This study is the main project I carry out as part of my postdoctoral position at the Division of Speech Language Pathology, CLINTEC, Karolinska Institutet. We are currently in the pilot phase (Aim 1 below) and are collecting data on children with typical language development in Swedish grades 4-9 as well as children with developmental language disorder in grades 4-6 and children with ADHD in grades 6-8 (part of aim 2). The pilot data collection and analyses are carried out as part of one final thesis and one master’s thesis in Speech and Language Pathology, where I am the supervisor.

Project summary:
Language skills, including reading and writing, are crucial for students to succeed academically and socially. Throughout the school years, language demands increase. As students mature, increasingly complex content is expected to be expressed through more advanced vocabulary and grammar for a range of different purposes, including to narrate, to explain, or to argue. These discourse-level tasks pose a challenge to students with developmental language disorder (DLD), which may be difficult to capture with existing standardized language tests. Further, there is a lack of clinically, culturally, and academically relevant elicitation protocols for oral and written discourse for Swedish-speaking school-age children. The objective with this study is to describe central characteristics of late language development, and to quantify linguistic characteristics and variability in students with LD through four structured and ecologically valid elicitation tasks. We will achieve this goal by collecting data from 10-16-year old students across three oral and one written task with increasing demands on language skills, executive functioning, and memory. The study has three interrelated aims:

Aim 1: Develop and pilot clinically useful discourse elicitation protocols. Based on previous research including from English-speaking children, we have created two age-appropriate and carefully matched stories for retelling, one with and one without picture support, as well as one oral and one written expository task on the same familiar topic (student’s favorite sport/game). The tasks enable analysis of interactions between different levels of support (with/without pictures), modality (spoken/written), word choices expressing target content (e.g., cause-effect) and grammatical complexity/accuracy. We will pilot these tasks on smaller groups of students (5-10 per grade).

Aim 2: Quantify typical patterns of interactions between language, task, and modality. Language assessment in the clinic has traditionally focused on aspects of language form (such as grammar). The protocols developed in aim 1 will enable analyses of developmental patterns of increasingly complex interactions between form, content, task, and modality, as well as the intra-individual variability in students with typical language development (TLD). We will collect data from students in 4th-9th grade (30 students/grade), which will form a comparison database. Through this aim, we expect to determine if and how age affects language complexity in terms of overall structure, word-choice, grammar, and content, and what tasks will elicit more complex language. The data will be made available for clinicians alongside with the elicitation protocols and analysis procedures.

Aim 3: Compare patterns of intra-individual variability in students with and without LD. We will collect data to describe which linguistic structures and contexts are particularly challenging for students with LD, and to specify measures that are clinically relevant to analyze. The data will provide the basis for the investigation of the selection of suitable language treatment targets, based on high-variability/vulnerable emerging language behaviors in students with LD. This variability is central in moving from one stable state in development to the next according to the dynamic systems view, and is an aspect often overlooked in traditional standardized language testing. The linguistic patterns across the four tasks from students with LD in grades 4-9 (at least 10 students/grade) will be compared to students with TLD (from data set collected in aim 2) to find patterns of similarity and change.

This project will allow us to better understand LDs, especially in the underserved school-age population. The project is unique in the Swedish context due to the focus on later academic language development, and the direct application to clinical practice.



Background: Language impairment (LI) in school-age children is less explored than in younger children, particularly in languages other than English. Emergentist theorists emphasize effects of frequency of words and morpho-syntactic patterns (lexical/structural frequency) on language processing. No studies have investigated frequency effects in morpho-syntactic error detection and correction tasks, which are common measures of metalinguistic awareness.

Aims: 1) Contribute to knowledge about LI in Swedish-speaking school-age children, 2) Investigate effects of frequency on error detection and correction, and 3) Discuss the metalinguistic demands of these tasks.

Methods: Ten-year-olds with LI or typical development (TD) participated in error detection (NLI=10, NTD=30) and correction (nLI=8, NTD=30). TD children also participated in error repetition (control experiment, n=29). Target sentences contained morpho-syntactic errors that characterize the language of young Swedish children with LI: the infinitive instead of past tense for regular/irregular verbs, and the omission of the indefinite article in common/neuter gender noun phrases. Target verbs and nouns were of high or low frequency (HF/LF).

Results: Error detection results showed that children with LI had specific difficulties with the target errors compared to a plural/singular control error. Children with LI had significantly weaker error detection/correction results than children with TD. All children showed effects of lexical and/or structural frequency in the tasks: LF verbs were associated with lower error sensitivity, fewer accurate corrections, and more accurate repetitions. Errors involving irregular verbs were more difficult to correct (but not to detect) than those involving regular verbs. Effects of lexical frequency were not seen for noun phrase errors, but the neuter article (with lower structural frequency) was associated with lower error sensitivity than the common article. Noun phrase errors were easier to correct than to detect for all children, and associated with lower accuracy in error repetition. The results indicate that error detection involves less explicit awareness than error correction and repetition. For error correction, LF words and verb errors seemed to involve more explicit awareness than HF words and noun phrase errors.

Conclusion: The results may help explain the variability in the morpho-syntactic performance of children with LI. Effects of frequency is important to consider when assessing grammatical/metalinguistic ability.

This dissertation was successfully defended on April 4, 2016 and was nominated for the NYU Steinhardt’s Outstanding Dissertation Award. Carol Miller (PSU) and Maria Grigos (NYU) served as readers.

The Dissertation Committee members were Christina Reuterskiöld (chair, NYU), Susannah Levi (NYU), Richard Schwartz (CUNY) and Sven Strömqvist (Lund University).


In this project the software SALT (Systematic Analysis of Language Transcripts) is combined with the Bloom & Lahey coding system to analyze content/form/use interactions in language samples from individuals with language disorders. The combination of these two models will hopefully be an aid for clinicians and researchers in planning goals for child-language intervention. This project is a collaboration with Dr. Christina Reuterskiöld, Dr. Klein, E. Altman, Emily Hadden, and Masters’ students from NYU.


The aim of this project was to find further support for the ”Dual Processing of Language” hypothesis, which postulates that formulaic utterances (familiar stereotyped expressions with conventionalized meaning, such as proverbs) are processed and stored holistically, in contrast to novel language, where lexicon and syntax is processed/retrieved separately. Tonal patterns and speaking rate were analyzed in proverbs and compared to matched control sentences in healthy adults and children. The results showed that personally and culturally familiar proverbs had common prosodic contours across speakers compared to matched control “novel” sentences. This gives support to the notion that words, syntax and prosody are stored and processed as wholes in formulaic expressions. This project was published in Applied Linguistics (2015). My advisor was Dr. Diana Sidtis.


I have been a member of Christina Reuterskiöld’s Language Development and Disorders Lab at New York University for six years and am still collaborating with them. For a description of members and current projects projects please visit the lab homepage.