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.

Many theories have been proposed to explain difficulty with center embedded constructions, most attributing the problem to some kind of limited capacity short-term memory. However, these theories have developed for the most part independently of more traditional memory research, which has focused on uncovering general principles such as chunking and interference. This article attempts to gain some unification with this research by suggesting that an interesting range of core sentence processing phenomena can be explained as interference effects in a sharply limited syntactic working memory. These include difficult and acceptable embeddings, as well as certain limitations on ambiguity resolution, length effects in garden path structures, and the requirement for locality in syntactic structure. The theory takes the form of an architecture for parsing which can index no more than two constituents under the same syntactic relation. A limitation of two or three items shows up in a variety o...

The most predominant language processing theories have, for some time, been based largely on structured knowledge and relatively simple rules. These symbolic models intentionally segregate syntactic information processing from statistical information as well as semantic, pragmatic, and discourse influences, thereby minimizing the importance of these potential constraints in learning and processing language. While such models have the advantage of being relatively simple and explicit, they are inadequate to account for learning and validated ambiguity resolution phenomena. In recent years, interactive constraint-based theories of sentence processing have gained increasing support, as a growing body of empirical evidence demonstrates early influences of various factors on comprehension performance. Connectionist networks are one form of model that naturally reflect many properties of constraint-based theories, and thus provide a form in which those theories may be instantiated. Unfortunately, most of the connectionist language models implemented until now have involved severe limitations, restricting the phenomena they could address. Comprehension and production models have, by and large, been limited to simple sentences with small vocabularies (cf. St. John & McClelland, 1990). Most models that have addressed the problem of complex, multi-clausal sentence processing have been prediction networks (cf. Elman, 1991; Christiansen & Chater, 1999a). Although a useful component of a language processing system, prediction does not get at the heart of language: the interface between syntax and semantics.

Recent developments in connectionist architectures for symbolic computation have made it possible to investigate parsing in a connectionist network while still taking advantage of the large body of work on parsing in symbolic frameworks. The work discussed here investigates syntactic parsing in the temporal synchrony variable binding model of symbolic computation in a connectionist network. This computational architecture solves the basic problem with previous connectionist architectures, while keeping their advantages. However, the architecture does have some limitations, which impose constraints on parsing in this architecture. Despite these constraints, the architecture is computationally adequate for syntactic parsing. In addition, the constraints make some significant linguistic predictions. These arguments are made using a specific parsing model. The extensive use of partial descriptions of phrase structure trees is crucial to the ability of this model to recover the syntactic st...

This article explores the use of Simple Synchrony Networks (SSNs) for learning to parse English sentences drawn from a corpus of naturally occurring text. Parsing natural language sentences requires taking a sequence of words and outputting a hierarchical structure representing how those words fit together to form constituents. Feed-forward and Simple Recurrent Networks have had great difficulty with this task, in part because the number of relationships required to specify a structure is too large for the number of unit outputs they have available. SSNs have the representational power to output the necessary O(n 2 ) possible structural relationships, because SSNs extend the O(n) incremental outputs of Simple Recurrent Networks with the O(n) entity outputs provided by Temporal Synchrony Variable Binding. This article presents an incremental representation of constituent structures which allows SSNs to make effective use of both these dimensions. Experiments on learning to ...

xi 1 Introduction 1 1.1 Intuitions and Evidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Previous Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.1 The Classical Dual Route Model . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.2 Seidenberg and McClelland 1989 . . . . . . . . . . . . . . . . . . . . . . 10 1.2.3 Plaut and Shallice 1993 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.2.4 Plaut et al. 1996: Naming . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.2.5 Bullinaria 1996 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.2.6 Plaut 1997: Lexical Decision . . . . . . . . . . . . . . . . . . . . . . . . . 15 1.2.7 Harm and Seidenberg 1998: Naming . . . . . . . . . . . . . . . . . . . . 16 1.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2 A New Computational Model 18 2.1 Principles . . . . . . . . . . . . . . . . . . . . . . . . ...

For connectionist networks to be adequate for higher level cognitive activities such as natural language interpretation, they have to generalize in a way that is appropriate given the regularities of the domain. Fodor and Pylyshyn (1988) identified an important pattern of regularities in such domains, which they called systematicity. Several attempts have been made to show that connectionist networks can generalize in accordance with these regularities, but not to the satisfaction of the critics. To address this challenge, this paper starts by establishing the implications of systematicity for connectionist solutions to the variable binding problem. Based on the work of Hadley (1994a), we argue that the network must generalize information it learns in one variable binding to other variable bindings. We then show that temporal synchrony variable binding (Shastri and Ajjanagadde, 1993) inherently generalizes in this way. Therebywe show that temporal synchronyvariable binding is a connect...

Efficient knowledge representation and reasoning is an important component of intelligent activity, and is a crucial aspect in the design of large-scale intelligent systems. This dissertation explores the design, analysis, and implementation of massively parallel knowledge representation and reasoning systems which can encode very large knowledge bases and respond to a class of queries in real-time, with reasoning episodes expected to span a fraction of a second. The dissertation attempts to design efficient, large-scale knowledge base systems by: (i) exploiting massive parallelism; and (ii) constraining representational and inferential capabilities to achieve tractability, while still retaining sufficient expressive power to capture a broad class of reasoning in intelligent systems. To this end, shruti, a connectionist reasoning system which models reflexive--- i.e., effortless and spontaneous---reasoning serves as the knowledge representation and reasoning framework. Shruti-based mas...

What precisely is the universal nature of the human syntactic parser, such that it copes easily with some embedded structures, yet fails so dramatically on others (e.g., classic double center-embeddings)? A theory is proposed in the form of an architecture for parsing based on two simple ideas. The first is that human short-term memory is an indexing structure which can give rise to interference effects (storage limitations) when contents overlap with respect to the indices. For parsing, the contents are syntactic structures, and the indices are potential structural relations. The second idea is that the capacity of STM is the minimum capacity required to support the basic functions of parsing. The theory successfully accounts for the contrasts between over 50 difficult and acceptable constructions from English, French, German, Hebrew, Japanese, Mandarin, and Spanish. The theory has independent psychological and computational motivation, and is a functional part of a broader cognitive ...

In recent years it has been claimed that connectionist methods of representing compositional structures, such as lists and trees, support a new form of symbol processing known as holistic computation. In a holistic computation the constituents of an object are acted upon simultaneously, rather than on a one-by-one basis as is typical in traditional symbolic systems. This thesis presents rstly, a critical examination of the concept of holistic computation, as described in the literature, along with a revised de nition of the concept that aims to clarify the issues involved. In particular it is argued that holistic representations are not necessary for holistic computation and that holistic computation is not restricted to connectionist systems. Secondly, an evaluation of the capacity of a particular connectionist representation, the Sequential RAAM, to generate representations that support holistic symbol processing is presented. It is concluded that the Sequential RAAM is not as eective a vehicle for holistic symbol processing as it initially appeared, but that there may be some scope for improving its performance. Acknowledgements I would like to thank the following people: My supervisor, Peter Hancox, for his support, advice and guidance throughout my PhD. The other members of my thesis group, Russell Beale, Riccardo Poli and Ela Claridge for the advice, questions and suggestions they oered during progress reviews. Additionally I'd like to thank Russell for his comments on this thesis and both Russell and Riccardo for the informal discussions I had with them about my work.