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.

this paper is to develop an alternative proposal whereby long-distance agreement is brokered via a correspondence-theoretic relation established between the participant segments. We term this approach Long-Distance Agreement through Correspondence or LDAC. A chief assertion of the LDAC proposal is that agreement is determined by Identity constraints which check feature matching in corresponding consonants, thereby obviating representations in which feature linkage skips over spans of neutral segments. Another key claim is that similarity plays a decisive role in identifying which segments stand in correspondence

Phonological structures and processes are determined by the functional principles of minimization of articulatory effort and maximization of perceptual contrast. We can solve many hitherto controversial issues if we are aware of the different roles of articulation and perception in phonology. Traditionally separate devices like the segment, spreading, licensing, underspecification, feature geometry, and OCP effects, are surface phenomena created by the interaction of more fundamental principles.

In this paper, I will be presenting some results bearing on a question about the basic architecture of Optimality Theory. This question was first framed by Prince and Smolensky (1993) in the quotation below: “Universal grammar must also provide a function Gen that admits the candidates to be evaluated.

Abstract not found

Abstract not found

Preprint version; to appear in Phonology, vol. 23, no. 1 In Hungarian, stems ending in a back vowel plus one or more neutral vowels show unusual behavior: for such stems, the otherwise-general process of vowel harmony is lexically idiosyncratic. Particular stems can take front suffixes, take back suffixes, or vacillate. Yet at a statistical level, the patterning among these stems is lawful: in the aggregate, they obey principles that relate the propensity to take back or front harmony to the height of the rightmost vowel and to the number of neutral vowels. We argue that this patterned statistical variation in the Hungarian lexicon is internalized by native speakers. Our evidence is that they replicate the pattern when they are asked to apply harmony to novel stems in a “wug ” test (Berko 1958). Our test results match quantitative data about the Hungarian lexicon, gathered with an automated Web search. We model the speakers’ knowledge and intuitions with a grammar based on the dual listing/generation model of Zuraw (2000), then show how the constraint rankings of this grammar can be learned by algorithm. *

Does the productive use of language stem from the manipulation of mental variables (e.g. "noun", "any consonant")? If linguistic constraints appeal to variables, rather than instances (e.g. "dog", "m"), then they should generalize to any representable novel instance, including instances that fall beyond the phonological space of a language. We test this prediction by investigating a constraint on the structure of Hebrew roots. Hebrew frequently exhibits geminates (e.g. ss) in its roots, but it strictly constraints their location: geminates are frequent at the end of the root (e.g. mss), but rare at its beginning (e.g. ssm). Symbolic accounts capture the ban on root-initial geminates as *XXY, where X and Y are variables that stand for any two distinct consonants. If the constraint on root structure appeals to the identity of abstract variables, then speakers should be able to extend it to root geminates with foreign phonemes, including phonemes with foreign feature values. We present findings from three experiments supporting this prediction. These results suggest that a complete account of linguistic processing must incorporate mechanisms for generalization outside the representational space of trained items. Mentally-represented variables would allow speakers to make such generalizations.

This paper proposes a generic feature geometry based lexical representation of phonological description using inheritance networks which has applications in various areas of speech technology. The feature geometry based lexicon model provides the linguistic knowledge for a generic lexicon tool which allows application-specific lexica to be generated. The relevance of this approach to interactive dialogue in multi-modal systems lies in the representation of parallel information which is structured with respect to temporal relations. This paper first presents the formal modelling of the feature geometry based lexical representation and then provides examples of specific output formats discussing in particular the temporal coordination of multilinear representations for speech technology applications.

iii ACKNOWLEDGMENTS I am very pleased to have had the encouragement and support of a committee of three linguists for whom I have the greatest respect and admiration: Mark Liberman, William Labov and Eugene Buckley. Each of them made my transition back to Penn pleasant after what seemed like a long absence. It was a great pleasure to have Mark Randolph both as an external reader and as a colleague at Motorola. Mark’s work at MIT a decade ago has served as an inspiration to me. Orhan Karaali made this dissertation possible in this millennium. As my manager for over two years at Motorola, Orhan insisted on making my dissertation a priority at work. Harry Bliss provided his voice to this project and our whole group is very grateful for his patience and cooperation. My colleagues at Motorola listened to my ideas and provided technical and theoretical assistance at every turn: Noel