This article describes a neural network model of speech motor skill acquisition and speech production that explains a wide range of data on variability, motor equivalence, coarticulation, and rate effects. Model parameters are learned during a babbling phase. To explain how infants learn language-specific variability limits, speech sound targets take the form of convex regions, rather than points, in orosensory coordinates. Reducing target size for better accuracy during slower speech leads to differential effects for vowels and consonants, as seen in experiments previously used as evidence for separate control processes for the 2 sound types. Anticipatory coarticulation arises when targets are reduced in size on the basis of context; this generalizes the well-known look-ahead model of coarticulation. Computer simulations verify the model's properties. The primary goal of the modeling work described in this article is to provide a coherent theoretical framework that provides explanations for a wide range of data concerning the articulator movements used by humans to produce speech sounds. This is carried out by formulating a model that transforms strings of phonemes into continuous articulator movements for

Introduction 2 0.1. Licensing: by cue or by prosody 2 0.2. Phonetics in phonology: the downward arrow and alternatives 3 0.3. An example of cue licensing: retroflexion 4 0.4. Cues 6 0.5. Cue weighting 9 0.6. Cue duration 10 0.7. The descriptive system 10 0.8. Excessive variability 13 0.9. Extensions 14 0.9.1 Direct reference to cues? 14 0.9.2 Intersegmental timing 15 0.9.3 Intrasegmental timing 15 0.9.4. Variable timing 16 Part I: Against syllable-based accounts of neutralization 1.1. Lithuanian 17 1.2. The representation of neutralized voicing 21 1.3. Word domain effects in voicing neutralization 24 2. Generalizing from Lithuanian 25 2.1. Greek and Sanskrit 25 2.1.1. Sanskrit 27 2.1.2. Greek 28 2.1.3 Aspiration neutralized 29 2.2. Voicing neutralization in Polish and Russian 30 2.3. German syllabification and devoicing 38 2.3.1 The facts 38 2.3.2. Correlations between neutralization and the syllable 40 3. A second voicing neutralization pattern: before obstr

Abstract not found

Abstract not found

This thesis explores approaches to the problem of spoken document retrieval (SDR), which is the task of automatically indexing and then retrieving relevant items from a large collection of recorded speech messages in response to a user specified natural language text query. We investigate the use of subword unit representations for SDR as an alternative to words generated by either keyword spotting or continuous speech recognition. Our investigation is motivated by the observation that word-based retrieval approaches face the problem of either having to know the keywords to search for a priori, or requiring a very large recognition vocabulary in order to cover the contents of growing and diverse message collections. The use of subword units in the recognizer constrains the size of the vocabulary needed to cover the language; and the use of subword units as indexing terms allows for the detection of new user-specified query terms during retrieval. Four

In foundational works of generative phonology it is claimed that subjects can reliably discriminate between possible but non-occurring words and words that could not be English. In this paper we examine the use of a probabilistic phonological parser for words to model experimentally-obtained judgements of the acceptability of a set of nonsense .words. We compared various methods of scoring the goodness of the parse as a predictor of acceptability. We found that the probability of the worst part is not the best score of acceptability, indicating that classical generative phonology and Optimality Theory miss an important fact, as these approaches do not recognise a mechanism by which the frequency of wellformed parts may ameliorate the unacceptability of low-frequency parts. We argue that probabilistic generative grammars are demonstrably a more psychologically realistic model of phonological competence than standard generative phonology or Optimality Theory.

Declarative phonology is a program of research that was motivated in part by the need for theories of phonology that can be implemented on a computer. While it is clear that such a development would be beneficial for both theoretical and field phonology, it is not immediately obvious how one should go about implementing phonological models. The

The study of phonotactics (e.g., the ability of English speakers to distinguish possible words like blick from impossible words like *bnick) is a central topic in phonology. We propose a theory of phonotactic grammars and a learning algorithm that constructs such grammars from positive evidence. Our grammars consist of constraints that are assigned numerical weights according to the principle of maximum entropy. Possible words are assessed by these grammars based on the weighted sum of their constraint violations. The learning algorithm yields grammars that can capture both categorical and gradient phonotactic patterns. The algorithm is not provided with any constraints in advance, but uses its own resources to form constraints and weight them. A baseline model, in which Universal Grammar is reduced to a feature set and an SPE-style constraint format, suffices to learn many phonotactic phenomena. In order to learn nonlocal phenomena such as stress and vowel harmony, it is necessary to augment the model with autosegmental tiers and metrical grids. Our results thus offer novel, learning-theoretic support for such representations. We apply the model to English syllable onsets, Shona vowel harmony, quantity-insensitive stress typology, and the full phonotactics of Wargamay, showing that the learned grammars capture the distributional generalizations of these languages and accurately predict the findings of a phonotactic experiment.

We examine the question of which aspects of language are uniquely human and uniquely linguistic in light of recent arguments by Hauser, Chomsky, and Fitch that the only such aspect is syntactic recursion, the rest of language being either specific to humans but not to language (e.g., words and concepts) or not specific to humans (e.g., speech perception). We find this argument problematic. It ignores the many aspects of grammar that are not recursive, such as phonology, morphology, case, and agreement. It is inconsistent with the anatomy and neural control of the human vocal tract. And it is weakened by experiments showing that speech perception cannot be reduced to primate audition, that word learning cannot be reduced to fact learning, and that at least one gene involved in speech and language was evolutionarily selected in the human lineage but is not specific to recursion. The recursion-only claim, we suggest, is motivated by Chomsky’s recent approach to syntax, the Minimalist Program, which de-emphasizes the same aspects of language. The approach, however, is sufficiently problematic that it cannot be used to support claims about evolution. We contest other arguments from Chomsky that language is not an adaptation, namely that it is “perfect, ” nonredundant, unusable in any partial form, and badly designed for communication. The hypothesis that language is a complex adaptation for communication which evolved piecemeal avoids all these problems.

this article. PRINCIPLES UNDERLYING THE CONNECTIONIST APPROACH The connectionist approach instantiates a number of computational principles that are relevant to morphological processing (see Figure 2). We discuss #ve central ones in some detail because they are important for understanding the conditions under which the approach predicts morphological effects in the absence of semantic and/or phonological similarity (for additional background on principles of connectionist modelling, see Chauvin &Rumelhart, 1995; Hertz, Krogh, &Palmer, 1991; McClelland et al., 1986; Rumelhart, Hinton, & Williams, 1986a; Smolensky, Mozer, & Rumelhart, 1996)