Wednesday, July 27, 2016

The production of complement clauses in children with language impairment.

Steel, G., Rose, M., & Eadie, P. (2016). The Production of Complement Clauses in Children With Language Impairment. Journal of Speech, Language, and Hearing Research59(2), 330-341.

Although a large body of research on children with language impairments has focused on problems in early language development, few studies have examined more complex, later-developing linguistic structures. One such structure is the complement clause; a complement clause is a type of subordinate clause, that is, a group of words including a subject and verb but cannot stand alone as a sentence. The complement clause completes the meaning of a noun or verb in the sentence. For example, in the sentence, ‘He never expected that she would come’, the clause ‘that she would come’ is the complement of the verb. One type of complement clause involves a simple infinitive such as ‘I need to wash the orange’ or ‘The girl wanted Dad to play hide-and-seek’. Note that in the second example, the two clauses have difference subjects, that is, the subject of the clause ‘The girl wanted...’ is different from the subject in the complement clause ‘Dad to play hide-and-seek’. In this study, the authors were interested in comparing the production of these complement clauses that contain a different subject but cannot stand alone (i.e., nonfinite clauses) to another type of complement clause, the sentential complement. A sentential complement can stand alone as in the sentence, ‘I pretended that Teddy had a bath’.

The purpose of the study was to compare the performance of children with language impairment between 5;1 and 8;10, and typically developing children between 3;11 and 5;3 on the production of sentences with nonfinite and sentential complement clauses. Children’s responses were coded based on inclusion of the target clause structure, semantic (meaning) accuracy, syntactic (grammatical) accuracy, and the number of different verbs used in the task. When the groups were compared on their production of nonfinite clauses with different subjects, no differences were found for target clause structure, syntactic accuracy, and number of different verbs used although children with language impairment made more semantic errors. For sentential complements, children with language impairment included fewer sentential complement structures, used less syntactically and semantically accurate structures, and used fewer different verbs than the typically developing group.

The authors suggested that while most typically developing children had mastered the sentential complement form, children with language impairment may have less processing capacity in order to accurately combine multiple ideas into a sentence to produce sentential complements. In addition, when children with language impairment produce either type of complex clause structure, the demands of the complex structure may compromise their ability to accurately access and combine other components of the sentence, leading to semantic and syntactic errors.  The authors argued that use of elicitation tasks in clinical assessment will capture complex sentence production more comprehensively, in order for interventions to be more directly tailored to the strengths and weaknesses of each individual child.

Blogger: Alex Cross is completing a combined MClSc and PhD in speech language pathology. Her work focusing on reading will be part of both the Language and Working Memory and the Language, Reading, and Cognitive Neuroscience labs.

Wednesday, July 13, 2016

Good-enough representations in language comprehension.

Ferreira, F., Bailey, K.G.D., & Ferraro, V. (2002). Good-enough representations in language comprehension. Current Directions in Psychological Science, 11, 11-15.

How do we generate the meaning of a sentence beyond just knowing what each individual word means? One source of information to help us understand a sentence comes from our knowledge of syntax, or the linguistic rules we know for how words go together. Another source of information comes from our ability to problem-solve, learn or discover information based on our past experience and present context, also known as a heuristic process. Although not as detailed and specific as linguistic rules, heuristics can be applied very rapidly. The application of heuristics can allow for the quick creation of a ‘good enough’ representation of a sentence, which could result in misinterpretations in some cases.

The authors reviewed two sources of evidence to support their idea that initial processing of sentences is ‘good enough’. In one study, participants were shown ‘garden-path sentences’, that is, sentences in which the most likely grammatical interpretation suggested by the first words of the sentence turns out to be wrong. The assumption has been that when readers become aware of the difficulty understanding, they re-interpret the sentence coming to a full and complete representation of the sentence. The results of the reported study, however, indicated that even though participants were able to indicate their understanding of the key confusion after reviewing the garden-path sentence, they persisted in misunderstanding another aspect of the sentence. The authors suggested that the processing of the sentences was ‘good enough’ to allow comprehension of the main idea required for the context but did not reflect the true content of the sentence. In another study, participants were asked to judge the possible truth of active sentences like ‘The man bit the dog’ and passive sentences like ‘The dog was bitten by the man’.  More errors were made on the passive sentences especially when the word order was unexpected. The authors argued that the linguistic representation of passive sentences was fragile, and was often outweighed by the heuristics favouring the error response.

The authors suggested that ‘good enough’ language processing is often successful because the context of the sentence or the conversation supports the linguistic interpretation generated. In cases where the context or heuristics are unclear or unknown, however, individuals with weak linguistic representations such as children with language impairment may fail to comprehend the sentence.

Blogger: Lisa Archibald

Tuesday, June 28, 2016

The sensitivity of children with SLI to phonotactic probabilities during lexical access.

Quemart, P. & Maillart, C. (2016). The sensitivity of children with SLI to phonotactic probabilities during lexical access. Journal of Communication Disorders, 61, 48-59.

This study examined how children with specific language impairment (SLI) use certain language cues to help with vocabulary learning. Children with SLI have typical intelligence, hearing, and motor skills, but struggle with learning language at the same rate as their peers. Examining different factors that lead to successful vocabulary acquisition may contribute to our understanding of the language difficulties seen in children with SLI. Sensitivity to phonotactic constraints may be one factor that helps language learners with vocabulary acquisition. Phonotactic constraints are how sounds can be ordered within words within a particular language. For instance, English words can end with /fs/ (e.g., ‘laughs’), but cannot begin with /fs/. Phontactic constraints vary within a language, with some constraints occurring across more words, or being more frequent, than others. These constraints are easily acquired: By about nine months of age, infants are able to differentiate between legal and non-legal phonotactic constraints in their native language. Once acquired, knowledge of phonotactic constraints is involved in word learning, spoken word recognition, and the selection of appropriate verb inflections. The authors wanted to examine whether children with SLI were sensitive to phonotactic constraints, and whether this sensitivity differed from that of typically developing children.

Children with SLI and typically developing children heard a list of words and non-words, one at a time, and had to correctly identify the real words and correctly reject the nonwords. Importantly, the nonwords were made up of phonotactic constraints that were highly frequent in their native language (French), or phonotactic constraints that were not frequent. That is, some of the nonwords were made up of sound combinations that occur in many French words, while others were made up of sound combinations that occur in few French words. If the children were sensitive to patterns involving phonotactic constraints, it should be easier to quickly reject the nonwords containing non-frequent sound combinations, and harder to reject the nonwords containing highly frequent sound combinations. The authors found that compared to typically developing children, children with SLI were worse at correctly rejecting nonwords that contained highly frequent phonotactic constraints. That is, the children with SLI had more difficulty rejecting nonwords that sounded similar to other words in their vocabulary. Typically developing children, on the other hand, could quickly and accurately reject all of the nonwords, perhaps because they had a better-developed vocabulary.

It seems that children with SLI are at least sensitive to the phonotactic constraints within their native language. But, they may be over-reliant on this cue compared to children with typical language skills, who can instead rely on their stronger vocabulary knowledge. These results underscore the importance of vocabulary knowledge in supporting language processing, and the need to support vocabulary learning in children with SLI perhaps on an extended basis.

Blogger: Nicolette Noonan is a Ph.D. student with Drs. Lisa Archibald and Marc Joanisse