LINGUIST List 2.801

Wed 20 Nov 1991

Disc: SE Asian Languages: Segments

Editor for this issue: <>

Directory

1. "Randy J. LaPolla", Re: 2.777 A Phonological Query
2. , Segment
3. David Stampe, 2.795 Segments and SE Asian
4. , Chinese, segmentless language?
5. Pierre Martin, 2.788 Are Segments Universal?
6. "Arthur S. Abramson", SE Asian Languages

Message 1: Re: 2.777 A Phonological Query

David Gil (RHLE813HAIFAUVM.bitnet) asked about the phonology of SE Asian
languages. The type of phonological analysis Gil mentions comes from the Chinese
philological tradition. Though the basic split is between initial and rhyme (or
rime), finer distinctions are also made when necessary. The syllable would then
include (in order) the initial, the medial (called the 'head' of the rhyme in
Chinese), the vowel (the 'belly' of the rhyme), and the coda (the 'tail' of the
rhyme).
As to the difficulty Gil has with communication in certain SE Asian areas, it
might correlate with the presence of tone systems in those languages. I
remember having similar difficulties when I first started learning Cantonese
(which has up to 11 tones depending on the dialect) many years ago.
--Randy LaPolla

Chinese secret languages typically break up syllables in ways which
do not seem to agree with the often-made claim that speakers cannot
go below the level of a syllable, even if they do not go down to
the level of the segment. Also, I have often heard claims that
Japanese speakers cannot segment utterances beyond moras. The
fact that Chinese is written syllabically and Japanese moraically
would suggest that perhaps these speaker perceptions are just
reflections of the writing system.

David Gil asks whether SE Asian languages have segments. These
languages represent a typological extreme at any level of analysis,
but they have abundant evidence of psychologically real segmentations
shorter than the word. In what follows I'm talking about Mon-Khmer
(including Vietnamese and Muong, and Khasi and Nicobarese, though
they're in Indian territory), and Daic, Chamic, and Chinese languages:
1 Measures or feet. The first division, of the finally-accented word
 into an accented ("major") syllable plus zero or more preceding
 unaccented ("minor") syllables, is widely accepted by SE Asianists on
 many kinds of evidence. (The widely used terms "major" and "minor"
 are Harry Shorto's, from his Word and syllable patterns in Palaung,
 BSOAS 23.544-577). The major syllable is bimoric (and tones
 typically contour tones, and nuclei typically diphthongal), and in
 Sinetic can even be bisyllabic (with the second syllable "minor").
 Minor syllables are "reduced" (they do not have their own tones,
 their consonants are minimal, and their nuclei are often just shwas
 or syllabic consonants). This is all typical of stress-timing, and
 so, not surprisingly, song and verse meters are often iso-accentual,
 not iso-syllabic or iso-metric. (This is from listening, not from
 analyses, but I recall seeing an article by Robbins Burling that
 recognized the similarity between Chinese and English folk meters.)
 Since Vietnamese and Muong dialects happen to have lost all minor
 syllables, you might think they were syllable-timed, if you'd never
 heard the Vietnamese pronunciation of other languages.
2 Syllables. As the above implies, syllables are less important than
 accentual measures in SE Asian languages. In level-tone languages, a
 game that reverses syllables reverses tones as well, but in SE Asian
 languages it doesn't: there, tones are mapped onto words or measures,
 not syllables. But I couldn't have stated the generalizations I did
 in part 1 without referring to syllables, and the usual sorts of
 evidence one cites for syllables, including the principles governing
 their internal structure, can be found in SE Asia languages as well
 as any other. Chinese and Nagari-derived writing systems provide
 additional evidence for syllables, and some song and verse meter
 systems do as well (cf. English, which has iso-syllabic as well as
 iso-accentual meters).
3 Onset vs rhyme. Most SE Asian languages, even Vietnamese and Muong,
 use matching rhymes as a coupling device in verse. The Chinese
 traditionally used rhyme tables as a way of classifying words. If
 there are infixes, as in Mon-Khmer, they go between onset and rhyme
 (even nonsensical ones, as in the language games Y.R. Chao wrote of,
 where e.g. mi + infix aik -> maici, see "Nonuniqueness", reprinted in
 the Joos reader, and the references to his Academia Sinica article on
 secret languages). Other morphological, prosodic, and phonological
 processes distinguish onset from rhyme in SE Asia, as anywhere else.
4 Phonemes. Speakers of SE Asian languages have no greater difficulty
 using alphabetic writing than speakers of other kinds of languages.
 Syllabic writing is practicable only with considerable modification in
 languages in which loss of unaccented vowels has left very complex
 syllable canons (e.g. the CCCCVC of Khasi, cf. the English CCCVCCCC).
 On the other hand, an alphabetic writing system that treats complex
 vocalic nuclei as sequences of vowels, or vowels and glides, is not
 appropriate for these languages, any more than it is for English.
 (Ask a speaker of an Indian language to say /ay/ backward, and the
 typical response is /ya/. Ask a Vietnamese or English speaker, and
 the typical response is /ay/, and a very funny look. You might as
 well ask them to say /a/ backward. The one good thing about the SPE
 analysis of English vowels was that it captured the fact that /ay/,
 /oy/, etc., are atomic phonemes for English speakers. On the other
 hand, it did that for the wrong reasons, and therefore missed other
 atomic phonemes like /ar/, /er/, etc. Ask an English speaker to say
 `barn' backward. The way I have written these, like Trager & Smith,
 wrongly suggests that they consist of simple vowel plus a glide.
 Email won't transmit a "tie" diacritic, but please read one in.) The
 fact that these are psychological atoms in SE Asia is confirmed by the
 way they act as units in affective phonology (cf. Gerard Diffloth's
 seminal BLS paper, sorry it's at the office), just as they do in
 morphology in English (divine/divinity, ride/rode, etc.). In real
 phonology (mental phonetics) they act exactly like any sequence, so
 /ay/ is probably no more or less likely to become [ash] in SE Asian
 languages or English (Ah feel fahn) than in, say, Hindi (Jah Hind).
 Consonants in SE Asian languages are shifty, like the vowels, and
 the shifts often result in tonal or registral differences, but those
 things aren't peculiar to SE Asia, or even to SE Asian type languages.
For an attempt to explain why SE Asian languages are the way they are,
and why S Asian languages are in most respects exactly opposite - from
phonetics to grammar and from verse to music - see P.J. Donegan & D.
Stampe, "Rhythm and the holistic organization of language structure",
in John Richardson et al., Parasession on the Interfaces of Phonology,
Morphology, and Syntax, Chicago: CLS, 1983. The article is terse and
speculative, but the bibliography includes the essential works on the
typologies and histories of these two areas, which are representatives
of the analytic and synthetic poles of the whole linguistic earth.

 It is not quite right to say that Chinese phonologists
 have not proposed units smaller than onset and rime. For
 example, in Y.R. Chao (1968), A Grammar of Spoken Chinese,
 and in many other works, the Chinese rime is divided into
 'initial', 'medial', and 'final' (or 'head', 'belly' and
 'tail'). Such units are, essentially, what one would call
 segments.
 It is possible that people like Y.R. Chao were
 influenced by segmental languages such as Greek, English,
 or Sanskrit, and wrongly extended the analysis of these
 languages to Chinese. It is also possible, however, that
 there is something real about segments (whether you call
 them phonemes or not). The test should not be solely based
 on how the native speaker feels. The reason is simple: we
 do not always feel what we are doing. For example, had
 vision studies solely depended on how we feel we see
 things, we would hardly have discovered, or believed, that
 we perceive all colors in terms of three primary colors.
 The fact that we do not feel we are doing something does
 not mean that we are not doing it. Indeed, had it been
 intuitively obvious how we manipulate sounds, it would
 have been a puzzle why the Greeks were the only people who
 invented/discovered alphabetic writing.
 More reliable evidence can be found. I will take a
 piece from language games. Here it is clear, I believe,
 that the rime cannot be the smallest unit. Consider the
 following game, which converts each syllable into two
 (from Y.R.Chao 1930, with diacritic adaptations, [ng]=velar
 nasal)
 (1) ma --> man ta
 sao --> san tao
 mai --> man tai
 mei --> men tei
 feng --> fen teng
 ing --> in ting
 ...
The simpliest analysis is stated in (2) and illustrated in
 (3)
 (2) a. Copy the syllable
 b. Chamge the first coda to [n]
 c. Change the second onset to [t]
(3) 2a 2b 2c
	ma --> ma ma --> man ma --> man ta
	sao --> sao sao --> san sao --> san tao
	mai --> mai mai --> man mai --> man tai
	mei --> mei mei --> men mei --> men tei
	feng --> feng feng --> en feng --> fen teng
	ing --> ing ing --> in ing --> in ting
Cf. Yip (1982) and Bao (1990), among others, for more
discussion of such cases.

I believe that distinctive features are the only ultimate basic
functional characteristics of phonological systems. Their structure
and their number vary from one language to another, which is why we
face different languages. As I see them, distinctive features consist
of phonic contrasts that ensure, in each language, the necessary
distinctions between monemes (or morphemes, if you prefer). Of course,
from the point of view of segmentation, these features can be analysed
as phonemes, accents, tones, mores, codas, etc. Segments (phonemes)
and supra-segments (syllables, prosodemes) are therefore specific, and
structurally different, bundles of distinctive features. No more, no
less. Accordingly, the phoneme (or segment) is not the basic
functional unit in phonology, although the structure it represents
seems to be present, in some form or another, in all known languages,
which is obviously not the case with prosodemes. Phonematics (phonemes)
and prosody (prosodemes) are indeed two different chapters of phonology,
but there is nothing surprising in the fact that, in languages, one of
those chapters will prevail here, and on the contrary, will tend to fade
out elsewhere. In some instances, it may even be more appropriate to
visualize the functional characteristics in purely syntagmatic terms.
That having been said, one should nevertheless be carefull with the use
of informant intuition as criteria for the identification
(=interpretation and definition, not the perception) of the basic units
of a language. While we cannot do without informants on functional
matters, the structure of the language is for the linguist to decide.
Segmentation, not the inner feeling of the language by the informants,
is the basis for the recognition of two categories of functional units
(phonemes and tones) in most languages of South East Asia.
Pierre Martin
Langues et linguistique
Universite Laval
Quebec, Canada
G1K 7P4

I am not on your list, but one of your members, Norman Miller, has forwarded
David Gil's query to me. I think that Gil's casual inspection of grammar books
written in Thailand, Vietnam and China--persumably written in a European
language--has misled him. Indeed, in Southeast and East Asian philological
tradition one speaks of syllable onsets and endings. This is important for the
tone languages of the area in that one can more efficiently state constraints
on the distribution of tones. In addition to this, however, there is awareness
of segmentation into vowels and consonants, as well as tones. (It is true that
in Thai tradition, for example, certain diphthongs and triphthongs are treated
simply as vowels.) Thai children learning to read work with primers that are
quite analytic, calling their attention to all possible syllable types in the
language, even nonsense syllables.That is, I really don't know why one would
speak of languages "with only onsets and rhymes." I am very willing to think
about phonetic and phonological proposals calling into question the psycho-
linguistic status of the segment in the production and perception of any speech
signal, but, whatever the ultimate truth of the matter, it is hard to believe
that language behavior is not universal in the basic aspects of the answer to
the question. Also, from my own lengthy times of residence in Pn Asia, I find
it hard to believe that any grammatical tradition will have much effect on the
speech behavior of ordinary people.
 As for the anecdotal evidence given, I am very skeptical. There are too
many sociocultural and psychological variables present in the various
situations. Also, the languages named as being hardest for a foreigner with
10-20 words to communicate in, note that these are all tone languages! If at a
tea-stall Gil uses the wrong tone and says, e.g., 'slow' instead of 'tea,' it
could cause at least momentary confusion. As for the point about the mainland
sprachbund, Gil should bear in mind that there are non-tonal languages on the
Pn Asian mainland, e.g., Malay.which, according to him, gave him no trouble. As
for Gil's point (C), let me remind him of the notorious trouble foreigners
claim to have when speaking French in France. Would his "cultural explanation"
cover this? I find this explanation hard to accept, and I wonder whether it was
offered half in jest.With some 37 years of contact with Thailsn, including long
stays adding up to a total of four years or more, I can only say that the Thai
are delighted with foreigners who try to learn about their language and
culture. One who makes an honest effort will get much help from the people.
Arthur S. Abramson
The University of Connecticut
abramsonuconnvm.bitnet