Time underlies many interesting human behaviors. Thus, the question of how to represent time in connectionist models is very important. One approach is to represent time implicitly by its effects on processing rather than explicitly (as in a spatial representation). The current report develops a proposal along these lines first described by Jordan (1986) which involves the use of recurrent links in order to provide networks with a dynamic memory. In this approach, hidden unit patterns are fed back to themselves; the internal representations which develop thus reflect task demands in the context of prior internal states. A set of simulations is reported which range from relatively simple problems (temporal version of XOR) to discovering syntactic/semantic features for words. The networks are able to learn interesting internal representations which incorporate task demands with memory demands; indeed, in this approach the notion of memory is inextricably bound up with task processing. These representations reveal a rich structure, which allows them to be highly context-dependent while also expressing generalizations across classes of items. These representations suggest a method for representing lexical categories and the type/token distinction.

Abstract. In this paper three problems for a connectionist account of language are considered: 1. What is the nature of linguistic representations? 2. How can complex structural relationships such as constituent structure be represented? 3. How can the apparently open-ended nature of language be accommodated by a fixed-resource system? Using a prediction task, a simple recurrent network (SRN) is trained on multiclausal sentences which contain multiply-embedded relative clauses. Principal component analysis of the hidden unit activation patterns reveals that the network solves the task by developing complex distributed representations which encode the relevant grammatical relations and hierarchical constituent structure. Differences between the SRN state representations and the more traditional pushdown store are discussed in the final section.

The problems of access -- retrieving linguistic structure from some mental grammar -- and disambiguation -- choosing among these structures to correctly parse ambiguous linguistic input -- are fundamental to language understanding. The literature abounds with psychological results on lexical access, the access of idioms, syntactic rule access, parsing preferences, syntactic disambiguation, and the processing of garden-path sentences. Unfortunately, it has been difficult to combine models which account for these results to build a general, uniform model of access and disambiguation at the lexical, idiomatic, and syntactic levels. For example psycholinguistic theories of lexical access and idiom access and parsing theories of syntactic rule access have almost no commonality in methodology or coverage of psycholinguistic data. This paper presents a single probabilistic algorithm which models both the access and disambiguation of linguistic knowledge. The algorithm is based on a parallel parser which ranks constructions for access, and interpretations for disambiguation, by their conditional probability. Low-ranked constructions and interpretations are pruned through beam-search; this pruning accounts, among other things, for the garden-path effect. I show that this motivated probabilistic treatment accounts for a wide variety of psycholinguistic results, arguing for a more uniform representation of linguistic knowledge and for the use of probabilisticallyenriched grammars and interpreters as models of human knowledge of and processing of language.

Abstract not found

Human language processing relies on many kinds of linguistic knowledge, and is sensitive to their frequency, including lexical

This paper describes a new, psychologically-plausible model of human sentence interpretation, based on a new model of linguistic structure, Construction Grammar. This on-line, parallel, probabilistic interpreter accounts for a wide variety of psycholinguistic results on lexical access, idiom processing, parsing preferences, and studies of gap-filling and other valence ambiguities, including various frequency effects. We show that many of these results derive from the fundamental assumptions of Construction Grammar that lexical idioms, idioms, and syntactic structures are uniformly represented as grammatical constructions, and argue for the use of probabilistically-enriched grammars and interpreters as models of human knowledge of and processing of language. Submitted to Cognitive Science ii 1 Introduction In the last twenty years, the field of cognitive science has seen an explosion in the number of computational models of cognitive processing. This is particularly true in the mode...

This introduction sets the stage for the papers making up this special issue. Its focus is on two major problems in the study of lexical processing-determining the phases involved in recognising a spoken word and identifying the nature of different types of contextual influences on these phases. An attempt is made to decompose the process of recognising a word into phases which have both theoretical and empirical consequences. A similar analytic approach is taken in the discussion of the problem of context effects by distinguishing qualitatively different types of context (lexical, intra-lexical, syntactic, semantic, and inter-pretative). We argue that such an approach is necessary to make explicit the relationship between a particular type of contextual information and the phase(s) of processing at which it has its impact. 1.

Current models of word recognition differ on the potential influence of lexical knowledge on sublexical processes. The present study addresses this issue through a paradigm based on the perceptual ‘‘migration’ ’ of linguistic units between two stimuli presented simultaneously, resulting in an illusory percept. For example, ‘‘kinrtrorverrsy’ ’ and ‘‘bosrglorrarfe’ ’ were played at the same time, with subjects judging, in one condition, if ‘‘controversy’ ’ was presented and, in another condition, if ‘‘bisrglorrarfe’ ’ was presented. The results showed that, in dichotic listening, the migration of the vowel, leading to erroneous detection of the prespecified target, occurred less often with real word targets than with nonsense word targets. However, when the two items of the pairs were played in both ears at different amplitudes so that they were perceived to be closer to each other than in the dichotic situation, the lexical effect only remained when the mispronounced phoneme was in an unstressed syllable. This observation suggests that the mispronunciation of stressed syllables of words impairs lexical access in such a way that no lexical influence on early processing stages is possible. In contrast, the mispronunciation of unstressed syllables does not affect lexical access, thus allowing the lexical influence to take place. These results are problematic for temporally sequential (left-to-right) models of lexical access. Rather, they support models that include the stress patterns of words as an important aspect of both lexical access and the interaction between sublexical and lexical knowledge.