James P. Kirby University of Edinburgh, UK ].kirby@ed.ac.uk Vietnamese, the official language of Vietnam, is spoken natively by over seventy-five million people in Vietnam and greater Southeast Asia as well as by some two million overseas, predominantly in France, Australia, and the United States. The genetic affiliation of Vietnamese has been at times the subject of considerable debate (Diffloth 1992). Scholars such as Tabard (1838) maintained a relation to Chinese, while Maspero (1912), despite noting similarities to Mon-Khmer, argued for an affiliation with Tai. However, at least since the work of Haudricourt (1953), most scholars now agree that Vietnamese and related Vietic1 languages belong to the Mon-Khmer branch of the Austroasiatic family. It is important to make a distinction between 'literary Vietnamese', a prescriptive construct in which several orthographic distinctions are maintained in production, and the colloquial or standard speech of a given dialect region. This illustration describes the modern Hanoi dialect of Northern Vietnamese; segmental and tonal inventories, as well as lexicon, vary considerably between Vietnamese dialects, including those spoken in areas adjacent to Hanoi. The earliest systematic account of Vietnamese phonology was given by de Rhodes (1651), whose analysis is reflected in the modern orthography. Other important descriptions include those of Maspero (1912), Le Van Ly (1948), Emeneau (1951), and Thompson (1965). Vietnamese historical phonology has played an important role in the broader study of Southeast Asian diachrony (Barker 1966; Barker & Barker 1970; Ferlus 1975, 1982, 1992, 1996, 1997; Gregerson & Thomas 1976; Thompson 1976; Diffloth 1992), and has proven central to our understanding of the process of tonogenesis (Haudricourt 1954; Matisoff 1973; Gage 1985; Diffloth 1989; Alves 1995; Ferlus 1998, 2004; Thurgood 2002, 2007; Stebbins 2010). Indeed, much of the synchronic phonetic and phonological research on Northern Vietnamese has similarly focused on its tonal system. The work of Vu Thanh Phifdng (1981, 1982) provides a comprehensive overview, but experimental studies have also been undertaken by Andreev & Gordina (1957), Earle (1975), Han & Kim (1974), Hoang Cao Cifdng (1986), Seitz (1986), Nguyln Van Ldi & Edmondson (1998), Brunelle (2003, 2009ab), Pham (2001, 2003), Michaud (2004), Michaud, Vu Ngoc Tuan, Amelot & Roubeau (2006), Brunelle & Jannedy (2007), Brunelle, Nguyln Duy Difdng & Nguyln Khac Hung (2010), and Kirby (2010). Other aspects of Vietnamese phonetics and phonology have been addressed by Nguyln Bat Tuy (1949, 1959), Gordina (1960a, b, 1961, 1964), Han (1966), Doan Thien Thuat (1977), Gordina & Bystrov (1984), Ngo Thanh Nhan (1984), and Nguyln 1 The Vietic branch is sometimes referred to as Viet-Mildng, although this latter term is also used to refer exclusively to a sub-branch of Vietic containing Vietnamese and Mildng. See Diffloth (1992) and Hayes (1992) for further discussion. Journal of the International Phonetic Association (2011) 41/3 doi:10.1017/S0025100311000181 © International Phonetic Association 382 Journal of the International Phonetic Association Dinh Hoa (1997). Studies of Vietnamese dialectology include Cadiere (1902), Thompson (1959, 1965), Gordina (1963), Cao Xuan Hao (1978, 1986, 1988), Hoang Thi Chau (1989), Ferlus (1991, 1995, 1997), Alves & Nguyln Duy HifOng (1998 [2007]), Alves (2002 [2007]), Pham (2005), and Honda (2006). The recordings accompanying this illustration are of a 32-year-old male native of Hanoi. Consonants Initials Labial Labiodental Dental Alveolar Palatal Velar Glottal Plosive ß t ť1 ď k ? Nasal m n Ji í) Fricative f V s z x Y h Approximant w Lateral approximant 1 ßa1 ba 'three' ďa1 da 'banyan tree' ka1 ca 'mug' mal ma 'ghost' nal na 'custard apple' rjal Nga 'Russia' ta1 ta 'we, our' tha1 tha 'to forgive' la^ lá (existential copula) fal pha 'to brew' va-l vä 'and' wanl oan 'unjustly' sal xa 'far' zal da 'skin' \ísA ha 'river' tga1 cha 'father' jia-J nhä 'house' hwa1 hoa 'flower' xa4 khá 'rather' va-J gä 'chicken' IsA ä (question particle) The voiced plosives are canonically, but not consistently, realized as implosives. Initial It th/ are apico-dental [t th], lamino-alveolar [t th], or contiguous apico-dental lamino-alveolar ('denti-alveolar', Harris 2006), while /cfn 1/ are apico-alveolar. Some previous treatments such as that of Thompson (1965) recognize an unaspirated, unaffricated palatal stop Id. However, in the speech of many younger Vietnamese native speakers from Hanoi, such as that of the present consultant, this segment is consistently realized as an affricate [tg], a well-attested areal feature (Harris 2006). The tongue body contacts the alveolar or post-alveolar region during the production of both the palatal nasal [n] and the palatal affricate [tg] in initial position (Henderson 1965). While some varieties of Vietnamese maintain a distinction in the phonetic realizations of orthographic (tr-) and (ch-), these onsets are completely merged in modern Hanoi Vietnamese. The highly salient (and socially stigmatized) merger of III and Inl > IV, characteristic of the speech of many lower- and working-class Vietnamese in the Red River Delta, is sometimes consciously manipulated to humorous and/or pejorative effect in colloquial Hanoi speech, as in e.g. /nfwl/ nau 'brown' + /norjnyl/ nong 'hot' — 'hot coffee with milk' > [lfwl lorjnyl]. In syllable-initial position /p j r/ occur in a small number of foreign (mainly French) loans, e.g. [pan1] < panne 'breakdown', [ya1 ra1] < garage, [bi1 ja1] < billiard. For many speakers, however, /p/ is realized as [b/6] and Irl as [z]. James P. Kirby: Vietnamese (Hanoi Vietnamese) 383 Finals Hanoi Vietnamese licenses eight segments in coda position: three unreleased voiceless obstruents /p t kl ([pn t1 kn]), three nasals /m n q/, and two approximants /j w/.2 In final position /t nl are canonically alveolar, though it is not clear if they are chiefly laminal or apical. While the EGG study of Michaud (2004) found no evidence of glottalization accompanying unreleased final stops /p t kl, the laryngoscopy study of Edmondson et al. (2010) suggests that glottal reinforcement (in the sense of Esling, Fraser & Harris 2005) may not always be absent in this context. Velar fronting Although the phonetic realization of the stops /rj kl following /i e el have sometimes been described as palatal [ji c], they are actually pre-velar [rj] and [k], with no point of alveolar contact (Henderson 1965). The conditioning vowels tend to be shortened and centralized, and may be produced with a noticeable palatal offglide. kirjl Kinh 'Vietnamese' kerjl kinh 'channel' kerjl canh 'broth' sikl xich 'chain' sekl xich 'slanting' sekl sdch 'book' There do exist a few instances of true velars following lei, e.g. [se:rjvl] xeng 'shovel'. Labial-velar finals Following back rounded vowels /u o ol, the velar stops Ik rj/ are produced as doubly articulated labial-velars [kp rjm]. This articulation is sometimes accompanied by a visible puffing of the cheeks as air becomes trapped in the oral cavity. urjml ung 'tumor' orjml ông 'grandfather' arjrrrl ong 'bee' ukp1 Úc 'Australia' okp1 ôc 'snail' 3kp1 óc 'mind, brain' Note the differences betweer sukpl xúc 'to scoop' sup1 súp 'soup' horjihl hông 'hip' homl horn 'day' hakpj hoc 'to study' hapj hop 'to meet' sarjnyl song 'wave' S3ny| xóm 'hamlet' Whether these segments are transcribed as final approximants /j w/ or as semivowels /; o/ is largely a matter of analytic perspective. From a phonological standpoint, these segments may be regarded as approximants (consonants) on the grounds that they may not be followed by another consonant. However, these segments are articulated somewhat differently from the initial approximants, with a lesser degree of closure. 384 Journal of the International Phonetic Association Figure 1 Location of monophthong and diphthong centroids in a schematic F1-F2 space, based on acoustic analysis of the accompanying sound files. As with velar fronting, there are rare exceptions to the realization of final velars as labial-velar after back rounded vowels: compare e.g. [barjml] bong 'to come loose' with [ba:rj~l] boong 'deck (of ship)' (< French pont; Nguyen Bat Tuy 1949; Haudricourt 1952; Sampson 1969). Vowels Hanoi Vietnamese distinguishes nine vowel qualities /ieumiuoo/ and three falling diphthongs /ia nia ua/. Length is normally distinctive only in closed syllables and then only for the vowels /a/ and M, although there do exist a small number of lexical pairs which provide evidence for a length distinction between the vowels It a/ such as [se:rjvl] x