Interference control in language processing: the effects of age and working memory

Language processing encompasses different levels of analysis, involving activation of semantic and perceptual features of concepts, binding of concepts, and understanding of the overall situation. Older adults or individuals with less working memory capacity have often been reported to have more difficulties in binding individual concepts in a sentence to form an integrated representation of ideas (e.g., thematic roles assignment). In contrast, situation understanding or simulation of language that draws on world knowledge and experience has been argued to be obligatory and relatively effortless, thus demonstrating relative resilience in the face of aging and working memory constraints. Recent models of language processing suggest that conceptual binding difficulties often observed in complex syntactic structures may be caused by semantic similarity of the to-be-integrated concepts and deficits in interference control in working memory. Furthermore, emerging evidence implies that situation construction/simulation can be compromised for individuals with less working memory resources or when binding concepts (that is the basis for enriched and refined situation simulation) is too resource-consuming. The study consisted of two experiments and was designed to 1) test whether similarity-based interference in working memory differentially impairs conceptual combination with age, and 2) whether such processes have downstream effects on situation simulation. In Experiment 1, the syntax (Object-relative clauses, ORC vs. Subject-relative clauses, SRC) and the semantic similarity between noun phrases (NPs) (same vs. different category) were manipulated (Gordon et al., 2006) to investigate their respective effect on conceptual binding among younger and older adults. In Experiment 2, the syntax and the similarity manipulations were crossed with match condition (match vs. mismatch between implied shape and actual shape) in a sentence-picture verification paradigm (Zwaan et al., 2002) to investigate the effect of controlling interference during binding concepts on generating perceptual-level inferences. Results of Experiment 1 revealed that the combination of syntactic and semantic complexity differentially affected older adults’ comprehension and there was evidence that older adults with less working memory resources showed poorer comprehension, particularly when the difficult syntax with similar NPs. Online eye-movement measures also showed that older adults with less working memory resources spent differentially more time at the critical relative clause region in this condition. Results of Experiment 2 demonstrated that there was a reliable mismatch effect in response latency for picture verification and that the effect did not vary as a function of syntax, similarity or age. Further analysis revealed that this mismatch effect was not associated with working memory for either age group across all the conditions, providing converging evidence that perceptual simulation was impermeable to syntactic complexity, semantic interference, and working memory constraint and might reflect distinctive and obligatory processes in language understanding beyond conceptual combination. Altogether, these results support a) a similarity-based interference control account of conceptual binding in working memory; b) a resource-independent and obligatory view of perceptual simulation that is immune to interference and working memory capacity; and c) a view of age differences in language understanding that processes requiring binding multiple concepts while controlling distractions in working memory is compromised, but that processes underlying situation-level inferences are preserved, even in face of interference and working memory constraint