|
Article Excerpt
ABSTRACT. Various places of articulation, e.g. coronal, velar, labial, and laryngeal, have been proposed as being unmarked for consonants. This paper provides a description of the rhymal systems in the two major dialects of Vietnam, Hanoi and Saigon, and argues that in all Vietnamese dialects there is a placeless consonant. In the Hanoi dialect this consonant is realized as a coronal from Coronal Default. In the Saigon dialect the placeless consonant is determined by the quality and quantity of the preceding vowel: a coronal after a short front vowel; otherwise a velar. The feature-sharing between the vowel and final consonant is carried out through a constraint on the rhyme, the Syllable Weight Constraint, which also accounts for the centralization of certain short vowels found in all dialects. The paper, therefore, provides further evidence for the claim that markedness is relative and its implementation is dialect/language-particular. *
1. INTRODUCTION. One of the central issues in phonology is underspecification, namely which place of articulation is unmarked, and various places of articulation have been argued as being unmarked for consonants: coronal (e.g. Avery & Rice 1989, Paradis & Prunet 1991), velar (Trigo 1988), both coronal and velar (Rice 1996), labial (Hume & Tsedanelis 2002), and laryngeal (Lombardi 2002). It is often noted that the restricted positions for licensing of place of articulation of consonants are root-final, stem-final, syllable-final, and word-final (Ito 1989, Goldsmith 1990, Prince 1984, Rice 1992, Yip 1991).
This paper provides a detailed description of the co-occurrence restrictions between the vowel and a syllable-final consonant in two major dialects of Vietnam, those of Hanoi and Saigon, and argues for a placeless consonant in Vietnamese. It will show that in the Hanoi dialect three places of articulation are licensed: labial, dorsal, and unspecified. In the Saigon dialect there are only two such places: labial and unspecified. The unspecified place in the former dialect is realized as a coronal by Default. In the latter dialect it is determined by the quality and quantity of the preceding vowel: the consonant surfaces as a coronal if the vowel is front and short; otherwise it as a velar. Thus the paper provides more evidence for the claim that markedness is relative and that the implementation of markedness is language-particular.
The Hanoi dialect is well-represented in the literature (Nguyen 1985, Cu et al. 1977, Thompson 1965, Huu & Vu'o'ng 1980) and the data were further confirmed by speakers now resident in Toronto and Hanoi. The data for the Saigon dialect come from the literature (Thompson 1957, 1965; Nguyen 1967, Vu'o'ng 1975) and from an experimental study conducted in Toronto in 2001. They were also confirmed by people from Saigon and its surroundings (Ben Tre) who now live in Toronto.
Section 1 presents certain theoretical assumptions adopted in the discussion that follows. Section 2 introduces the issue of asymmetry in Vietnamese rhyme and comparative data from the two dialects. Section 3 provides a description of Hanoi rhyme and discusses the problem with /e/ and /a/ in terms of the distribution of final consonants. An account making use of the Syllable Weight Constraint is provided. Section 4 provides a description of Saigon rhyme and discusses the distributional problem. Section 5 is the conclusion.
2. THEORETICAL ASSUMPTIONS. Certain assumptions are made in this paper. According to the constricted-based theory for phonological representations of consonants and vowels (Clements & Hume 1995) features are privative. A feature represents a place of articulation in consonants and vowels and can be used to group sounds into natural classes: [labial] groups labial consonants and rounded vowels into a natural class; [coronal] groups coronal consonants and front vowels; and [dorsal] groups dorsal consonants and back vowels. Therefore, front vowels are [coronal], back vowels are [dorsal], and rounded vowels are [labial]. Central vowels are placeless (Clements 1991).
Second, the model is one in which contrasts are built up in a step-by-step fashion (Avery & Rice 1989). A feature is underspecified unless a minimal phonemic contrast exists, in which case the marked value is specified. In this model Coronal is the unmarked place with no dependent, Labial is the least marked with [peripheral] under the Place node, and Dorsal is the most marked with both [peripheral] and [dorsal] under the Place node (see Rice & Avery 1989, Rice 1994, Rice & Avery 1993 and Figure 1 below).
[FIGURE 1 OMITTED]
It has been argued that both coronals and velars pattern as if they were unspecified for place features, with both equally unmarked (Paradis & Prunet 1989, 1991, 1993, 1994; Yip 1991, Ghini 1995). The Default Variability Hypothesis (DVH) (Rice 1996) reconciles claims that both coronal and velar consonants are unspecified and share the same underlying representation, i.e. a place node with no underlying dependent as shown in Figure 1.
Given Figure 1, there are three possibilities for realizing an unmarked consonant: (i) default fill-in of the unmarked feature, creating a coronal; (ii) failure of fill-in of the unmarked feature, with interpretation of the node resulting in a velar; and (iii) assimilation to an adjacent node. It is important to note that the DVH distinguishes the term 'velar' from the term 'dorsal', a distinction this paper also recognizes. Dorsal is specified for Dorsal and has the fullest place structure as in Fig. 1e, while velar is a consonant that lacks a specified place feature in both the phonological and phonetic representations as in Figs. 1a and 1c. The paper will show that there are rhymal dorsals in the Hanoi dialect but rhymal velars in the Saigon dialect.
Syllable weight is an important focus of this paper: all Vietnamese dialects are sensitive to it for their phonetic realization. To represent segment length, early generative theory used a skeletal tier to arrange the segments in a word independently from the gestural tier. Consequently, a geminate tt is represented with two slots on the skeletal tier that map to t on the gestural tier. This theory is well established for geminates and long vowels in certain phonological processes. However, the skeletal tier faces a problem when it does not appear to differentiate between vowel and consonant in certain processes involving vowel lengthening (see Broslow 1995 for a comprehensive overview and discussion).
Later, the mora was proposed as a prosodic unit; this theory uses syllable weight instead of number of segments to describe and account for shape-dependent processes (Hyman 1985, Hock 1986, McCarthy & Prince 1986 among others). The moraic tier replaces the skeletal tier. Only segments that bear weight can occupy slots in the moraic tier. In moraic theory a short vowel usually contributes one mora while a long vowel contributes two moras. Depending on languages, a structure such as CVC can be regarded as either LIGHT or HEAVY. Thus moraic theory accounts for certain phenomena based on syllable types, i.e. light or heavy, that depend on their mora count. Figure 2 represents possible moraic structures in which CV and CVC are equivalent and are, therefore, CVV and CVVC. A short vowel (2a) or a short vowel and a final consonant (2b) is monomoraic; a long vowel or a diphthong (2c) is bimoraic. A long vowel with a final consonant is also bimoraic since the consonant is linked to the second mora of the long vowel by a rule of Adjunction-to-mora (Broslow 1995). Therefore, a heavy syllable, 2d, is equal to two light syllables, 2a and b. In Vietnamese weight contrast is found in some vowels; however, short vowels occur only in closed syllables. Thus, in Figure 2 CVC is light but CVV or CVVC is heavy. In Vietnamese VV can be either a long vowel or a diphthong. Final consonants are not affected in CVVC syllables.
[FIGURE 2 OMITTED]
In nonmoraic theory branching and nonbranching in the nucleus are used to distinguish light and heavy syllables. Ewen & van der Hulst (2001:134) claim that there are two types of languages. One type is rhyme-weight languages in which the nucleus plays no role in the distinction between light and heavy syllables. The other type is nucleus-weight languages in which branching nuclei are heavy and non-branching nuclei are light. Figure 3 represents the possibilities of nucleus-weight languages.
Using Figure 3 as a working model, this paper will show that Vietnamese is a nucleus-weight language in which a light rhyme, i.e. a monomoraic rhyme with a short vowel and a final consonant as in Fig. 2b, is the domain of feature sharing and centralization of the vowel.
[FIGURE 3 OMITTED]
3. THE RHYMAL DISTRIBUTION. As observed in the literature (Thompson 1965, Huynh 1996, Cao 1988, Vu'o'ng 1975), the most striking difference between the Hanoi and Southern dialects (from Hue southward to the southernmost area) is the distribution of coronals and velars in the rhymal position. However, there are scattered observations on this topic and the various local dialects have not been thoroughly investigated.
This paper presents the consonantal rhyme systems of the two major dialects, i.e. those of Hanoi (North) and Saigon (South), and focuses on two problems within these: (i) one dialect has rhymal coronal consonants in situations in which the other has rhymal velar consonants; and (ii) the asymmetry in the realization of coronal and velar consonants after vowels since certain vowels appear with both coronals and velars but other vowels can occur only with either coronals or velars.
The data from two dialects given in 1 illustrate these facts. Tones are not a focus in this paper. The tonal diacritics, therefore, are shown only in the orthography but not in the phonetic transcription.
(1) Comparative phonetic data on final consonants in Vietnamese dialects ORTHOGRAPHY HANOI SAIGON GLOSS a. dut [dit] [dik] 'be broken]' b. ot [[??]] [[??]] 'pepper' c. khat [xa:t] [xa:k] 'thirsty' d. mat [mat] [mak] 'eyes' e. het [he:t] [he:k] 'to scream' f. chuot [cu[??]t] [cu:k] 'mouse' g. dich [dic] [dit] 'target' h. lenh [len] [l[??]:n] 'order' i. khach [xac] [xat] 'guest' j. khac [xa:k] [xa:k] 'different' k. khac [xak] [xak] 'to engrave'
As 1 shows (and ignoring the correspondences involving vowels), coronals in the Hanoi dialect correspond to velars in the Saigon dialect (la-f) and palatals in the Hanoi dialect correspond to coronals in the Saigon dialect (1g, h, and i). In addition, both coronals and velars occur after [i], [[??]:], and [a] in the Saigon dialect (1a & g, b & h, d & i, respectively); however, only velars occur after the other vowels in the Saigon dialect, i.e. [a:], [e:], and [u:]. After the long [a:] and short [a], velars in Hanoi correspond to velars in the Saigon dialect (1j, k).
This paper will show that velars are specified for [dorsal] in the Hanoi dialect but result from phonetic implementation in the Saigon dialect; coronals result from Coronal Default in the Hanoi dialect, but result from spreading of [coronal] from short front vowels in the Saigon dialect.
4. THE HANOI DIALECT. This section presents a description of rhyme in the Hanoi dialect and discusses the asymmetrical problems with/a/and/e/as well as solutions.
4.1. DESCRIPTION OF RHYME IN THE HANOI DIALECT. The Hanoi dialect is generally considered to represent standard Vietnamese. All vowels are followed by the full range of final consonants: labial, coronal, and dorsal. It also displays a centralization phenomenon which is much discussed in the Vietnamese literature.
A Vietnamese syllable maximally has an onset, nucleus, coda, and a tone, and minimally has an onset, a nucleus, and a tone. A nucleus can be a single vowel, short or long, or a diphthong, i.e. an element that consists of two vocalic segments both of which occupy the nucleus. There is a single initial cluster Cw and there are no final clusters. The patterns of syllables in the Hanoi dialect are shown in 2.
(2) Patterns of syllables in the Hanoi dialect C(w)V(C) C(w)VV(C) (C in final position represents a consonant or glide, V represents a single vowel, and VV represents a long vowel or a diphthong.)
Only a limited number of segments occur syllable-finally and there is slight variation from dialect to dialect, e.g. there are no palatals in the Saigon dialect. In the Hanoi dialect, ten consonants and two glides occur in the coda position. The phonetic final inventory is shown in Table 1.
Phonetically, five stops [p, t, [sup.j]k, k, [k.sup.p]], five nasais [m, n, [sup.j]n, n, [n.sup.m]], and two glides [w, j] occur in the coda. (1) Phonemically, there are only three places of articulation finally: labial, coronal, and dorsal. Plain, palatalized, and labialized dorsals are allophones of the same phoneme in this position. In the literature and in Vietnamese orthography the palatalized velars are presented as plain velars 'c' and 'n', phonetically and phonemically, depending on the account (see Cao 1998 for an overview). These two plain palatals are treated as phonemes in the initial inventory in all accounts, as seen in Table 2.
The Hanoi dialect has nineteen phonem. [UNREADABLE IN ORIGINAL SOURCE.] --initial consonants, as in Table 2. (2) /p/does not occur initially nor do the glides/j/and/w/; however,/w/can be prevocalic but must follow an initial consonant other than a labial, i.e. *bw, *fw, *vw, *mw, except in some borrowed words, e.g. boa [bwa] 'garden pea' (Fr. petitpois) or 'tip' (Fr. pourbois), voan [vwa:n] 'veil' (Fr. voile), phuy [fwi] 'barrel' (Fr. fut). Some examples with non labial clusters are hoa/hwa/ 'flower', toan/twa:n/ 'mathematics', quen/kwen/ 'to forget'.
A full set of examples of initial consonants is given in 3. The following section will discuss the phonemic status of certain vowels and final consonants but here brackets are used for the transcription (except for initials, which are phonemic in the examples below).
(3) A full set of examples of initial consonants in the Hanoi dialect
ba [ba] 'three' phai [fa:j] 'right' ve [ve] 'to return' me [me] 'mother' tam [ta:m] 'eight' du [du] 'enough' tha [tha] 'to forgive' sa [sa] 'lemon grass' dai [za:j] 'be long' hat [ha:t] 'to sing' lam [lam] 'very' nam [nam] 'five' chi [chi]...
|
