Language Sample Measures Research Articles

Relationships between narrative language samples and norm-referenced test scores in language assessments of school-age children.

Both narrative language samples and norm-referenced language tests can be important components of language assessment for school-age children. The present study explored the relationship between these 2 tools within a group of children referred for language assessment. The study is a retrospective analysis of clinical records from 73 school-age children. Participants had completed an oral narrative language sample and at least one norm-referenced language test. Correlations between microstructural language sample measures and norm-referenced test scores were compared for younger (6- to 8-year-old) and older (9- to 12-year-old) children. Contingency tables were constructed to compare the 2 types of tools, at 2 different cutpoints, in terms of which children were identified as having a language disorder. Correlations between narrative language sample measures and norm-referenced tests were stronger for the younger group than the older group. Within the younger group, the level of language assessed by each measure contributed to associations among measures. Contingency analyses revealed moderate overlap in the children identified by each tool, with agreement affected by the cutpoint used. Narrative language samples may complement norm-referenced tests well, but age combined with narrative task can be expected to influence the nature of the relationship.

Spanish and early English development in young dual language learners: A preliminary study

The objective of this study was to describe ongoing Spanish and early English development in young dual language learners (DLLs) in the context of standardized measures and language sample measures. Participants were eight DLLs (M age in months = 44.4) attending preschool. Data were collected at four time points over the school year. Child outcome measures were composite and raw scores on the Clinical Evaluation of Language Fundamentals Preschool-2 (Spanish and English), and language sample measures of lexical and grammatical development. Median standard scores in Spanish at school entry (Time 1) were not substantially different from scores in May (Time 4); however, an increase in verb use was observed. Children's English receptive, content, and structure standardized and raw scores in May were higher than at school entry. Children's performance was highly variable across measures and languages. Findings provide a starting point for future research on language development of young Spanish–English DLLs and helpful considerations for speech-language pathologists and educators who assess DLLs.

Stability of Measures From Children's Interviews: The Effects of Time, Sample Length, and Topic

The purpose of this study was to examine the reliability of, and sources of variability in, language measures from interviews collected from young school-age children. Two 10-min interviews were collected from 20 at-risk kindergarten children by an examiner using a standardized set of questions. Test-retest reliability coefficients were calculated for 8 language measures. Generalizability theory (G-theory) analyses were completed to document the variability introduced into the measures from the child, session, sample length, and topic. Significant and strong reliability correlation coefficients were observed for most of the language sample measures. The G-theory analyses revealed that most of the variance in the language measures was attributed to the child. Session, sample length, and topic accounted for negligible amounts of variance in most of the language measures. Measures from interviews were reliable across sessions, and the sample length and topic did not have a substantial impact on the reliability of the language measures. Implications regarding the clinical feasibility of language sample analysis for assessment and progress monitoring are discussed.

Development and Validation of the Spanish–English Language Proficiency Scale (SELPS)

This study examined the development and validation of a criterion-referenced Spanish-English Language Proficiency Scale (SELPS) that was designed to assess the oral language skills of sequential bilingual children ages 4-8. This article reports results for the English proficiency portion of the scale. The SELPS assesses syntactic complexity, grammatical accuracy, verbal fluency, and lexical diversity based on 2 story retell tasks. In Study 1, 40 children were given 2 story retell tasks to evaluate the reliability of parallel forms. In Study 2, 76 children participated in the validation of the scale against language sample measures and teacher ratings of language proficiency. Study 1 indicated no significant differences between the SELPS scores on the 2 stories. Study 2 indicated that the SELPS scores correlated significantly with their counterpart language sample measures. Correlations between the SELPS and teacher ratings were moderate. The 2 story retells elicited comparable SELPS scores, providing a valuable tool for test-retest conditions in the assessment of language proficiency. Correlations between the SELPS scores and external variables indicated that these measures assessed the same language skills. Results provided empirical evidence regarding the validity of inferences about language proficiency based on the SELPS score.

Assessing bilingual Chinese–English young children in Malaysia using language sample measures

One reason why specific language impairment (SLI) is grossly under-identified in Malaysia is the absence of locally- developed norm-referenced language assessment tools for its multilingual and multicultural population. Spontaneous language samples provide quantitative information for language assessment, and useful descriptive information on child language development in complex language and cultural environments. This research consisted of two studies and investigated the use of measures obtained from English conversational samples among bilingual Chinese–English Malaysian preschoolers. The research found that the language sample measures were sensitive to developmental changes in this population and could identify SLI. The first study examined the relationship between age and mean length of utterance (MLUw), lexical diversity (D), and the index of productive syntax (IPSyn) among 52 typically-developing (TD) children aged between 3;4–6;9. Analyses showed a significant linear relationship between age and D (r = .450), the IPsyn (r = .441), and MLUw (r = .318). The second study compared the same measures obtained from 10 children with SLI, aged between 3;8–5;11, and their age-matched controls. The children with SLI had significantly shorter MLUw and lower IPSyn scores than the TD children. These findings suggest that utterance length and syntax production can be potential clinical markers of SLI in Chinese–English Malaysian children.

Longitudinal stability in genetic effects on children's conversational language productivity.

The authors examined the longitudinal stability of genetic and environmental influences on children's productive language sample measures during the early school-age years. Twin study methodology with structural equation modeling was used to derive univariate estimates of additive genetic (A), shared environmental (C), and nonshared environmental (E) effects on language measures at each of 2 time points, based on 487 twins at the 1st-grade time point and 387 twins at the 2nd-grade time point. To address questions of stability over time, the authors used longitudinal latent factor analysis. Stability in the Conversational Language factor was accounted for almost entirely by shared genetic effects between 1st and 2nd grade, meaning no new genetic effects were observed at the 2nd time point. In contrast, nonshared environmental effects were entirely time point specific, meaning whatever nonshared environmental influences were operating at the first time point were not influencing individual variation in the language factor at the second time point. The discussion in this article centers on possible candidates for both genetic and nonshared environmental effects as well as implications for clinical practice and future research.

Language Sampling: Does the Length of the Transcript Matter?

Language sample analysis is considered by many to be the gold standard for documenting children's oral language skills. One limitation, however, is the time required for collection and transcription of language samples. The goal of this study was to determine if stable language sample measures could be generated using relatively short language samples. Measures were generated from children's conversational and narrative language samples after they were broken into three lengths (1-, 3-, and 7-min samples). The measures were compared to determine the stability of measures from the short samples (1 and 3 min) when compared to measures from the long sample (7 min). The measures were further analyzed to determine if differences across transcript cuts varied as a function of age group (2;8 [years;months]-5;11 vs. 6;0-13;3) or sampling context (conversation vs. narrative). Overall, the language sample measures were quite consistent across the transcript cuts. Measures of productivity, lexical diversity, and utterance length were the most reliable when short samples were used. Implications for the efficient use of language sample analysis in clinical protocols are discussed. A framework for eliciting reliable short samples is provided.

Language sample measures and language ability in Spanish-English bilingual kindergarteners

Measures of productivity and sentence organization are useful metrics for quantifying language development and language impairments in monolingual and bilingual children. It is not yet known what measures within and across languages are most informative when evaluating the language skills of bilingual children. The purpose of this study was to evaluate how measures of language productivity and organization in two languages converge with children's measured language abilities on the Bilingual English Spanish Assessment (BESA), a standardized measure of language ability. 170 kindergarten age children who produced narrative language samples in Spanish and in English based on a wordless picture book were included in the analysis. Samples were analyzed for number of utterances, number of different words, mean length of utterance, and percentage of grammatical utterances. The best predictors of language ability as measured by the BESA scores were English MLU, English grammaticality, and Spanish grammaticality. Results are discussed in relationship to the nature of the measures in each of the languages and in regard to their potential utility for identifying low language ability in bilingual. Learning outcomes: The reader will be able to identify (1) why narratives are useful for evaluating bilingual language ability, (2) which measures of productivity and sentence organization in narratives best capture language ability and development in bilingual children, (3) how a bilingual or composite language ability across two languages can be determined, and (4) what relationship exists between measures of productivity and sentence organization in each language with a bilingual language ability score.

Using Language Sample Databases

Over the past 50 years, language sample analysis (LSA) has evolved from a powerful research tool that is used to document children's linguistic development into a powerful clinical tool that is used to identify and describe the language skills of children with language impairment. The The Systematic Analysis of Language Transcripts (SALT; J. F. Miller & A. Iglesias, 2008) Software Project has developed several databases of language samples from more than 6,000 typical speakers. This article presents an overview of the SALT databases and then demonstrates the power of these databases in classifying children with language impairment. Conversational language samples were elicited from 244 children with language impairment who were between 3 and 13 years of age. Language production measures generated from these transcripts were compared to measures from 244 transcripts in the SALT conversational database. A series of discriminant function analyses were completed to document the sensitivity and specificity of the language sample measures. The language sample measures were effective in classifying children based on their language status, with correct identification of 78% of the children with language impairment and 85% of the children who were typically developing. The SALT databases provide a useful tool for the clinical management of children with language impairment.

Clinician–Child Interactions: Adjustments in Linguistic Complexity

This pilot study examined the extent and nature of associations in the linguistic complexity used by child and clinician within conversational interactions. Correlation analyses focused on semantic and morphosyntactic language sample measures from an experienced speech-language clinician and 29 children with language impairment. Positive associations emerged between a variety of child and clinician measures, even when the effect of child age was removed. The most robust effect related to clinician adjustments in both morphosyntactic complexity and vocabulary diversity associated with differences in children's developmental sentence scores. Within a conversational exchange, the clinician in this study made significant adjustments in her linguistic complexity that were due, at least in part, to the linguistic complexity used by the children with whom she was interacting. Associations were similar to adjustments reported in prior studies of parent and teacher interactions with children with differing language abilities. However, the extent to which these findings generalize to other clinicians needs to be examined. Results from the present study challenge clinicians to dedicate conscious thought toward how their linguistic input should be structured, taking into consideration both the goal of the interaction and each child's profile of linguistic strengths and weaknesses. Directions for future research are also provided.

Early lexical and syntactic development in Quebec French and English: implications for cross‐linguistic and bilingual assessment

Although a number of studies have been conducted on normal acquisition in French, systematic methods for analysis of French and normative group data have been lacking. To develop a systematic method for the analysis of language samples in Quebec French, and to provide preliminary normative data on early lexical and syntactic development in French with a comparison with English. Language samples were collected for groups of monolingual French- and English-speaking children (n=39, age range 21-47 months) with normal language development. Coding conventions for French were developed based on similar principles as English SALT conventions. However, due to structural differences between the languages, coding of inflectional morphology was considerably more complex in French than in English. The French procedure provided developmentally sensitive measures of lexical and syntactic development, including mean length of utterance in morphemes and in words, and number of different words, and should be an important addition to the assessment procedures available for French. Cross-linguistic similarities and differences were noted in the language sample measures. Although the same elicitation context was used in the English and the French language samples, and the analysis methods were designed to rest on similar principles across languages, systematic differences emerged such that the French-speaking children exhibited a higher mean length of utterance, but smaller vocabulary sizes. Differences were also noted in error patterns, with much lower error rates occurring in samples of the French-speaking children. The findings have important implications for language assessment involving cross-linguistic comparisons, such as occurs in the assessment of bilingual children, and in the matching of participants in cross-linguistic studies. Given differences in the mean length of utterance and vocabulary scores across the languages, the finding of the same mean length of utterance or vocabulary obtained in the two languages for a given bilingual child or for monolingual speakers of the two languages does not imply equivalent levels of language development in the two languages.

Language sampling for kindergarten children with and without SLI: mean length of utterance, IPSYN, and NDW

Language sample analysis measures have long been promoted as exhibiting greater ecological validity than formal testing in the assessment of language disorder in children. In practice, their use is often restricted to preschool children, owing to lack of normative information, as well as criticisms of the validity of commonly used measures for the language of older children. This study compared scores of kindergarten children (mean age 6 years) with and without specific language impairment (SLI) on three commonly used language sample analysis measures: mean length of utterance in morphemes (MLU-m), the index of productive syntax (IPSyn), and number of different words (NDWs). Mean scores of the children with SLI were significantly lower for all three measures, though not for all subtests of the IPSyn. A number of individual differences were observed; notably, several children with SLI scored as well as those without. The problems and promise of language sampling for children beyond the preschool years are discussed in light of these results. Learning outcomes: (1) readers will gain an understanding of strengths and weaknesses of language sample measures in assessing kindergarten children with language impairment. (2) The reader will become aware of the utility of MLU in differentiating between young school age children with and without language impairment.

Case Studies of Children Using Fast ForWord

We report five case studies in which the children were enrolled in Fast ForWord (FFW). The purpose of the case studies was twofold: (a) to obtain independent objective data with the commercially available computer-based language intervention program, FFW; and (b) to identify patterns of performance with FFW related to the children's pre-intervention language profiles. Five children (3 boys, 2 girls) between the ages of 5;10 and 9;2 (years; months) enrolled in a private school for children with language-learning difficulties participated. Outcome measures included both standard scores from general tests of language and measures taken from language-sample analyses. There were modest changes in standardized measures of language after experience with FFW for 3 of the 5 children using the criterion of 95 % confidence interval. There were no clinically significant changes in language sample measures. Several interesting patterns of response to FFW intervention are discussed.

Language Change Following Computer-Assisted Language Instruction With Fast ForWord or Laureate Learning Systems Software

This exploratory study was designed to evaluate functional language changes during and after treatment with language intervention software. Two children with language impairments received Fast ForWord (FFW; Scientific Learning Corporation, 1997), and two other children received a bundle of intervention programs published by Laureate Learning Systems (LLS). The children received intervention for 1 hour and 40 minutes per day for 20 days (4 weeks). Treatment was delivered according to a multiple-probe design in which one child was enrolled in Fast FFW immediately after a baseline phase. Another child remained in an extended baseline phase before beginning FFW. The design was replicated for the bundle of LLS programs. The children with extended baselines were identical twins. Progress was measured by gains on the Oral and Written Language scales (OWLS) and by visual and mathematical examination of trends for language sample measures. All four children made clinically significant gains (posttest scores outside the 95% confidence interval of the pretest scores) on the OWLS. Two children who received the LLS software and one child who received FFW software made clinically significant gains on mean length of utterance in morphemes (MLU), but only one child, who received treatment with the LLS software, had fewer grammatical errors after treatment. The three children with improved MLU also produced a higher proportion of utterances with mazes. Measures of language content (percent of response errors) and language use (percentage of assertive utterances) were not informative due to high variability and floor effects. The similarity of the treatment effects (especially in the case of the identical twins) was surprising since FFW and the bundle of LLS programs targeted different levels of language, used different types of auditory stimuli, and were designed to promote different kinds of learning.

The structure of verbal abilities in young and older adults.

Four language sample measures as well as measures of vocabulary, verbal fluency, and memory span were obtained from a sample of young adults and a sample of older adults. Factor analysis was used to analyze the structure of the vocabulary, fluency, and span measures for each age group. Then an "extension" analysis was performed by using structural modeling techniques to determine how the language sample measures were related to the other measures. The measure of grammatical complexity was associated with measures of working memory including reading span and digit span. Two measures, sentence length in words and a measure of lexical diversity, were associated with the vocabulary measures. The fourth measure, propositional density, was associated with the fluency measures as a measure of processing efficiency. The structure of verbal abilities in young and older adults is somewhat different, suggesting age differences in processing efficiency.

Parental perceptions of children's communicative development at stuttering onset.

There has been clinical speculation that parents of young stuttering children have expectations of their children's communication abilities that are not well-matched to the children's actual skills. We appraised the language abilities of 15 children close to the onset of stuttering symptoms and 15 age-, sex-, and SES-matched fluent children using an array of standardized tests and spontaneous language sample measures. Parents concurrently completed two parent-report measures of the children's communicative development. Results indicated generally depressed performance on all child speech and language measures by the children who stutter. Parent report was closely attuned to child performance for the stuttering children; parents of nonstuttering children were less accurate in their predictions of children's communicative performance. Implications for clinical advisement to parents of stuttering children are discussed.

Mean length of utterance and other language sample measures in early Icelandic

Adaptations of the widely used MLU measure have been developed in several languages. Such adaptations require numerous modifications, especially in languages that are highly inflected. This study involved the development of a systematic procedure for coding language samples from Icelandic toddlers. Results are reported in terms of mean length of utterance in morphemes (MLU), total vocabulary, total number of different words and type-token ratio (TTR). These measures are analogous to their English counterparts, though not directly comparable. The Icelandic MLU measure was found to be developmentally sensitive in the age range of the study, which included a cross- sectional sample of 36 children aged 15 to 36 months. MLU correlated more strongly with sentence complexity than did age. Consistent with studies in Dutch and Irish, MLU in morphemes was very highly correlated with MLU in words in normally developing children. This relationship remains to be tested in children with language impairments. A secondary goal of this study was to provide descriptive data on the early acquisition of inflectional morphology in Icelandic, derived from language sample analysis.