Georgian, a South Caucasian language belonging to the Kartvelian family, is characterised by the ability of its consonants to combine in extensive clusters. Among the possible combinations are a series of two-member clusters which are argued to behave phonologically as single segments (Tschenkeli 1958, Vogt 1958, 1971, Aronson 1982, 1991, Deprez 1988 and others). They are known as ‘harmonic’ clusters, because the laryngeal quality is constant across the cluster. Its two members are both voiced ([dg bg dγ bγ]), both aspirated ([thkh tshkh thχ tshχ]) or both ejective ([t'k' ts'k' p'k' t'q' ts'q']). They can occur either word-initially or in word-medial position. Harmonic clusters do not contrast with identical sequences of segments, except for sequences formed at the junction of two words. There is no evidence that across word boundaries harmonic clusters are derived by some sort of restructuring.

The purpose of the present study is to review the phonological arguments brought in the literature in favour of treating harmonic clusters as single segments, and to look for acoustic evidence that would motivate the distinction made between harmonic clusters behaving as single segments, on the one hand, and simple sequences of consonants, on the other hand. The study uses phonetic data to address the issue of phonological representation. If the difference between a harmonic cluster and a simple sequence of segments is present in the phonology, then it should ideally also be visible in the acoustic signal, for example in the presence or absence of a release burst, or in timing differences, as suggested by previous studies of complex vs. simple segments in various languages (Maddieson & Ladefoged 1989, Maddieson 1989, 1990). The results show that the treatment of Georgian harmonic clusters as complex segments is not supported by the acoustic data.

The paper is organised as follows: § 2 presents the phonological behaviour of consonant clusters in Georgian, § 3 reviews phonetic evidence for complex segments, and spells out the predictions made by the present study. The acoustic study is described in § 4, followed by the presentation and discussion of results in § 5. The conclusions and areas for further study are presented in § 6.

The present study proposes a new interpretation of the underlying distortion in APRAXIA OF SPEECH. Apraxia of speech, in its pure form, is the only neurolinguistic syndrome for which it can be argued that phonological structure is selectively distorted.

Apraxia of speech is a nosological entity in its own right which co-occurs with aphasia only occasionally. This…conviction rests on detailed descriptions of patients who have a severe and lasting disorder of speech production in the absence of any significant impairment of speech comprehension, reading or writing as well as of any significant paralysis or weakness of the speech musculature.

(Lebrun 1990: 380)

Based on the experimental investigation of poorly coarticulated speech of patients from two divergent languages (German and Xhosa) it is argued that apraxia of speech has to be seen as a defective implementation of phonological representations at the phonology–phonetics interface. We contend that phonological structure exhibits neither a homogeneously auditory pattern nor a motor pattern, but a complex encoding of sequences of speech sounds. Specifically, it is maintained that speech is encoded in the brain as a sequence of distinctive feature configurations. These configurations are specified with differing degrees of detail depending on the role the speech segments they underlie play in the phonological structure of a language. The transfer between phonological and phonetic representation encodes speech sounds as a sequence of vocal tract configurations. Like the distinctive feature representation, these configurations may be more or less specified. We argue that the severe and lasting disorders in speech production observed in apraxia of speech are caused by the distortion of this transfer between phonological and phonetic representation. The characteristic production deficits of apraxic patients are explained in terms of overspecification of phonetic representations.

A universal characteristic of speech is that utterances are generally broken down phonologically into smaller phrases which are marked by suprasegmental features such as intonational events and/or final lengthening. Moreover, phrases can be further divided into smaller-sized constituents. These constituents of varying size, or ‘prosodic units’, are typically characterised as performing the dual function of marking a unit of information and forming the domain of application of phonological rules. However, there is less agreement about how prosodic units are defined in generating an utterance. There are at least two different approaches (for a general review, see Shattuck-Hufnagel & Turk 1996). One approach posits that prosodic constituents are hierarchically organised and that prosodic constituents larger than a word are derived indirectly from the syntactic structure by referring to the edge of a maximal projection (Selkirk 1986), to the head–complement relation (Nespor & Vogel 1986) or to the c-command relation (Hayes 1989). This position, which I call the SYNTACTIC APPROACH, has been called the Prosodic Hierarchy theory, Prosodic Phonology or the Indirect Syntactic Approach (Selkirk 1984, 1986, Nespor & Vogel 1986, Hayes 1989).

The other position, which I call the INTONATIONAL APPROACH, also assumes a hierarchical prosodic structure, but defines the prosodic units larger than a word based on the surface phonetic form of an utterance by looking at suprasegmental features such as intonation and final lengthening (e.g. Beckman & Pierrehumbert 1986, Pierrehumbert & Beckman 1988, Jun 1993, Beckman 1996). Both approaches assume a prosodic hierarchy in which prosodic units are hierarchically organised and obey the Strict Layer Hypothesis (Selkirk 1984, 1986, Nespor & Vogel 1986; a prosodic unit of a given level of the hierarchy is composed of one or more units of the immediately lower prosodic unit, and is exhaustively contained in the superordinate unit of which it is a part). The prosodic units which are higher than a word, and which are commonly assumed by proponents of the syntactic approach, are the Phonological Phrase and the Intonation Phrase, while those assumed by the intonational approach are the Accentual Phrase, the Intermediate Phrase and the Intonation Phrase. The prosodic units below the Phonological Phrase, i.e. the Syllable, Foot and Prosodic Word, do not differ much in the two approaches, since these units have more fixed roles vis-à-vis syntax or intonation.

The intonational unit corresponding to the Phonological Phrase is the Intermediate Phrase in English (Beckman & Pierrehumbert 1986) or the Accentual Phrase in Korean (Jun 1993), in that these are the units immediately higher than a Word. The Phonological Phrase is defined based on the syntactic structure, but the intonational units are defined by intonational markers. The Intermediate Phrase in English is the domain of downstep, and is delimited by a phrase accent, H- or L-; the Accentual Phrase in standard (Seoul) Korean is demarcated by a phrase-final High tone. The next higher level, the Intonation Phrase, is much more similar in the two approaches. Even though the proponents of the syntactic approach define this level in terms of syntax (e.g. a sister node of a root sentence), they claim that this level is the domain of the intonational contour and is sensitive to semantic factors (Selkirk 1980, 1984, 1986, Nespor & Vogel 1986). In this paper, we will focus on the prosodic level corresponding to the Phonological Phrase.

The kinds of rhythmical variation I will examine in this paper can be observed in some double-accented Hungarian words, and can be exemplified by the changes that the stressing of the numeric compound ′tizen′három ‘thirteen’ may undergo when it is embedded in phrases. The stressing of this word may become tizen′három in the phrase ′pont tizen′három ‘exactly thirteen’ and ′ tizen′három in the phrase ′ tizen′három ′pont ‘thirteen points’. The two processes, however, are not quite symmetrical in their range of application; the first kind of change occurs in only a subset of the words affected by the second kind. The symbol ′ in the examples represents a pitch accent on the syllable whose orthographic form it precedes.

These changes are the Hungarian counterparts of the two kinds of rhythmical variation in English, manifested, for instance, by the different realisations of the word ′thir′teen, when it is embedded in phrases like ′thirteen ′points or ′just thir′teen. However, while rhythmical variation in English is extremely widespread, in Hungarian it is restricted to certain classes of words. Besides, rhythmical variation in English lacks the asymmetry between the two processes; both kinds of change affect the same set of words.

This article concentrates on rhythmical variation in Hungarian and on the implications of this variation for metrical theory.