LINGUIST List 7.1378

Sat Oct 5 1996

Sum: Consonant harmony

Editor for this issue: Ann Dizdar <>


  1. Adamantios Gafos, summary: consonant harmony

Message 1: summary: consonant harmony

Date: Wed, 02 Oct 1996 15:28:19 EDT
From: Adamantios Gafos <>
Subject: summary: consonant harmony
Dear Colleagues:

In March 1996, I sent out an inquiry for instances of what appear to
be long-distance assimilations affecting noncontiguous sequences of
consonants. I would like to thank the several people who have
responded to my query. Their responses follow (after the copy of the
original message below).

Based on their information and my own research I have proposed a
typology for the cases of consonant harmony involving coronal
consonants, that is, consonants articulated with the tongue tip-blade
(in my PhD dissertation titled "The Articulatory Basis of Locality in
Phonology", The Johns Hopkins University, August 1996; copies
available from the author).

Diamandis Gafos Department of Linguistics	Univ. of Massachusetts, Amherst

WARNING: The remainder of this message is long (approx. 700 lines)

	>>>>>>>>>>>>>>>>>>>>>ORIGINAL MESSAGE<<<<<<<<<<<<<<<<<<<<<<<

I am trying to compile a typology of the (perhaps) rather limited
phenomenon of consonant harmony, a process of assimilation which
affects sequences of non-contiguous consonants (in a similar fashion
to vowel harmony).

To give a simple example, Chumash has been reported to have had such a
process, causing a prefix apical fricative /s/ to alternate with a
laminal fricative /sh/ in the environment before, but not necessarily
immediately adjacent to, another laminal fricative or affricate in the
word. Vowels are transparent to this process, and hence consonant
harmony has a striking long-distance character.

I have so far found the following cases of languages with consonant

American Indian:				Other languages:
	Chilcotin (N. Athapaskan)		Kinyarwanda (Bantu)
	Tahltan (N. Athapaskan)			Moroccan Arabic (Semitic)
	Navajo (S. Athapaskan)			Tamazight/Ntifa (Berber)
	Kiowa Apache (S. Athapaskan)		Sanskrit (Indo-European)
	Chiricahua Apache (S. Athapaskan)
	Chumash (Hokan)	
	Tzeltal (Penutian Mayan)
	Southern Paiute (Uto-Aztecan)

There are also claims that consonant harmony is attested in Quechua
and Greenlandic Eskimo, but my preliminary search of some grammars did
not confirm these claims.

I am looking for any other cases of consonant harmony which you may
have seen. References to sources, especially those with some
articulatory and/or acoustic characterizations of the sounds, will be
most helpful. Please reply to me and I'll post a summary.

Diamandis Gafos				Department of Cognitive Science			The Johns Hopkins University
					Baltimore, MD 21218

	<<<<<<<<<<<<<<<<<<<<<<<< responses >>>>>>>>>>>>>>>>>>>>>>>>>>>>>


I'm not 100% sure if this is what you are looking for, but I recall a
brief mention of a process where rounding can spread to a consonant
over a vowel in Passamaquoddy, an algonquian (native american)
language. The main linguistic source that I know of is Philip
LeSourd's MIT dissertation, which was recently (1993) published by
Garland. I've forgotten now exactly when the dissertation was
completed -- 1988? Anyway, it's called "Accent and Syllable Structure
in Passamaquoddy".

Ah, here we go. I wrote a fairly long paper about Passamaquoddy from
the point of view of Optimality Theory, and I just found the section
where I talked about this in particular. The phenomenon is this:

Two nearly identical consonants when they surround an underlying schwa
will become identical and become a geminate, deleting the vowel which
separates them. So it doesn't *look* much like consonant harmony over
a vowel, because the vowel is gone on the surface. This is sounding
less and less like what you want.

Nevertheless, an example is /n-skkwi-.../ -> n-skwkwi-pn 'we
(du. exc.) vomit' (292) where the second  (schwa) deletes and the
two consonants surrounding take on the [+round] that is underlyingly
only associated with the second consonant. This also can happen in a
case like the following: /tot-cssi-w/ -> toc-csso 'he is dark'. If
you get ahold of LeSourd's book, this stuff is discussed on p. 285 and
forward, section 5.7.3.

 -Paul Hagstrom

- --------------------------------------------------------------------
From: kvthusc.HARVARD.EDU (Karl Teeter)

Dear Diamandis: It doesn't look as if you have gone very far in
checking out consonant harmony (LINGUIST 10,869). Among other things
you have not listed any of the languages I know which have it, and to
the best of my knowledge it is I who invented the term in my third
published paper in 1959: Consonant Harmony in Wiyot (with a note on
Cree), IJAL 25.41-43. I described it there and in my book The Wiyot
Language (UCPL 37). Languages I know which have it are Wiyot, Karuk,
Wishram, etc., and, in fact, consonant harmony appears to be a
Northwest Coast areal phenomenon, as Boasians used to use the term, a
phonological process which crosses genetic boundaries. Good idea to
study it further, it may be more extensive than you think (some
dialects of Igbo, perhaps)! Yours, kvt (=Karl V. Teeter, Professor of
Linguistics, Emeritus, Harvard University)

- ---------------------------------------------------------------------
From: (Glenn Ayres)

In (one dialect of) Ixil, a Mayan language, it is fairly common in CVC
roots for the first fricative or affricate to assimilate (across the V
of the root) to the second C, assuming it is also a fricative or
affricate. Is that the kind of thing you are interested in?

More information may be found in my book:

Glenn Ayres (1991?) La gramatica ixil. Antigua Guatemala: CIRMA.

Good luck with your research.

- --------------------------------------------------------------------
From: (Nicholas Ostler)

Latin -alis/-aris adjectival alternation might be an example.
Notionally, -aris is used only in roots with a preceding l



Unfortunately, the long-distance form seems to have broken down fairly
early in Latin's long history: hence fluvialis, legalis, laminalis

Another possible example is the noted Indo-European phonotactic
constraint against having two aspirated consonants in the same root.


Nicholas Ostler

- --------------------------------------------------------------------
From: (cpeust)

A phenomenon similar to what you mentioned can be found in Coptic, the
last stage of the extinct Egyptian language, written in Greek
characters. Here /s/ shifts to /sh/ (a palatal fricative) if
preceding another palatal in the same word, as in /shace/ (c = palatal
stop) "to speak" < egyptian sct (egyptian vowels are unknown as not
represented in hieroglyphic script, c is here to be taken as a
non-aspirated, perhaps ejective palatal stop, t is dropped in Coptic
in word-final position). This is only true, however, for the Sahidic
dialect, which is the most widespread; in Bohairic, the dialect of the
Nile Delta, for example, the same word is 'saci' with /s/ retained.

Carsten Peust
Seminar of Egyptology
- ----------------------------------------------------------------

From: Larry Trusk

Basque has a limited form of consonant harmony, in the form of
sibilant harmony. Basque has two contrasting voiceless fricatives: a
lamino-alveolar one, notated <z>, and an apico-alveolar one, notated
<s>. The first resembles English or American Spanish /s/; the second
resembles Castilian Spanish /s/, but is sometimes more retracted.
There are also the two corresponding affricates, notated <tz> and
<ts>. Basque also has a voiceless palato-alveolar fricative, notated
<x> and resembling English <sh>, and the corresponding affricate <tx>,
resembling English and Spanish <ch>, but these, like all "palatal"
consonants in Basque, have historically occupied a rather special
place in the system, and they do not participate in the sibilant
harmony: they can co-occur with any of the others.

The basic rule is that a lexical item can contain only laminals or
only apicals. Here are some typical items:

<zezen> `bull'
<izotz> `ice'
<zazpi> `seven'
<zortzi> `eight'
<ziztor> `type of sausage; icicle'
<zotz> `small stick'
<zurtz> `orphan'
<zuzen> `straight'
<zirtzil> `insignificant'
<zintzo> `honest'
<zozo> `blackbird'
<zorrotz> `sharp'

<sahats> `willow'
<saihets> `rib'
<sats> `ordure, manure'
<sits> `moth'
<itsusi> `ugly'
<itsatsi> `stick, adhere'
<saski> `basket'
<seska> `cane' (plant)
<seaska> `cradle' (a compound)
<isats> `broom' (plant)
<ostots> `thunder' (a compound)
<itsaso> `sea'

There are no words like *<itsuzi>, *<sezen>, *<zits>, *<zahats>.

There is a marked tendency (but not an absolute law) for non-harmonic
sibilants in compounds, derivatives, and loan words to undergo
assimilation, almost always to the apical. Examples:

<zin> `oath, truth' + <-etsi> `consider' > <sinetsi> `believe',
attested as <zinetsi> in some early texts.

Spanish <france's> `French' was borrowed as (attested) <frantzes>, but
the usual form today is <frantses>.

The word for `bramble' is everywhere <sasi> today, but it is attested
in the 17th century as <zarzi>. Since <rz> > <s> is a common change
in Basque, we probably have <zarzi> > *<zasi> > <sasi>, though the
second form is not attested and the development is not certain.

Spanish <sazo'n> `season', borrowed as <sazoi(n)> and still so in the
east, is <sasoi(n)> in the west.

A word for `mole' is <satsuri>, from <sagu> `mouse' and <zuri>
`white'; without assimilation, we would have expected *<satzuri>.

Spanish <solaz> `recreation' was borrowed as <solaz>, attested and
still used in the east, but the west has <solas> `conversation'.

Inflectional suffixes do not assimilate. So, the instrumental
case-ending <-z> is always so: <hots> `noise', instr. <hotsez>;
<itsaso> `sea', instr. <itsasoz>. Likewise, the gerund-forming suffix
<-tze> does not assimilate: <sartu> `enter', stem <sar->, forms gerund

If you should want to check this with any Basque-speakers you may have
handy, note that the <z>/<s> contrast and the <tz>/<ts> contrast have
recently been lost in the western third of the country (Vizcaya and
much of Guipu'zcoa); you will need a Basque from farther east.

There is a brief discussion of sibilant harmony in section 14.3 of the
following book:

Luis Michelena, _Fone'tica histo'rica vasca', San Sebastia'n:
Publicaciones del Seminario de Filologi'a Vasca "Julio de Orquijo".
First ed. 1961; 2nd ed. 1977; 3rd ed. 1985.

There will also be a brief treatment of it in sections 2.2.3 and 3.4
of my forthcoming book:

R. L. Trask, _The History of Basque_, London: Routledge. Forthcoming
November 1996.

Larry Trask
University of Sussex
Brighton BN1 9QH

- --------------------------------------------------------------
From: (Markus Hiller)

in my reply to your query i forgot to include a rather tricky
(i.e. interesting) one:

this one is sometimes referred to as ``lyman's law'' in japanese
rendaku and can be (roughly, not quite exactly) characterized as
consonant disharmony in d i s t i n c t i v e (!) v o i c i n g .

rendaku = in (native) compounds, an initial obstruent in the right
	 ``becomes'' voiced.

``lyman's law'' = if the right component contains an underlying voiced
		 obstruent a n y w h e r e , no rendaku
		 (itou/mester 1986).
	/ori/ + /kami/ ---> /origami/	``paper folding''
	/onna/ + /kotoba/ -X-> */onnagotoba/	``feminine speech''

this even holds if that obstruent is voiced because of
postnasal voicing (itou et al. 1995):
	/aka/ + /tombo/ -X-> */akadombo/	``red dragonfly''

markus hiller
(tuebingen, germany)

p.s.: references:

ITOU, junko and MESTER, armin, 1986. ``the phonology of voicing in japanese''.
 l i n g u i s t i c i n q u i r y 17, pp.49-73.

ITOU, junko; MESTER, armin and PADGETT, jaye, 1995(?). ``licensing and
 underspecification in optimality theory''(?). l i n g u i s t i c
 i n q u i r y 24.

- ------------------------------------------------------------------
From: (Carol Stoel-Gammon)

Although consonant harmony is relatively uncommon in adult languages,
it is fairly frequent in early child speech. You might want to look at
the following:

Stemberger, J.S. and Stoel-Gammon, C. (1991) The underspecification of
coronals: Evidence from language acquisition and speech errors. In C.
Paradis and J.F. Prunet (eds.) The Special Status of Coronals (pp.
181-199(. New York: Academic Press.

Stoel-Gammon, C. and Stemberger, J.S. (1994) Consonant harmony and
underspecification in child speech. In M. Yavas (ed.) First and Second
Language Phonology (pp. 81-105). San Diego: Singular Publishing Group.

Carol Stoel-Gammon, Ph.D.
Speech and Hearing Sciences
University of Washington

- --------------------------------------------------------------
From: ("")

Regarding your posting looking for consonant harmony, I have a paper
that appeared in Language vol. 70 ('Adjacency Parameters in
Phonology') which mentions other cases that you might be interested

- ----------------------------------------------------------------

I don't know if this phenomenon qualifies, as it is morphologically
triggered, but in some Algonquian languages (many Cree dialects for
sure) s>sh and t>ch (as in chip) when the diminutive suffix (ish or
ishish) is added to a noun: otaapaan > ochaapaanish(ish), vehicle >
small vehicle, and siipii > shiipiishish(ish), river > creek. In
Algonquian it is referred to as 'sound symbolism'. I could send
information and a reference if you would like.

The article which treats this topic is "Diminutive Consonant Symbolism
in Algonquian" by David Pentland in the Papers of the Sixth Algonquian
Conference, 1974, William Cowan, editor, National Museum of Man
Mercury Series, Canadian Ethnology Service Paper No. 23, National
Museum of Canada, Ottawa, pp.237-252. Pentland referes to a paper by
Johanna Nichols 1971 "diminutive Consonant symbolism in Western North
America in Language 47:826-848.

As the Mercury Series publication is somewhat obscure, I could send
you a photocopy of Pentland's article if you have difficulty obtaining
it. However I leave for the field tomorrow for three weeks and would
not be able to do anything about it until early May. Let me know.

I work with dialects of Cree-Montagnais-Naskapi and the consonant
harmony occurs in some but not all of the dialects. It is also
optional in the ones where it is used.

I also see a reference to consonant harmony in Wiyot by Karl Teeter,
1959, IJAL 25:41-43 and one by Mary Haas "Consonant symbolism in
northwestern California" in Earl H. Swanson, Jr., ed., Languages and
Cultures of western North America: essays in Honor of Svan S. Liljblad
(Pocatello: Idaho State University Press) 86-96.

Hope this all helps, mm

- -------------------------------------------------------------
From: (Allan Wechsler)

I have two marginal cases for you.

The first example is from the history of Warlpiri (Australian,
Pama-Nyungan, Southeast, "Ngarrka"). A no-longer-productive rule
fronted the retroflex stop to a pre-retroflexed alveolar flap. This
rule was suppressed, however, when the next coronal was retroflexed.
The result is a vague sort of harmony in the domain of retroflection,
similar in spirit to Sanskrit Rnati.

The other marginal case is the conflicting evidence for manner harmony
of some sort between C1 and C2 of the root in Proto-Indoeuropean.


- --------------------------------------------------------------
From: Torben Andersen

In Paeri, a Western Nilotic language of Sudan, there is consonant
harmony to the effect that interdentals are assimilated to
non-contiguous non-liquid alveolars, see section "2.6. Consonant
harmony" in my article "Consonant alternation in the verbal morphology
of Paeri" in "Afrika und Uebersee" 71: 63-113, 1988.

Torben Andersen
Department of Communication
Aalborg University
Langagervej 8
9220 Aalborg Oest

- ---------------------------------------------------------------
From: (Markus Hiller)

in linguist list posting 7-474 you asked about of consonant harmony, a
> process of assimilation which affects sequences of non-contiguous
> consonants (in a similar fashion to vowel harmony).

i have not found your summary in the list, so i assume you are
still looking for examples.

i have seen another consonant harmony quoted, namely one of gothic, in
some intro to ling. lecture script. unfortunately, no reference is
given, in particular, none for the reconstructed phonemic repres-
entations. ( ``g'' is for ipa gamma):

 1sg pres. 1sg past

 haita /hE:ta/ haihait /hEhE:t/ ``call''
 leta /le:ta/ lailot /lElo:t/ ``let''
 slepa /sle:pa/ saislep /sEsle:p/ ``sleep''
 faha /fa:ga/ faifah /fEfa:g/ ``catch''
 ' '
 fraisa /frE:sa/ faifrais /fEfrE:s/ ``try''

of course, the lecture script cites this as an example for
``reduplication''. but then, why isn't the /snV/- prefix in klamath
(which odden (1993: 263f. in phonology 8) calls ``vowel harmony'')
one? sure, there have been attested cases of total vowel harmonies
that cannot be analyzed as reduplication (e.g. kolami epentheses;
cit. by clements (1993: 132 in laks/rialland (eds.):
archit.d.repr.phonlg., paris: cnrs)), so the klamath prefixes are
vowel harmony alright. still, it is very artificial, then, to call the
gothic past tense prefix a ``reduplication'' instead of consonant
harmony (autosegmentalists might want to argue, gothic /E/ had no
specification besides its skeletal slot, thusly transparent to
consonant assimilation).

sorry, no reference for the gothic data, but at least the translit-
erated orthography should not be too hard to find somewhere else; and
that <ai> denotes some vowel commonly found in epentheses seems to be

if this is interesting enough, i hope this is still in time
for your summary.

c u


- ----------------------------------------------------------------------

From: kvthusc.HARVARD.EDU (Karl Teeter)

Dear Mr. Gafos: Consonant harmony may be more extensive than you
think! Among others you have not listed any of the languages I know
which have it, and it is I who invented the term, to the best of my
knowledge, in my third published paper in 1959: Consonant Harmony in
Wiyot (with a note on Cree) IJAL 25.41-43. It is described there and
in my book The Wiyot Language (UCPL 37). The phenomenon is present in
Wiyot, Karuk, Wishram, to my direct knowledge, and, in fact, I believe
it to be an Native American Northwest Coast AREAL feature (as Boasians
used the term, a phonological process which does not respect genetic
boundaries). As I recall, you may also find it in some dialects of
Igbo in Africa. Good idea to study it further! Yours, KVT (=Karl
V. Teeter, Professor of Linguistics, Emeritus, Harvard University)

- -------------------------------------------------------------------

Dear Diamandis,

Just read your query on consonant harmony. There is a consonant
harmony in Kalabari-Ijo (see Jenewari's Chapter on Ijoid in 'The
Niger-Congo Languages' book edited by John Bendor-Samuel (1989)).

There is an implosive harmony in which only implosives (voiced
bilabial and voiced alveolar) or plosives may occur in simple
morphemes. That is there is a kind of restriction in which the two
types of consonants do not combine in simple morphemes. Examples are
available in the Chapter I referred to above. I don't know of any
language that has this kind of consonant harmony. Kalabari also has
an ATR-harmony system (vowel harmony).

If you are interested in further communication on this and other
phenomena in phonology you are welcome.

Presently, I am about completing a Ph.D. dissertation on 'Tone and
Syntax in Kalabari at the University of the West Indies at Mona,
Jamaica. I am a native speaker of the language.

Otelemate G. Harry

- ------------------------------------------------------------------
From: (Vincent DeCaen)

curious business you noted....

there is something in semitics, but it involves long distance
dissimilation. the m- prefix, e.g., in derived stem participles in
the ancient Semitic language of Akkadian, is sensitive to the feature
labial, and m>n if the lexical root has a labial consonant. (makes
good sense in an autosegmental approach, coronal unmarked.)

best wishes,

Vincent DeCaen
Hebrew Syntax Encoding Initiative

c/o Near Eastern Studies
4 Bancroft Ave., 3d floor
University of Toronto
Toronto ON, M5S 1A1

- -----------------------------------------------------------------------
From: (Sharon Rose)


Hi - it seems you and I are interested in similar topics!

In Pat Shaw's article in the Special Status of Coronals book, she
lists a number of languages reported to have some form of either
phonological or morphological consonant harmony.

Here's one from her list which I have been working on, although I'm
not sure if you want to count it as consonant harmony as it's
triggered by a vowel:

In Harari (Ethio-Semitic), also known as Adare, there is a form of
consonant harmony triggered by a high front vowel suffix. The only
source I've found on this is Wolf Leslau's The Verb in Harari (1958)
UC Berkeley Press. Pat cites a CLA paper by Girma Halefom on the
topic, but Girma tells me he no longer has that paper - in any case, I
think his data came from Leslau's book. I've been trying to find an
Adare speaker, but so far no luck.

The suffix -i signalling 2nd singular feminine subject in
non-perfective forms causes palatalization of a final stem coronal. In
addition, other coronals in the word may be affected (the -i in other
Ethio-Semitic languages does not do this, affecting only final
coronals). The interesting thing about the Adare case is that stops
are palatalized, whereas other cases of consonant harmony are reported
to involve only fricatives (the s/sh alternation).


- ---------------------------------------------------------------
From: (Rick Mc Callister)

If you'll permit me to throw in a monkey wrench. Spanish has a
tendency toward consonant DISharmony, (more so in colloquial

Latin carcer Spanish c=E1rcel
Latin tortor Spanish t=F3rtola

- ---------------------------------------------------------------

Sanskrit has a nice case of retroflexion harmony: the dental nasal
becomes a retroflex nasal if retroflex s, or r (which is a retroflex,
too, in Skt) precedes anywhere in the word. The only segments which
block this rule are intervening [+coronal] segments.

Dorothy Disterheft
Linguistics Program
University of South Carolina

- -----------------------------------------------------------------
From: (Malcolm Ross)

There is Consonant harmony of a kind in Javanese and certain other
western Austronesian lgs. The easiest source to lay hands on is
probably R. Armin Mester (1988), Studies in Tier Structure, Garland,
New York. He refers to some of the older literature.

The phonologist who knows the most about the distribution of consonant
harmony in western Austronesian is Adrian Clynes. He lives in Brunei,
and probably does not see LINGUIST, but you can contact him via a
colleague at the following e-mail address:

I hope you will post a summary to LINGUIST: I would be interested to
read it.

Malcolm Ross
Senior Fellow
Department of Linguistics
Research School of Pacific and Asian Studies
Australian National University
PO Box 1428
Australia 2601

- ------------------------------------------------------------------

From: (Peter Daniels)

I don't know whether this counts as consonant harmony for you, but it
has been well known since Greenberg 1950 that no "root" in Arabic can
contain two homorganic consonants. A similar rule in Akkadian was
discovered by F. W. Geers some years earlier, that no root can contain
more than one emphatic consonant; I extended this in 1975 (unpub.)
that no root can contain two consonants sharing certain features of
place and articularion in Akkadian. Koskinen did a similar stydy of
Classical Hebrew; it is safe to say it holds in stronger or weaker
form fo rthe Semitic family (but not for the Afroasiatic phylum).

		<<<<<<<<<<<<<<<< EOF >>>>>>>>>>>>>>>>>>
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue