Sentence processing: The role of memory and expectations in understanding relative clauses

Levy, R., Fedorenko, E., & Gibson, E. (2013). The syntactic complexity of Russian relative clauses. Journal of Memory and Language, 69, 461-495.

Human language is unique amongst communication systems because of its expressivity: We can hear a sentence we have never heard before and understand its meaning. But how does this happen? We do not have the memory resources to pursue every possible interpretation of a sentence, so there must be some sort of cognitive constraint that guides our understanding. Possible cognitive constraints have been investigated by examining how long it takes for people to understand or read sentences that present some confusion, that is, syntactic ambiguity resolution.

Research in processing relative clauses is important in understanding syntactic ambiguity resolution because these are one of the most complex sentence forms we encounter, and they offer an avenue to explore language’s rich expressive capacity. There are two types of relative clauses examined in this paper:

1)   Subject-extracted relative clause

The reporter who attacked the senator hoped for a story

            subject                 relative clause verb        object

2)   Object-extracted relative clause

The reporter who the senator attacked hoped for a story

                 object                                      subject  relative clause verb

The finding is that object relative clause sentences are more difficult to process than subject relative clause sentences. The authors distinguish between two theories that explain this difference in processing: Memory Based Theories and Expectation Based Theories. Memory Based Theories state that the comprehender is actively storing, integrating, and retrieving incoming information from the sentence in an online fashion. In these theories, the more material that needs to be processed, the more overloaded the memory system gets and the higher the processing load. Based on this theory, sentence 2 is harder to process because you need to hold more information in memory before you get to the verb attacked, which is the point where the meaning of the sentence becomes clear.

Expectation Based Theories assume that your previous language experience and semantic content shapes your sentence processing abilities. In fact, for Expectation Based Theories, greater semantic information (more words) in the sentence helps sharpen your expectations and guides your processing, whereas Memory Based Theories would predict that this impairs in processing. Based on this theory, sentence 2 is harder to process because it follows an Object-Subject-Word order. We are used to hearing Subject-Verb-Object orders in English, so this violates our expectations.

Even though the theories were difficult to disentangle in this paper, there are two interesting ideas in this paper: Memory limits may constrain language processing, and language processing may be limited by a lack of language knowledge. Considering which of these resources may be impacting language processing may help to understand a particular individual’s language functioning.

Blogger: Nicolette Noonan

Sensitivity to Word Order Cues by Normal and Language/Learning Disabled Adults

Plante, E., Gómez, R., & Gerken, L. A. (2002). Sensitivity to word order cues by normal and language/learning disabled adults. Journal of Communication Disorders, 35, 453–462.

The ability to recognize statistical regularities in inputs is considered to be an important learning mechanism emerging early in life and supporting language acquisition (Gómez & Gerken, 1999). Children with specific language impairment (SLI) on the other hand, have been found to have difficulty extracting statistical regularities from the input (Bishop, 1982).

 

In this article, Plante, Gómez, and Gerken (2002) examined statistical learning of sequential word order strings generated by an artificial grammar. Participants included 32 adults with and without personal or familial history of language/learning disabilities (L/LD). In the artificial grammar task, adults listened to novel consonant-vowel-consonant (CVC) words, with grammatical rules regarding syllable sequences for five minutes. After listening to the language, participants were asked to judge whether a new sentence spoken in the novel language obeys or violates the rules of the language.  After only five minutes of exposure to the training set, typically developing adults were able to exceed chance performance, whereas adults with L/LD showed significantly lower performance compared to the control group.

These results suggest that adults with L/LD may have difficulty in recognizing word order compared to typically developing adults. It may be adults with L/LD have more difficulty extracting the regularities available in linguistic input. Another study reviewed in this blog by Evans et al. (2009)  suggested that children with language impairment requirement more repetitions to learn the statistical patterns available in the input.

Blogger: Areej Balilah

What Causes Specific Language Impairment in Children?

Bishop, D.V. (2006). What causes specific language impairment in children? Current Directions in Psychological Science, 15 (5), 217-221.

This week’s article is an update of Bishop’s "The underlying nature of specific language impairment" discussed in the last posting looking at the more broad hypotheses of genetic vs. multiple underlying defictis as explanations for SLI.

            Studies showing that SLI runs in families suggest a genetic influence; however these results can be confounded by the fact that families share environments. Twin studies have been used to test this idea and have shown that monozygotic twins have more similar SLI profiles than dizygotic twins. Nevertheless, SLI does not follow a simple inheritance pattern suggesting that there is no ‘gene for language’. Rather, SLI may be described as a complex genetic disorder, that is, a disorder of several genes interacting with environmental factors. Indeed, Bishop’s own twin studies have provided evidence of separable genetic influences on a phonological short-term memory task (non-word repetition), and a grammatical task (adding inflectional ending to verbs), but not on an auditory perception task.  

            Overall, children with multiple deficits were more likely to be identified as SLI. Multiple routes to language acquisition may provide a protective factor such that redundancies and compensations occur in the system even if a single weakness or deficit exists. Multiple deficits, however, would impair language learning and may result in identification as SLI.

Future directions will continue to examine the development of language abilities over time, and investigate genetic and environmental interactions. 

Blogger: Lauren Perduk is currently completing her Masters of Clinical Science in Speech Language Pathology. She often participants in lab meetings with the Language and Working Memory lab.