LINGUIST List 2.788

Thu 14 Nov 1991

Disc: Are Segments Universal?

Editor for this issue: <>


  1. H.Samual Wang (035, "segmentless languages"
  2. Eric Schiller, Re: 2.777 A Phonological Query
  3. Jacques Guy, David Gils' Query on Phonology
  4. Richard Ogden, RE: 2.777 A Phonological Query
  5. Jean-Francois Prunet, Re: 2.777 A Phonological Query
  6. Frank Anshen, Re: 2.777 A Phonological Query

Message 1: "segmentless languages"

Date: Wed, 13 Nov 91 19:51:40 -0600
From: H.Samual Wang (035 <>
Subject: "segmentless languages"
In response to David Gil's query about "segmentless languages", I would like
to share this research experience with you. I am a native speaker of Taiwanese
(the Minnan dialect spoken on Taiwan). Recently I worked in a Taiwanese
syllable structure project (which is part of a cross-linguistic syllable
structure project by Bruce Derwing of the Univ of Alberta, Canada). I tried to
have my subjects break the syllables into segments and recombine the segments,
only to find that most of my subjects could not manipulate segmentation of
syllables. (Although I worked with only three subjects, the frustration
already convinced me that this is the case.) This incidence was mentioned in a
paper presented at the XIIth international congress of phonetic sciences this
past august in France (reference: Bruce L. Derwing, Sook Whan Cho and H.
Samuel Wang, 1991, "A cross- linguistic experimental investigation of syllable
structure: Some preliminary results.") At present I am thinking toward the
direction that Taiwanese, and perhaps other Chinese dialects as well, is
segmented differently from Western languages, or perhaps it is "segmentless"
as suggested by David Gill. Traditionally the Chinese syllables are divided
into onset and rhyme, not without reasons. My personal experience in learning
English seems to speak for this. When I started to learn English (at grade 7),
I had a hard time trying to fight off the idea that the "n" in "not" is
different from the "n" in "can". The reason, I reckon, is that one is in the
onset and the other is in the rhyme, thus the notion that the syllable is
broken into onset and rhyme seems well-grounded. A cognitive explanation might
be suggested for such under-segmentation: because the Chinese syllable
structures are simple, and the number of possible syllables is rather limited
(mostly under 500), compared to those of Western languages, the native
speakers do not have to spend the effort to further segment them as the number
is cognitively manipulable. Probably because of this under-segmentation that
David found the Chinese people difficult to communicate with.
H. Samuel Wang <>
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Re: 2.777 A Phonological Query

Date: Tue, 12 Nov 91 19:35:09 CST
From: Eric Schiller <>
Subject: Re: 2.777 A Phonological Query
I have no similar experiences to report with Cambodians speaking
Khmer. In fact, the absorption of phonotactically inappropriate items
into the Khmer spoken in Chicago (e.g. MacDonalds) seems to progress
very smoothly. Alas, I no of no studies along these lines and am
unlikely to have time to investigate personally. I will, however,
try to be observant when I travel to Thailand in January for the
Pan-Asiatic conference. I don't know any Thai, and unless I bump
into some Khmer-speaking refugees will have to rely on Western
languages to get around.
Eric Schiller
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 3: David Gils' Query on Phonology

Date: Wed, 13 Nov 91 13:32:31 EST
From: Jacques Guy <>
Subject: David Gils' Query on Phonology
Gil's query brings two old memories to my mind.
The older one is Prof. Rygaloff's passing remark (when I was a student
of his at the Langues Orientales) that Chinese did not lend itself
gracefully to phonemic analysis. In those days I knew nothing of
phonemics and his comments went way above my head (and fellow
students'). If you do a phonemic analysis of Chinese, said Rygaloff,
you end up with just two vowels, and you could perhaps even end up with
one, pushing your analysis further. (The two vowels, I remember, were a
and shwa).
The second remembrance is from John Wellfield, then studying Japanese
history at ANU. He had, of course, learnt Japanese. "It is strange", he
used to say, "how easy Japanese is to pronounce, and yet the Chinese
and especially the Vietnamese students in my class had no end of
trouble. There was this poor Vietnamese who just couldn't say the
simplest word: "sakana" (fish) would come out as something like [sakp
There is an algorithm discovered 30 ago by a Soviet scientist, B.V.
Sukhotin, which, given a text in any alphabetical writing system, very
successfully separates its symbols into two classes, one of which
happens to contains all vowels, the other all consonants. The closer to
a phonemic representation the system is, the more accurate the
classification. Even on such gems as traditional English spelling,
Sukhotin's algorithm is remarkably successful. It yields perfect
results on such languages as Hungarian, despite its large number of
vowels, and on Georgian, despite its consonant clusters. I have tried
on Chinese (Guoyu, pinyin transcription, tones omitted); the results
were incredibly bad. There must be something very wrong indeed with
pinyin, for Sukhotin's is an extremely robust algorithm, or rather,
"strange" with Chinese phonology. In fact, I doubt that the
dichotomy consonant/vowel is much relevant to Chinese phonology.
On the other hand, I don't think the cultural explanation holds too
well. The same is said of Japanese (speak Japanese to a Japanese and he
won't understand you). I've had that experience, and I used to believe
it until recently in Singapore airport: I was in bad need of a trolley
and saw one just about to be relinquished by a group of Japanese; so I
darted off and asked if I could have it. Without turning his back to
look at me, one of them answered "yes" (doozo), then only turned, and I
looked shocked out of his senses. I surmised that, had he seen me
first, he would have assumed that whatever I would say would be in
English or German or failing that some obscure European language that
might just sound remotely like Japanese. The same happens with
Australians. They think my accent is Dutch or German, sometimes
Danish. But when they have been forewarned that I am French, or have
guessed it from my name, they most often comment on how strong a French
accent I have after so many years in Australia. Britons, strangely,
usually mistake me for English. So, to me, the cultural explanation can
explain everything, and therefore explains nothing.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 4: RE: 2.777 A Phonological Query

Date: Wed, 13 Nov 91 9:54 GMT
From: Richard Ogden <>
Subject: RE: 2.777 A Phonological Query
Re David Gil's query about phonological analyses with segments/onset+
rime and S E Asian languages:
part of this recapitualtes the old (and quite erroneous!) arguments which
developed in the early 1930's in response to Firthian prosodic phonological
analysis - namely that some languages were 'prosodic' and hence amenable
to the kind of anlysis which Firth was proposing and others were 'segmental'
and hence *not* amenable to the kind of analysis theat Firth was proposing.
What the Firthian analysts showed was that any language was amenable to their
kind of analysis. The kinds of claims Eugenie Henderson makes about Thai are
also to be found in Jack Carnochan's treatments of Hausa, Keith Sprigg's
analyses of Tibetan, Eileen Whitley's analyses of english and JOhn Kelly's
analyses of a number of Bantu languages.
All these analyses use notions of prosodic phonological categories having
relevance for parts of structures, rather than treating phonological
representations as unstructured or minimally bracketed strings.
The issue is one of methodological approach to the anlysis of sound systems
of languages rather than one of language typology. It makes no sense to
talk about languages having inventories of segments; this simply reflects
*one mode* of analysis for some specific purpose, eg in Pike's terms
'reducing languages to writing' -- not the 'Truth' about the language under
Despite some of the claims to be found in psycholinguistic literature
concerning the 'salience of segmental perception' none of the work which
purports to demonstrate this actually succeeds in doing so. However, there
is some very interesting work carried out on illiterate populations by
Morais and colleagues which should raise serious questions in the mids
of any linguist/phonologist who truly believe in the segment.
ONe thing we find perturbing is the anecdotal 'evidence' which David Gil
presents. Anecdotes may provide a starting point for work; in themselves
they have no value whatever for linguistic analysis.
John Local
Richard Ogden,
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 5: Re: 2.777 A Phonological Query

Date: Wed, 13 Nov 91 10:30:10 EST
From: Jean-Francois Prunet <R22534%UQAMpucc.PRINCETON.EDU>
Subject: Re: 2.777 A Phonological Query
David Gil asks for references about traditional analyses of the syllable in SE
Asian languages, and studies arguing that some languages (e.g. IE languages) ha
ve segments whereas others (e.g. Thai) have syllables.
The following reference is a whole study devoted to showing that SE Asian langu
ages do not have segments but only syllables. The author divides languages into
 microphonemic (e.g. IE) and macrophonemic (SE). It is a good case for this the
sis, though it is perhaps not as empiricically minded as it could have been:
 Cao Xuan Hao (1985) *Phonologie et linearite (Reflexions critiques sur les pos
tulats de la phonologie contemporaine)* SELAF (Societe d'etudes linguistiques e
t anthropologiques de France), 323 p.
 The adress of SELAF is 5, rue de Marseille, 75010 Paris, France.
 The book reads very well and is well-informed about various phonological scho
ols. I hope this helps.
 Jean-Francois Prunet
 Universite du Quebec a Montreal
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 6: Re: 2.777 A Phonological Query

Date: Wed, 13 Nov 91 13:05:24 EST
From: Frank Anshen <>
Subject: Re: 2.777 A Phonological Query
One obvious suggestion why it should be easier to pick up a bit of
understandable Indonesian is that Bhasa Indonesia developed from a
pidginized Malay used as a Lingua Franca, i.e., presumably modified in ways
that make minimal acquisition easy.
 Frank Anshen
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue