Article Contentsp. 1281p. 1282p. 1283p. 1284p. 1285p. 1286p. 1287p. 1288p. 1289p. 1290p. 1291p. 1292p. 1293Issue Table of ContentsPhilosophical Transactions: Mathematical, Physical and Engineering Sciences, Vol. 358, No. 1769, Computers, Language and Speech: Formal Theories and Statistical Data (Apr. 15, 2000), pp. 1225-1431Front Matter [pp. ]Introduction: Combining Formal Theories and Statistical Data in Natural Language Processing [pp. 1227-1238]Formal Grammar and Information Theory: Together Again? [pp. 1239-1253]Finite-State Models, Event Logics and Statistics in Speech Recognition [pp. 1255-1266]Statistical and Logical Reasoning in Disambiguation [pp. 1267-1280]Towards a Psycholinguistic Computational Model for Morphological Parsing [pp. 1281-1293]Information Extraction from Broadcast News [pp. 1295-1310]Incorporating Linguistic Structure into Statistical Language Models [pp. 1311-1324]Incorporating Linguistic Theories of Pronunciation Variation into Speech- Recognition Models [pp. 1325-1338]The Role of Taxonomy in Language Engineering [pp. 1339-1355]Learning Dependency Transduction Models from Unannotated Examples [pp. 1357-1372]Stochastic Text Generation [pp. 1373-1387]Probabilistic Methods in Spoken-Dialogue Systems [pp. 1389-1402]Concept-to-Speech Synthesis by Phonological Structure Matching [pp. 1403-1417]Prosody Modelling in Concept-to-Speech Generation: Methodological Issues [pp. 1419-1431]Back Matter [pp. ]Towards a Psycholinguistic Computational Model for Morphological ParsingAuthor(s): R. Harald Baayen and Robert SchreuderSource: Philosophical Transactions: Mathematical, Physical and Engineering Sciences, Vol. 358,No. 1769, Computers, Language and Speech: Formal Theories and Statistical Data (Apr. 15,2000), pp. 1281-1293Published by: The Royal SocietyStable URL: http://www.jstor.org/stable/2666818 .Accessed: 17/05/2011 12:17Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unlessyou have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and youmay use content in the JSTOR archive only for your personal, non-commercial use.Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .http://www.jstor.org/action/showPublisher?publisherCode=rsl. .Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected] Royal Society is collaborating with JSTOR to digitize, preserve and extend access to PhilosophicalTransactions: Mathematical, Physical and Engineering Sciences.http://www.jstor.orgKM THE ROYAL MM SOCIETY Towards a psycholinguistic computational model for morphological parsing By R. Harald Baayen and Robert Schreuder Interfacuity Research Unit for Language and Speech, Max Planck Institute for Psycholinguistics, PO Box 310, 6500 AH Nijmegen, The Netherlands ([email protected]; [email protected]) Psycholinguistic experiments on visual word recognition in Dutch and other lan? guages show ubiquitous effects of word frequency for regular complex words. The present study presents a simulation experiment with a computational model for mor? phological segmentation that is designed on psycholinguistic principles. Results sug? gests that these principles, in combination with the presence of form and frequency information for complex words in the lexicon, protect the system against spurious segmentations and substantially enhance segmentation accuracy. Keywords: morphological segmentation; full-form storage; mental lexicon 1. Introduction When encountering the Dutch word bestelauto, Dutch readers understand this ortho? graphic string to denote 'delivery van'. They hardly ever become aware of an alter? native legitimate interpretation, 'berry counting car', corresponding to the segmen? tation bes-\- tel+ auto instead of the correct segmentation be+ stel+ auto. Neither do readers seem to have any difficulty in discounting uninterpretable sequences of Dutch morphemes that likewise span the orthographic string, such as bes + t + el + auto. The question addressed in this study is how readers might accomplish the selection of the correct segmentation of a morphologically complex word. The traditional approach in computational linguistics to morphological parsing proceeds in two steps. First, the set of possible segmentations that span the input string is calculated. Next, the combinatorial properties of morphemes are used to rule out illegal segmentations such as bes + t + el + auto, in which the verbal inflectional suffix -t follows a noun instead of a verb. In some statistically enhanced methods, co-occurrence frequencies are used to distinguish between probable parses ('delivery van') and improbable parses ('berry counting car'). The algorithm for determining the most probable segmentation described in the present paper is based on a rather different, psycholinguistically motivated concep? tual framework, that of parallel lexical activation and lexical competition. The wet- ware of the human brain makes use of massively parallel and interactive processing, in contrast to the hardware of the present-day single-processor PC which operates sequentially. The lexicon on which this algorithm operates also differs from the lexicons tradi? tionally found in linguistics and computational linguistics. The traditional approach Phil. Trans. R. Soc. Lond. A (2000) 358, 1281-1293 ? 2000 The Royal Society 12811282 R. H. Baayen and R. Schreuder in linguistics to the problem of morphological parsing is to assume that irregular com? plex words are stored in a lexicon along with the basic formative elements (stems and affixes), and that rules are used to segment regular complex words into their component constituents. However, various psycholinguistic studies report that high- frequency complex