 The effect of the method of data collection

on the resultant data

 The effect of the sample size and variability
on the resultant data.
 The issues of naturalness and
representativeness
 Collection and analysis procedures
 The value of cross language studies
 4 goals of child language research Bennett-
Kaster, 1988:
1. To confirm general linguistic principles
2. To discover principles of language development
3. To clarify the relationship of language to
developments in other areas, such as cognition.
4. To provide a more or less theoretical description
of language development.
 Purpose and the researcher’s theoretical
predisposition will influence the data
collection procedure used.
 The researcher’s behavioural, linguistic,
cognitive, or eclectic theories will influence
specific language features studied and overall
study design.
 Collecting and analyzing child language data
is very complex.

 Several decisions must be made prior to data

collection.

 Methods and procedures used can influence

the resultant data and may unintentionally
color the conclusions drawn from these data.
 Language development data are usually
collected in 2 ways:
1. Spontaneous conversational sampling/natural
observation
2. Structured testing/experimental manipulation.
 Both methods raise issues of appropriateness
for the language feature being studied.

 Either one alone may be insufficient to

describe the child’s linguistic competence.

 Data yielded in one context may not appear

in another.
E.g. Haas & Owens, 1985:
In a study of pronouns, children produced a
wider variety in conversation and produced
more advanced forms in more formal testing.

Eisenberg, 1997:
Formal elicitation tasks, such as testing
procedures produce more advanced child
language than conversational sampling.
 Advantages:
 Control variables
 Various linguistic elements may be elicited using
verbal and nonverbal stimuli in a structured
presentation.
 Enable researchers to gather data that may not be
readily available using conversational or
observational techniques.
E.g. difficult to assess childrens’ comprehension or
metalinguistic skills without direct testing.
 Disadvantages:
 Control of context may result in a rather narrow
sampling.

 Experimental factors can have unintended

consequences e.g. a researcher may highlight an item
in a picture with an attempt to ensure the child’s
accurate comment. While the accuracy of the
comment is not always increased when one item is
marked, the amount of redundancy or inclusion of
irrelevant information does increase ( Lloyd &
Banham, 1997)
 Testing and experimental tasks do not necessarily
reflect the child’s performance in everyday use,
e.g. noncompliance with testing or experimental
procedures may not mean noncomprehension or
lack of knowledge, but may indicate a lack of
attention or interest.
 Advantages:
 More naturalistic
 Ensures analysis of real-life behaviours
 Disadvantages:
 Data collected may be affected by several
variables, such as amount of language, the
intelligibility of child, and the effect of the
context.

 Single conversational sample is inadequate to

demonstrate the full range of child’s
communication abilities.
 Sampling techniques exist along from a very
unstructured to, open-ended situations to more
structured.
 Any given naturalistic situation may be
insufficient for eliciting the child’s systematic
knowledge of language.
 No certainty that a given test situation will
represent the child’s naturally occurring
communication.
 Researcher concerned about 2 samples:
1. The sample or group of children from whom
data are collected.
2. Sample of language data.

 With both samples, a researcher is

concerned with the size and variability.
 If the sample is…
 Too small: restrict the conclusions that can be
drawn about all children
 Too large: unwieldy

 The 2 samples, subjects and language, also

interact, one influencing the other.
 Population sample:
 Must be large enough to allow for individual
differences and to enable group conclusions to be
drawn.
 Design of the study will influence the number of
subjects considered adequate.
e.g. Appropriate to follow a few children for a period
of time in a longitudinal study, but inappropriate to
administer a one-time only test to the same limited
number of children.
 The sample of children should accurately
reflect the diversity of the larger population
from which they were drawn.

 The sample should represent all

socioeconomic, racial and ethnic, and
dialectical variations found in the same
population and in the same proportion.
 Other variables that may be important
include:
 Size of family
 Birth order
 Presence of one or both parents in the home
 Amount of schooling.
 Children may also be excluded during the
course of the study either if they move away
or if they are uncooperative .

 With each exclusion, the sample group

becomes more restricted and thus less
representative.
 Language Sample
 Sample size depends on the purpose for which it
is collected.

 Adequate sample: usually at least 100 utterances.

 High reliability on measures such as number of

different words, MLU, and mean sentence length
in morphemes may require atleast 175 complete
and intelligible utterances (Gavin & Giles, 1996).
 Elements that occur less than once in 100
utterances may not occur within the typical
sample of that length.
 The amount of language collected will vary with
the language feature being studied.
 Pragmatic aspects of language which vary with
the context, may require the inclusion of several
contexts to provide an adequately varied
sample.
 Resources such as personnel, time and
money are always limited.
 Linguist must decide on appropriate sample
size and adequate level of analysis.
 In general:
 The larger the sample of children and/or speech,
the fewer data it is possible to analyze.
 The more detailed the analysis, the fewer the
children or the smaller the amount of speech it is
possible to sample.
 Any sample should fulfill the twin
requirement of naturalness and
representativeness.

 The conversational sample will be more

natural if the participants are free to move
about and inhibited by the process of sample
collection.
 A representative sample should include as
many of the child’s everyday experiences as
possible.
 3 potential problems in obtaining natural and
representative data ( Wells, 1985):
1. Observer paradox
2. Child’s physical and emotional state at the time
the information is being collected
3. Context in which sample is collected.
 The presence or absence of the researcher and
the actual recording method.
 Several collection techniques exist:
 Diary accounts
 Checklists
 Parental reports
 Direct and taped observation.
 First 3 are less time consuming and considered
reliable and valid while remaining highly
representative (Marchman & Bates, 1994).
 Electronic means of collection seem essential for
microanalysis.
 Videotaping is better than audiotaping as it allows
the researcher to observe the nonlinguistic
elements.
 Written transcriptions:
▪ - least desireable method for microanalysis
1. Easy to miss short utterances
2. Nearly impossible to transcribe the language of both the
child and the conversational partner because of the large
number of utterances to within a short period of time.
 Since transcription offers many opportunities for
error, studies should ensure intra-transcriber
reliability.
 The use of more than one transcriber reduces the
possibility of errors if the transcribers compare their
transcriptions and resolve their differences in a
consistent manner.
 In general, the more defined the speech sampled, the
better the intelligibility; the greater the number of
transcribers, the larger the unit of comparison; the
more experienced the transcriber, the better the
chance of having an accurate transcript.
 MLU is the most common measure of
language growth although it’s value is
questionable.

 Numerical scores and measures such as MLU

are inadequate for describing language
development in detail.

 Other values include total number of words

or number of words per clause.
 Difficult to determine when a child or group of
children actually knows or has mastered a
language feature.
 Criteria for establishing that a child knows a
word or a feature have not been pre-established.
E.g. with word knowledge, the researcher must
have clear evidence that a child comprehends the
word. In contrast, production criteria would
probably be based on spontaneous use and
consistent semantic intent.
 Young children: the researcher would also note
consistent phonetic form and semantic intent.
 Mastery can be based on:
 children using a feature in 90% of the obligatory
locations
 90% of children using the feature consistently.
(these percentages vary with individual researchers).

 Some researchers consider the average age for

acquisition to be that at which 50% of children
use a language feature consistently.
Berman, 1986:
 Usually undertaken to investigate
universality, linguistic specificity, relative
difficulty, or acquisitional principles.

 Studies of universality attempt to determine

which aspects of language, such as nouns and
verbs appear in all languages.
 Studies of linguistic specificity attempt to
determine whether development is the result of
universal cognitive development or unique
linguistic knowledge.
 E.g. The development of spatial and temporal
terms seems to be based on cognitive temporal
knowledge as well as on specific linguistic forms
used to mark that knowledge:
 English: in for containment, on for support
 Spanish: en for both
 German: auf, an, and um for on
 Relative difficulty studies look for language
development differences that may be
explained by the ease or difficulty of learning
structures and forms in different languages.
e.g. the passive sentence form is very difficult
to master in English than the relatively easier
form in Egyptian Arabic, Turkish, Sesotho and
Zulu (Demuth, 1990; Perera, 1994).
 Studies that investigate acquisitional
principles try to find underlying language
learning strategies that children apply
regardless of the language being acquired.
 2 basic methods of collecting cross-linguistic
data:
1. Gather a range of studies completed in different
languages, although these studies may differ in their
aims and methods.
+ quicker as studies have already been completed.
- Difficult to draw conclusions from such a diverse
collection.
2. Use a similar design across subjects from different
language groups
+ yields much more definitive data, with fewer complicating variables.
- takes much more time and effort to organize, coordinate, and
collect