Three Dimensional Phonology





        About Jean-Roger Vergnaud


Jean-Roger Vergnaud has made major contributions to the theory of generative grammar, both in the area of syntax and in phonology. His current research projects include work for GRECO Communication Parlee, a cooperative research project on speech recognition in France; a second project on the acquisition of phonology, and a third project, Outils pour le development d' interfaces en langage naturel. He is also involved with the new university program in cognitive science which centers on those aspects of language that are of interest to neurological and psychological theory and about computer science.


Adapted from:


        About Morris Halle



   Adapted from:



During the early years, before MIT had a graduate program in linguistics, my work focused mainly on speech acoustics, a subject about which I knew very little when I arrived. I very quickly learned a lot about it thanks to various colleagues who were working in nearby offices and with whom I socialized a great deal. Not being married at the time, I would have dinner at various nearby restaurants, none of which was memorable for its cuisine, but all of which provided excellent settings for long discussions and hot exchanges. These discussions would continue after dinner in Building 20. The topics discussed ranged from cybernetics and information theory to technical problems in circuit theory, and the grounding I received in those areas has stood me in good stead for the rest of my career. There was a great variety in the backgrounds of the people. There were no computer scientists in the 1950s, but there were many electrical engineers, psychologists, neurophysiologists, and linguists. One of my frequent discussion partners was Ken Stevens, who was then in the Acoustics lab, and some of our discussions have resulted in joint publications, the first appearing in 1959 and the latest in 1991, with another planned for 1998.



 The fundamental features of this new theory will be best understood when we compare it to the theory of SPE (Sound Pattern of English), henceforth the ¡§Standard Theory¡¨ (Chomsky and Halle 1968). Within the Standard Theory, there are two levels of representation: the level of SYSTEMATIC PHONEMICS and the level of SYSTEMATIC PHONETICS. A representation at either one of these levels is a linear sequence of units (junctures and segments in the case of systematic phonemics), and representations at systematic phonemics are mapped onto representations at systematic phonetics through the application of phonological rules, which are essentially context-sensitive rules (in the technique sense); some of these rules apply cyclically. In the early seventies, it was point out that this model of phonological competence was in principle in capable of handling tonal phenomena commonly found in various African languages. William1971 and Goldsmith1973, 1974, 1976a, 1976b proposed to overcome these difficulties by introducing into the theory what turned out to be essentially an elaboration of Z. Harris¡¦s long components (Z. Harris, 1944,1951). According to Williams and Goldsmith, each representation is to be analyzed as a pair of autonomous sub-representations spliced together by simple principles. These two sub-representations can be called the ¡§tone tier¡¨ and the ¡§segment tier¡¨, respectively. Each tier is a linear sequence of units (tones in the case of the tone tier and junctures and segments unspecified for tone in the case of the segment tier) and can be affected independently by rules specific to it. This extension of the standard model has been elaborated and refined in the work by Goldsmith and others, and the set of principles that were developed there has come to be known under the name ¡§Autosegmental phonology¡¨. In provocative contributions, McCarthy (1976,1979) showed that the language and concepts of autosegmental phonology could help solve some outstanding problems in Semitic phonology and morphology. McCarthy proposed that new autonomous tiers be introduced, beside the tone tier. In particular, he demonstrated that, by splitting the segment tier of ¡§Standard¡¨ autosegmental phonology into three separate and autonomous tiers, the vowel tier, the consonant tier, and the ¡§syllabic skeleton¡¨, the basic structure of the paradigm of Semitic verbal stems could be accounted for in a simple and natural fashion. In addition, he suggested that each one of the previously defined tiers should be split in its turn into subtiers, corresponding to the different morphological units.


 The autosegmental approach has been extremely successful, and fruitful. In particular, it has led to significant discoveries in several sub-domains of phonology.


On the other hand, it is clear that there are some difficulties with this model, of a formal and conceptual kind, as well as of an empirical kind. It was the recognition of this fact that led the authors to propose and develop a different kind of extension of the Standard Theory, based essentially on work by Liberman (1975,1977), Prince(1975,1976,1977a,b) and Selkirk(1978a,b,1979,forthcoming) and also Vergnaud 1975, Halle and Vergnaud 1976,Vergnaud1977, and Sportiche1977. The authors extended the ¡§metrical treatment¡¨ beyond its original domain of definition by introducing an abstract notion of binary branching tree and by showing that the basic concepts of metrical theory could be elaborated and generalized, and at the same time combined with the models sketched in Halle and Vergnaud 1976, for example, in such a way as to yield illuminating accounts of such diverse phenomena as harmony, stress, or syllabic structure (Vergnaud and Halle 1979). A corollary result of this paper was to show that the metrical formalism allowed for an adequate characterization of these various aries in terms of separate and autonomous ¡§modes of representation¡¨. This consequence of the analysis must be taken as an indication of its-at least, partial-correctness, since the various domains that happen to fall under metrical theory are actually independent from each other. In this sense, Vergnaud and Halle (1979) points to a more radical revision of the SPE theory than was explicitly recognized there.


 The core of the theory is constituted by a linear sequence of slots-the skeleton. Each morpheme of the word is represented by a sequence of distinctive feature complexes, which we have called the MELODY. The phonetic content of each skeleton slot is specified by means of lines that link the slot to a distinctive feature complex in one or more morphemes. The Well-formedness Condition defines what constitutes a proper link between a slot in the skeleton and a segment in a morpheme melody. The lines that link the melody with skeleton define a plane. Thus, phonological representation of a word contains as many planes as there are morphemes in the world. The three dimensional phonology noted that skeleton slots do not simply succeed one another in linear order but have a special organization of their own. The most basic of these is the organization into syllables which consist of three distinct constituents-onset time and appendix- of which only the rime is obligatory. The constituent organization of the syllable is expressed not by means of boundary markers, but rather by means of trees on a separate plane anchored in the C and V slots of the skeleton. These trees provide a natural way for capturing the oft-noted equivalence between closed syllables and syllables with long vowels. The distinction between rime and appendix explains certain peculiarities in the sequential constraints on phoneme sequences as well as the differential treatment of certain closed syllables. Finally, the tree structures that express the constituent structure of syllables are formally identical with tree structure encountered in the ¡§metrical¡¨ treatment of other prosodic phenomena, most notably stress and vowel harmony. Moreover, constituents of the syllable-specially rimes-are the terminal elements in which the stress trees are anchored. Thus, in some languages, a word may exhibit a number of independent constituent organizations, each of which is anchored in the CV skeleton.


Adopted from Morris Halle and Vergnaud Jean-Roger.1980. Three Dimensional Phonology. Journal of Linguistic Research, 1980,1.1,83-105.                    





Recent Publications of Marris Halle:


Bibliography of Materials on the Morphology-Phonology Interface-- John Alderete, Rutgers University