, from http://
Bowles, M. A. (2008). Task type and reactiv- zif.spz.tu-darmstadt.de/jg-09-2/beitrag/
ity of verbal reports in SLA: A first look at a kormos2.htm
L2 task other than reading. Studies in Second Larsen-Freeman, D. (1991). Research on lan-
Language Acquisition, 30, 359–387. guage teaching methodologies: A review of
396 FALL 2012

the past and an agenda for the future. In K. gies used by students in SNS courses. In L. A.
de Bot, R. Ginsberg, & C. Kramsch (Eds.), Ortiz López & M. Lacorte (Eds.), Contactos y
Foreign language research in cross-cultural contextos: El Español en los Estados Unidos y
perspective (pp. 119–132). Amsterdam: John en contacto con otras lenguas (pp. 323–334).
Benjamins. Madrid/Frankfurt: Iberoamericana/Vervuert.
Leow, R. P., & Morgan-Short, K. (2004). To Swain, M. (1985). Communicative compe-
think aloud or not to think aloud: The issue tence: Some roles of comprehensible input
of reactivity in SLA research methodology. and comprehensible output in its develop-
Studies in Second Language Acquisition, 26, ment. In S. Gass & C. Madden (Eds.), Input
35–57. in second language acquisition (pp. 235–253).
Rowley, MA: Newbury House.
Martínez, G. A. (2007). Writing back and
forth: The interplay of form and situation Swain, M. (1995). Three functions of output
in heritage language composition. Language in second language learning. In G. Cook &
Teaching Research, 11, 31–41. B. Seidelhofer (Eds.), Principle and practice
in applied linguistics: Studies in honor of H.G.
Meissner, C. A., Brigham, J. C., & Kelley, C.
Widdowson (pp. 125–144). Oxford: Oxford
M. (2001). The influence of retrieval pro-
University Press.
cesses in verbal overshadowing. Memory &
Cognition, 29, 176–186. Swain, M. (2000). The output hypothesis and
beyond: Mediating acquisition through collab-
Montrul, S. (2010). Current issues in herit-
orative dialogue. In J. P. Lantolf (Ed.), Sociocul-
age language acquisition. Annual Review of
tural theory and second language learning (pp.
Applied Linguistics, 30, 3–23.
97–114). Oxford: Oxford University Press.
Roca, A. (1999). Nuevos mundos. New York,
Swain, M., & Lapkin, S. (1995). Problems in
NY: John Wiley.
output and the cognitive processes they gen-
Rossomondo, A. E. (2007). The role of lexical erate: A step towards second language learn-
temporal indicators and text interaction for- ing. Applied Linguistics, 16, 371–391.
mat in the incidental acquisition of the Span-
Valdés, G. (1997). The teaching of Spanish
ish future tense. Studies in Second Language
to bilingual Spanish-speaking students: Out-
Acquisition, 29, 39–66.
standing issues and unanswered questions.
Sachs, R., & Polio, C. (2007). Learners’ uses In M. C. Colombi & F. X. Alarcon (Eds.),
of two types of written feedback on a L2 writ- La enseñanza del español a hispanohablantes.
ing revision task. Studies in Second Language Praxis y teoría (pp. 8–44). Boston: Houghton
Acquisition, 29, 67–100. Mifflin.
Sachs, R., & Suh, B. R. (2007). Textually Williams, A. M., & Davids, K. (1997). Assess-
enhanced recasts, learner awareness, and L2 ing cue usage in performance contexts: A
outcomes in synchronous computer-mediated comparison between eye-movement and
interaction. In A. Mackey (Ed.), Conversa- concurrent verbal report methods. Behavior
tional interaction in second language acquisition: Research Method, Instruments, & Computers,
A collection of empirical studies (pp. 197–227). 29, 364–375.
Oxford: Oxford University Press.
Wolfe-Quintero, K., Inagaki, S., & Kim, H. Y.
Sanz, C., Lin, H. J., Lado, B., Bowden, H. W., & (1998). Second language development in writ-
Stafford, C. A. (2009). Concurrent verbaliza- ing: Measures of fluency, accuracy and com-
tions, pedagogical conditions, and reactivity: Two plexity. Technical report 17. Manoa, Hawai’i:
CALL studies. Language Learning, 59, 33–71. University of Hawai’i Press.
Schmidt, R. (1993). Awareness and second Yoshida, M. (2008). Think-aloud protocols
language acquisition. Annual Review of Applied and type of reading task: The issue of reac-
Linguistics, 13, 206–226. tivity in L2 reading research. In M. Bowles,
R. Foote, S. Perpiñán, & R. Bhatt (Eds.),
Schwartz, A. M. (2003). Heritage Spanish
Selected proceedings of the 2007 Second
speakers’ writing strategies. In A. Roca & C.
Language Research Forum (pp. 199–209).
Colombi (Eds.), Mi lengua: Spanish as a her-
Somerville, MA: Cascadilla Proceedings
itage language in the United States: Research
Project.
and practice (pp. 235–257). Washington, DC:
Georgetown University Press.
Received April 20, 2011
Schwartz, A. M. (2005). Exploring differ-
ences and similarities in the writing strate- Accepted April 3, 2012
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 397

APPENDIX A

Participants

⫹TA Group
Family Born/schooling Language Spanish at home Spanish with
from abroad pref. friends
1 El Salvador USA/no Both Very often Often
2 Dominican DR/5 years English Often Never
Republic
3 Peru/ Venezuela/8 Both Very often Often
Argentina years
4 El Salvador El Salvador/no English Very rarely Very rarely
5 El Salvador/ USA/no English Always Very often
Ethiopia
6 Uruguay Israel/no English Often Very rarely
7 Guatemala/ USA/no Both Very often Very often
Peru
8 El Salvador El Salvador/yes Both Very often Very often
9 Honduras/ USA/no Both Often Very rarely
USA
10 Puerto Rico/ USA/no English Rarely Always
USA
11 Colombia Colombia/no English Rarely Never
12 Bolivia/El USA/no English Often Often
Salvador
13 Chile/ USA/no Both Very often Often
Jamaica
14 Romania Romania/11 Both Often Often
years
15 Ecuador USA/no Both Very often Rarely
16 Colombia/ USA/no English Often Rarely
Peru
17 Dominican DR/no English/Both Often Never
Republic
18 Costa Rica/ USA/no Both Always Often
Nicaragua
19 Puerto Rico USA/no English Very rarely Never
20 El Salvador USA/no English Rarely Never
398 FALL 2012

–TA Group
Family from Born/schooling Language Spanish at Spanish with
abroad pref. home friends
1 Puerto Rico USA/no Both Often Rarely
2 Ecuador USA/no English Always Very rarely
3 El Salvador El Salvador/yes Both Always Very often
4 Mexico USA/no Both Often Rarely
5 Puerto Rico USA/no Both Always Often
6 Guatemala USA/no English Always Often
7 Honduras Honduras/no English Always Very often
8 Puerto Rico/ USA/no English Rarely Never
Cuba
9 El Salvador USA/no Both Often Rarely
10 Costa Rica/ USA/no Both Always Often
Colombia
11 El Salvador USA/no Both Always Very rarely
12 Dominican USA/no English Often Rarely
Republic
13 Puerto Rico USA/no English Often Very rarely
14 El Salvador USA/no English Often Very rarely
15 El Salvador USA/no Both Always Often
16 El Salvador/ USA/no Both Always Often
Honduras
17 El Salvador USA/no Both Always Often
FOREIGN LANGUAGE ANNALS · VOL. 45, NO. 3 399

APPENDIX B

Semi-Guided Writing Activity

BALDO © Baldo Partnership. Used by permission of Universal Uclick. All rights reserved.
Copyright of Foreign Language Annals is the property of Wiley-Blackwell and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written permission.
However, users may print, download, or email articles for individual use.

You might also like