Working memory and the revision of syntactic and discourse ambiguities.

Two hundred participants, 50 in each of 4 age ranges (19 -29 years, 30 - 49 years, 50 - 69 years, 70 -90 years) were tested for short-term working memory, speed of processing, and online processing of 3 types of sentences in which an initially assigned syntactic structure and/or semantic interpretation had to be revised. Self-paced reading times were longer for the segments that signaled the need for revision; there also were interactions of age and sentence type and speed of processing and sentence type, but not of working memory and sentence type on reading times for these segments. The results provide evidence that working memory does not support the processes that revise the structure and interpretation of sentences and discourse.Keywords: short-term memory, working memory, age, speed of processing, sentence and discourse comprehensionResumeDeux-cent participants, repartis egalement a raison de 50 dans quatre fourchettes d'âge (19 a 29 ans, 30 a 49 ans, 50 a 69 ans, 70 a 90 ans) ont ete soumis a des tests sur la memoire de travail a court terme, la vitesse de traitement et le traitement en ligne de trois types de phrases dans lesquelles la structure de syntaxe et/ou l'interpretation semantique initialement attribuee devait etre revisee. Les temps de lecture, au rythme de chacun, etaient plus longs pour les segments qui exigeaient une revision; on a aussi note des interactions entre l'âge et les types de phrases et entre la vitesse de traitement et le type de phrases, mais pas entre la memoire de travail et le type de phrases quant au temps de lecture pour ces segments. Les resultats demontrent que la memoire de travail n'a pas d'influence sur les processus de revision de la structure et de l'interpretation des phrases et des discussions.Mots-cles : memoire a court terme, memoire de travail, âge, vitesse de traitement, comprehension de phrases et de discussions.In this paper we investigated the role of short-term working memory in comprehension. Short-term working memory is often conceived as an aspect of the memory system responsible for effortful, conscious, and controlled manipulation of active information. Although some aspects of sentence comprehension may intuitively seem to require this type of deliberate effort (e.g., making sense of a complex run-on sentence), most operations involved in typical online sentence comprehension are highly skilled, automatic, and unconscious. In previous work, we argued that skilled operations of this type did not rely on short-term working memory resources (e.g., Caplan & Waters, 1999, 2013). In this paper we explore a middle ground between these extremes of automatized processing: namely, the role of short-term working memory in contexts where the reader is required to revise an initial interpretation during online sentence comprehension.We begin with a brief overview of literature related to this topic. There is broad agreement among psychologists that one component of human memory is a short-term memory (STM) that maintains small amounts of item and order information for short periods of time (Barrouillet & Camos, 2012; Brown, Neath, & Chater, 2007; Cowan, 2001; Henson, 1998; Jones, 1993; Jones, Hughes, & Macken, 2006, 2007; Jones & Macken, 1993; Jones, Macken, & Murray, 1993, Macken & Jones, 1995; Oberauer & Lewandowsky, 2011; Page & Norris, 1998; Salame & Baddeley, 1986; Verhaeghen & Basak, 2005). A longstanding question has been what role STM plays in cognitive functions. One of the points Crowder (1982) made in his famous critique of the concept of STM was that there was no convincing evidence for a functional role of STM, as he said, "STM, as traditionally defined, is remarkably insensitive to almost everything (p. 305)."Baddeley and Hitch (1974) addressed this issue in a series of studies in which they had participants perform memory, comprehension, and reasoning tasks while retaining sequences of digits of various lengths. …