Toward a connectionist model of recursion in human linguistic performance
➤ Gửi thông báo lỗi ⚠️ Báo cáo tài liệu vi phạmNội dung chi tiết: Toward a connectionist model of recursion in human linguistic performance
Toward a connectionist model of recursion in human linguistic performance
COGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performancereserved.Toward a Connectionist Model of Recursion in Human Linguistic PerformanceMorten II. ChristiansenSouthern Illinois UniversityNick ChaterUniversity of WarwickNaturally occurring speech contains only a limited amount of complex recursive structure, and this is reflected in the empirically docu Toward a connectionist model of recursion in human linguistic performancemented difficulties that people experience when processing such structures. We present a connectionist model of human performance in processing recursToward a connectionist model of recursion in human linguistic performance
ive language structures. The model is trained on simple artificial languages. We find that the qualitative performance profile of the model matches huCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performances and right-branching recursive constructions. We analyze how these differences in performance are reflected in the internal representations of the model by performing discriminant analyses on these representations both before and after training. Furthermore, we show how a network trained to process Toward a connectionist model of recursion in human linguistic performance recursive structures can also generate such structures in a probabilistic fashion. This work suggests a novel explanation of people's limited recursiToward a connectionist model of recursion in human linguistic performance
ve performance, without assuming the existence of a mentally represented competence grammar allowing unbounded recursion.I. INTRODUCTIONNatural languaCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performanceple are only able to deal easily with relatively simple recursive structures. Thus, for example, a doubly center-embedded sentence like (1) below is extremely difficult to understand.(I)The mouse that the cat that the day chased bit ran away.Direct all correspondence to: Morten II. Christiansen. Dep Toward a connectionist model of recursion in human linguistic performanceartment of Psychology. Southern Illinois University.Carbondale. IL 62901-6502; E-Mail: niorten@siu.edu.1571 58CHRISTIANSEN AND CHATERIn this paper, weToward a connectionist model of recursion in human linguistic performance
present a connectionist network which models the limited human abilities to process and generate recursive constructions. The “quasi-recursive” naturCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performanceatural language originates not from the project of trying to understand human linguistic performance which is the focus of this paper, but from the very different enterprise of specifying a “competence grammar" a set of rules and or principles which specify the legal strings of a language. It is sta Toward a connectionist model of recursion in human linguistic performancendardly assumed that, if the competence grammar allows a recursive construction to apply at all, it can apply arbitrarily many times. Thus, if (2) isToward a connectionist model of recursion in human linguistic performance
sanctioned by a recursive analysis with one level of recursion, then the grammar must thereby also sanction (1) with two levels of recursion and (3) wCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performanceway.Thus, the very idea that natural language is recursive requires a broadening of the notion of which sentences are in the language, to sentences like (3) which would presumably never be uttered or understood. In order to resolve the difference between language so construed and the language that h Toward a connectionist model of recursion in human linguistic performanceumans are able to produce and comprehend, a distinction is typically made between linguistic competence and human performance. C ompetence in this conToward a connectionist model of recursion in human linguistic performance
text refers to a speaker hearer’s knowledge of the language, and is the subject of linguistic inquiry. In contrast, psycholinguists study performance—COGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performanceuse of that knowledge. Il is here lhal "performance factors”, such as memory limitations, can be invoked to show that some sentences, while consistent with linguistic competence, will never actually be said, or understood, rhe competence performance distinction is also embodied in many symbolic mode Toward a connectionist model of recursion in human linguistic performancels of language processing, such as CC-READER (.lust & Carpenter, 1992). In this model, grammatical competence consists of a set of recursive productioToward a connectionist model of recursion in human linguistic performance
n rules w hich are applied to produce slate changes in a separate working memory. By imposing constraints on the capacity of the working memory systemCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performancen alternative account of people’s limited ability to do recursion, without assuming an internally represented grammar which allows unbounded recursion i.c., without invoking the competencc/performance distinction.In light of this discussion, it is clear that, from the point of view of modeling psych Toward a connectionist model of recursion in human linguistic performanceological processes, wre need not take the purported unbounded recursive structure of natural language as axiomatic. Nor need we take for granted the sToward a connectionist model of recursion in human linguistic performance
uggestion that a speaker hearer's knowledge of language captures such infinite recursive structure. Rather, the view’ that “unspeakable” sentences whiCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performancew' dominant in linguistics and many areas of the psychology of language. The challenge for a computational model such as the coimectionist model we propose isA CONNECTIONIST MODEL OF RECURSION1 59to account for those aspects of human comprehension production performance which are suggestive of the s Toward a connectionist model of recursion in human linguistic performancetandard recursive picture. If this can be done without making the assumption that the language processor really implements recursion, or that arbitrarToward a connectionist model of recursion in human linguistic performance
ily complex recursive structures arc really sentences of the language, then it presents an alternative to adopting this assumption. Therefore, in asseCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performancedle recursive structures: we need not require that conncclionisl systems be able to handle recursion in full generality.In this paper, we shall consider the phenomenon of natural language recursion in a ‘pure* and highly simplified form. Specifically, we train connectionist networks on small artific Toward a connectionist model of recursion in human linguistic performanceial languages, which exhibit the different types of recursive structure found in natural language. We do tills in order to address directly the classiToward a connectionist model of recursion in human linguistic performance
c arguments by Chomsky (1957) that recursion in natural language in principle rules out associative and finite state models of language processing. InCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performancermits us to address the in principle viability of comiectionist networks in handling recursion, in much the same way as simple artificial languages have been used, for example, to assess rhe feasibility of symbolic parameter-setting approaches to the learning of linguistic structure (Gibson & Wexler Toward a connectionist model of recursion in human linguistic performance. 1994; Niyogi & Berwick. 199Ố).The structure of this paper is as follows. We begin by distinguishing varieties of recursion in natural language, consToward a connectionist model of recursion in human linguistic performance
idering the three kinds of recursion discussed in Chomsky (1957). We then summarize past connectionist research dealing with natural language recursioCOGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performanceations using conncclionisl networks trained on these languages, rhe results suggest that the networks are able to handle recursion to a degree comparable with humans. We close by drawing conclusions for the prospects of conncelionist models of language processing.II. VARIETIES OF RECURSIONChomsky (1 Toward a connectionist model of recursion in human linguistic performance957) proposed that a recursive generative grammar consists of a set of phrase structure rules, complemented by a set of transformational rules (we shaToward a connectionist model of recursion in human linguistic performance
ll not consider transformational rules further below). Phrase structure rules have the form A —> BC, with the interpretation that the symbol A can be COGNITIVE SCIENCE Vol 23 (2) 1999, pp. 157-205ISSN 0364-0213Copyright '© 1999 Cognitive Science Society, Inc. All rights of reproduction in ony form r Toward a connectionist model of recursion in human linguistic performancelf (c.g., A —> /?/!). rhe new symbol can then itself be replaced by a further application of the recursive rule, and so on. Recursion can also arise through the application of a recursive set of rules, none of which need individually be recursive. When such rules are used successively to expand a pa Toward a connectionist model of recursion in human linguistic performancerticular symbol, the original symbol may eventually be derived. A recursive construction in a natural or artificial language is one that is modeled usToward a connectionist model of recursion in human linguistic performance
ing recursive rules: a language has recursive structure if it contains such constructions.160CHRISTIANSEN AND CHATERs --> NP VPNP --> N (comp S)VP -4Gọi ngay
Chat zalo
Facebook