Does knowledge of language consist of mentally-represented rules? Rumelhart and McClelland have described a connectionist (parallel distributed processing) model of the acquisition of the past tense in English which successfully maps many stems onto their past tense forms, both regular (walk/walked) and irregular (go/went), and which mimics some of the errors and sequences of development of children. Yet the model contains no explicit rules, only a set of neuron-style units which stand for trigrams of phonetic features of the stem, a set of units which stand for trigrams of phonetic features of the past form, and an array of connections between the two sets of units whose strengths are modified during learning. Rumelhart and McClelland conclude that linguistic rules may be merely convenient approximate fictions and that the real causal processes in language use and acquisition must be characterized as the transfer of activation levels among units and the modification of the weights of their connections. We analyze both the linguistic and the developmental assumptions of the model in detail and discover that (1) it cannot represent certain words, (2) it cannot learn many rules, (3) it can learn rules found in no human language, (4) it cannot explain morphological and phonological regularities, (5) it cannot explain the differences between irregular and regular forms, (6) it fails at its assigned task of mastering the past tense of English, (7) it gives an incorrect explanation for two developmental phenomena: stages of overregularization of irregular forms such as bringed, and the appearance of doubly-marked forms such as ated, and (8) it gives accounts of two others (infrequent overregularization of verbs ending in t/d, and the order of acquisition of different irregula...

This paper appeared in N. Gilbert and R. Conte (Eds.), Artificial Human language provides, among other things, a mechanism for distinguishing between relevant objects in the natural environment. This mechanism is composed of two components – forms and meanings – which must be shared by the community of language users. The lexicon constitutes

We present Embodied Construction Grammar, a formalism for linguistic analysis designed specifically for integration into a simulation-based model of language understanding. As in other construction grammars, linguistic constructions serve to map between phonological forms and conceptual representations.

Abstract not found

Cross-linguistic similarities and differences in early lexical and grammatical development are reported for 1001 English-speaking children and 386 Italian-speaking children between 1;6 and 2;6. Parents completed the English or Italian versions of the MacArthur Communicative Development Inventory: Words and Sentences, a parent report instrument that provides information about vocabulary size, vocabulary composition and grammatical complexity across this age range. The onset and subsequent growth of nouns, predicates, function words and social terms proved to be quite similar in both languages. No support was found for the prediction that verbs would emerge earlier in Italian, although Italians did produce a higher proportion of social terms, and there were small but intriguing differences in the shape of the growth curve for grammatical function words. A strikingly similar nonlinear relationship between grammatical complexity and vocabulary size was observed in both languages, and examination of the order in which function words are acquired also yielded more similarities than differences. However, a comparison of the longest sentences reported for a subset of children demonstrates large cross-linguistic differences in the

We describe a computational model of the acquisition of early grammatical constructions that exploits two essential features of the human language learner: significant prior knowledge of concepts and individual lexical items, and sensitivity to the statistical properties of the input data. Such principles, previously applied to lexical acquisition, are shown to be useful and necessary for learning the structured mappings between form and meaning needed to represent phrasal and clausal constructions. We describe an algorithm based on Bayesian model merging that can induce a set of grammatical constructions based on simpler previously learned constructions in combination with new utterance-situation pairs. The resulting model shows how cognitive and computational constraints can intersect to produce a course of learning consistent with data from studies of child language acquisition.

Complex words consist of morphemic subunits that can recombine to form other words. Thus midnight is standardly analyzed as consisting of the prefix mid- and stem night, which also occur in words such as midstream and nightly. A considerable body of empirical and theoretical research suggests that morphological structure governs the representation of words in memory and that many words are decomposed into morphological components in processing. We investigated an alternative approach in which morphology arises from the interaction of semantic, phonological, and orthographic codes. Five cross-modal lexical decision experiments show that the magnitude of priming (e.g., for pairs such as teacher-teach) is affected by the degree of semantic and phonological overlap between words. Crucially, items that are only moderately similar produce intermediate facilitation effects (e.g., latelylate) . This pattern is observed both for words standardly treated as morphologically related (e.g., teacher-teach) and for morphologically unrelated words that exhibit similar degrees of semantic and phonological overlap (e.g., snarl-sneer). The results can be understood in terms of connectionist models employing distributed representations rather than discrete morphemes. Graded semantic and phonological similarity effects in priming: Evidence for a distributed connectionist approach to morphology One of the fundamental problems in the study of language is to characterize knowledge of words and how this knowledge is used in comprehension and production. The focus of the present article is on derivational morphology, the aspect of lexical knowledge concerning the structure and formation of complex words. Words such as baker and talking appear to consist of components, traditionally called m...

Like many cryptosystems, PGPfone[13] requires a method of reliably exchanging binary data over noisy phone lines. This paper describes a method of encoding binary data into a "radio alphabet," using a feature-based distance metric to measure phonetic confusibility, then using this metric in a GA to select appropriate words from a larger list of candidate words. This work indicates several larger issues that should be addressed in any (human) language engineering project.

Abstract not found

WH-questions and the mechanisms of language acquisition, page 2 It is no understatement to say that the central issue in the theory of language acquisition is whether children actually learn language and construct a grammar based on the data to which they are exposed, or whether they set the parameters of an autonomous language acquisition device