LINGUIST List 10.1872

Sun Dec 5 1999

Disc: What Exactly Are Allophones?

Editor for this issue: Karen Milligan <>


  1. Lotfi, On "Disc: What Exactly Are Allophones?"
  2. Andreas Gather, What exactly are allophones?

Message 1: On "Disc: What Exactly Are Allophones?"

Date: 4 Dec 1999 23:03:34 EDT
From: Lotfi <>
Subject: On "Disc: What Exactly Are Allophones?"

 I guess this theme has been the most popular one on the list for the
whole year (though most probably NOT the most important one). But why
is it so? I doubt it is (as somehow suggested) due to this linguistic
education of ours. It's simply because we linguists today have so
little in common to talk about. This divergence of interests can be
primarily a symptom of a crisis rather than the maturity of the
science of language. We've become so much involved in our own 'field
of specialty' that linguistics is already on its way to turn into a
private science. I think that's why Linguists have welcome allophones
so warmly: they come from the already scarce terminolgy we have
inherited from the 'wise' linguists of the past. Ironically enough,
and in want of a common language, scientists of *language* are
'babbling' (rather than talking) among themselves with the very rare
occasion of some good old words of the past (like 'phone', 'phoneme',
and 'allophone') heard from time to time: does it mean the tower of
Babel may be left unfinished another time? 

Ahmad R. Lotfi
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: What exactly are allophones?

Date: Sun, 5 Dec 1999 12:25:57 -0500 (EST)
From: Andreas Gather <>
Subject: What exactly are allophones?


	1. Allophones: Things or Relations?
	One could perhaps stress the difference between phones and allophones
	(and prevent 'abusive' uses of the terms) by pointing out that 'phone'
	is a
	non-relational term, whereas 'allophone' is a relational one (roughly
	comparable to common sense notions like 'man' as compared to 'husband'
	or 'girl' as compared to 'daughter'; thus, a simple statement like 'X is
	an allophone' should be as "logically incomplete" as 'X is a daughter'.
	This view perhaps comes close to what Dan Moonhawk had in mind when he
	considered allophones as relations, not things, although this kind of
	terminological opposition doesn't seem felicitous.
	2. Allophones, Assimilation and Coarticulation
	<< As for the issue of coarticulation vs. complementary distribution,
	the added notion of physiological inevitability strikes me as a red
	herring, if not a red blue-whale. There are plenty of traditional
	allophones (allophones as traditionally analyzed) which are in
	complementary distribution but
	do not result from coarticulation: e.g., the aspirated realization of
	English /p/ which occurs initially in stressed syllables, as opposed to
	the unreleased word-final variant which tends to close, say, _yep_ and
	_nope_. Thus, without arguing that this is an ironclad characterization,
	I would suggest that complementary distribution, assimilation, &
	coarticulation are a set of increasingly specific terms which seem to
	involve proper inclusion, although they may turn out not to be quite so
	neatly related. English /h/ and angma are notorious PHONEMES occurring
	in complementary distribution but obviously not synchronically
	conditioned by assmilation. >>
	This view is challenged by JOAQUIM BRANDAO DE CARVALHO who argues
	against Richard Janda's position concerning the nature of aspirated and
	non-aspirated stops in English:
	<< This is not true. If aspiration (as well as voice) is to be defined
	in terms of voice onset time within the C/V transition, it does result
	(CV-)coarticulation: C-voicelessness continues after the vowel release;
	the final plosive is not aspirated because there is no vowel at its
	and therefore no possibility of coarticulation.>>
	And JOAQUIM even seems to maintain a stronger thesis arguing along the
	following lines:
	<< I don't mean that all cases of complementary distribution imply
	coarticulation, though I'd like to suppose it. But, honestly, I confess
	I can't find a really good example of context-sensitive allophonic
	that could be described without reference to some sort of
	I agree with the skepticism RICHARD JANDA exhibits towards the position
	that assimilation should be considered a prerequisite for complementary
	distribution and with his remarks on the notion of 'physiological
	inevitability'. But I'm not so sure about the relation of proper
	inclusion he suggests to capture the relation between 'complementary
	distribution' (CD), 'assimilation' and 'coarticulation' (CA). Leaving
	aside the notion of 'assimilation' for the moment, the notions of CD and
	CA, as I understand them, refer to different levels of linguistic
	description, namely phonology and phonetics, respectively. And it is
	precisely this distinction, it seems to me, which is in the heart of
	much of the present debate about the relation between phonology and
	phonetics. Where exactly do we draw the dividing line, as far as a
	synchronic description of a language is concerned? 
	Of course, that type of allophony which - contra JOAQUIM BRANDAO - is
	not conditioned by assimilation or CA seems to be quite unproblematic.
	It is virtually ever assigned to the phonology of a language:
	Disagreeing with JOAQUIM, I think the case of aspirated and
	non-aspirated voiceless stops in English is one good example. In German,
	aspirated voiceless stops can also occur at the end of a word, without a
	following vowel, so that in German <Tat> 'deed' both occurrences of the
	phoneme /t/ are aspirated, even though aspiration of the first [t],
	followed by a vowel that bears the word-stress, normally is stronger
	than that of the second, word-final one. According to JOAQUIM's
	CA-thesis, aspiration in word-final position should not occur. And, by
	the way, how could a language like Hindi exploit the contrast between
	voiceless aspirated and voiceless unaspirated stops for phonological
	purposes, if this constrast were determined by CA and hence (supposedly)
	physiological inevitability? Contra JOAQUIM, I would also argue that
	Marc Picard's example of the distribution of tense and lax high vowels
	in Canadian French is another good example of CD which can not
	reasonably be traced back to CA or assimilation. In German tenseness or
	laxeness of vowels is a concomitant feature of phonemic vowel length.
	Short/lax as well as long/tense vowels both occur in closed as well as
	in open syllables. But there are other, possibly still clearer cases of
	CD which can hardly be considered the result of CA or assimilation. One
	case are vocalic and consonantal realizations of the phoneme /R/ in
	German, which are determined by syllable structure: consonantal
	allophones in the syllable onset and vocalic realizations in syllable
	nuclei or syllable codas. Another case is vowel length in Standard,
	Tuscan-based Italian: Leaving aside some aspects which are not relevant
	to the point I'd like to make, short and long vowels are distributed as
	follows: Vowels have short allophones before long consonants
	(consonantal length being phonemic in Italian) and long allophones
	before short consonants. Thus, using the terminology of derivational
	phonology for expository purposes, <fato> 'fate' is phonologically
	/fato/ and its phonetic representation is [fa:to], whereas <fatto>
	'fact' is underlyingly /fat:o/ and at the level of phonetic
	representation [fat:o]. Vowel lengthening in Italian is entirely
	determined by well-formedness conditions on syllable structure and not
	by assimilation or CA. If we tried to subsume any kind of CD under the
	notions of assimilation or CA, the latter notions would become so wide
	that they are at best of very limited use to phonetics as well as to
	Coming back to the dividing line between phonology and phonetics, I
	think that cases which originate in some kind of assimilation (or CA),
	but where the outcomes of these processes (at least after some time) are
	sufficiently perceptually different, are likewise uncontroversial: A
	case in point is the standard textbook example of the distribution of
	the voiceless palatal [�] and velar/uvular fricatives [x] in Standard
	High German as realizations of one phoneme. Another is the distribution
	of voiceless velar stops and alveopalatal affricates in the very early
	history of Romance languages (i.e. before the allophones got
	phonologized as a consequence of other developments in the languages) or
	in Canadian French. Once again, we obviously have to do with
	language-specific phenomena which should be
	handled within the phonology of the languages (though these phenomena
	seem to have purely phonetic origins, and may be, as pointed out by
	Richard Janda, the result of Ohalian 'exaggerations'). 
	But there are hundreds of phenomena which are mainly phonetically
	conditioned, due to CA or some other phonetically driven mechanisms,
	which are only more or less perceptible (or only measureable and not
	perceptible at all) and whose status is quite difficult to determine.
	One such example is the status of velar and less velar/more palatal
	"allophones" of English /k/ in <cow> and <key>, originally mentioned by
	Martin Salzmann at the beginning of the whole debate. Do phenomena like
	these have to be treated within the phonology of a language and hence be
	given the status of allophones? Or do they have to be assigned to the
	phonetic component? And are there different phonetic components,
	language-specific and universal? Or do we need graded notions of
	allophony, i.e. should we speak of allophones that are 'more
	phonological' (belonging to the phonology, the grammar proper, of the
	language) and allophones that are 'more phonetic' (belonging to
	language-specific or universal phonetics)? And how can we decide what is
	language-specific phonetics and what is phonology? 
	In fact, I personally would restrict the notions 'allophone' and CD to
	cases of phoneme variation within the phonology of a language. Or should
	we speak of allophones and CD when, for instance, we consider the fact
	that English, like virtually every language, has nasalized realizations
	of vocalic phonemes in the context of nasal consonants
	(e.g. in <man>)? This case of a nasalized vowel surely is very different
	from allophonic (or surface phonemic) nasalization as in AE <can't>, in
	French or in Portuguese. (I think work by Abigail Cohn on nasalization
	contains very important insights into the question of determining
	differences between phonetic and phonological nasalization.) Another
	notoriously difficult topic seems to be vowel lengthening, perhaps
	covering the whole range from clear allophony/CD-cases as in Italian
	(see above), to purely phonetic (universal?) cases like differences in
	vowel length before voiceless and voiced stops with such intermediate
	stages as allophonic vowel lengthening before voiced continuants in
	Standard French (given that appropriate prosodic conditions are
	Thus, I think it is reasonable to restrict notions like 'allophone' and
	CD to phonological, structural factors. Perhaps, one could do the same
	with the notion 'assimilation' which seems more appropriate to describe
	the structural, phonologically relevant aspects of a process where some
	features of a sound or one autosegment carry/carries over to another
	sound, whereas the notion of coarticulation focusses on the phonetic,
	physiological aspects of the same process or the origins of that
	Best regards
	Andreas Gather
	Ruhr-Universitaet Bochum
	Romanisches Seminar
	GB 8/133
	Universitaetsstr. 150
	D-44780 Bochum
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue