LINGUIST List 6.254

Sun 19 Feb 1995

Sum: Consonant Insertion

Editor for this issue: <>

Directory

1. , Summary: C-insertion

Some weeks ago I posted a query as to whether anyone knew examples of
consonant insertion to separate impermissible vowel clusters, in which the
choice of consonant was NOT conditioned by its surrounding vowels. Thanks
who all those who responded. Their names and e-mail addresses appear with
their edited remarks. I would welcome further comments on anything brought
up in this summary.

REASON FOR THE QUERY:
My query was prompted by work I've been doing on the Czech verb system in
which, in certain forms from certain paradigms, a glide /j/ appears that is
not present in other forms. (Since most e-mail won't support Czech
diacritics, I'll spell all Czech forms phonetically. An apostrophe after a
consonant indicates its palatalization.) Typical of such verbs is the verb
[d'elat] "to do":

[d'ela:] "he/she/it does"
[d'elaji:] "they do"
[d'elaji:c] active participle (plural)

In Old Church Slavonic and in Russian, this /j/ appears in all persons of the
present tense in related paradigms (e.g., Russian [d'elajet] "to do" 3rd per.
sing.) and creates a VjV sequence, which historically contracted into a long
vowel in Czech (e.g., [d'ela:]). In most current phonological accounts of
West Slavic languages, such as Rubach (1993), this /j/ is considered to be
present underlyingly, but then needs to be deleted in almost all forms of
these verbs. This representation seemed too abstract to me, and when I
considered Czech phonotactic restrictions (see charts in Palkova (1994)) from
an autosegmental viewpoint, it occurred to me that the /j/ (where it does
survive) may simply be inserted to break up impermissible V clusters or to
prevent formation of overlong syllable nuclei.

The only sticking point in this was that /j/ often appears between non-high
Vs or Vs with which it shares no obvious features. As I found out from the
responses and from searching I've done since, the epenthesis of consonants
between non-homorganic vowels is not as uncommon as one might assume. So
far, the norm seems to be that Cs inserted at morpheme boundaries will be
glides (even between non-high Vs) unless there is a relic consonant that has
been retained, and sometimes analogically extended, in a given environment.
 In all cases I've found where the inserted C is a relic, that C is coronal.
 More data would be greatly welcome.

THE RESPONSES:

Albert Ortmann (ortmannsapir.ling.uni-duesseldorf.de) mentioned several
examples: English intrusive /r/, French t-epenthesis in interrogative forms,
Dutch insertion of /j/ and /v/ between non-high Vs morpheme internally (e.g.,
theater [te:'(j)a:tr]; zovals [zo:'(v)als] "like, as if"), and Swiss German
insertion of /n/ between certain hosts and clitics (e.g. groesser wie-n-i
"taller than-EP-I").

==================================================================
 Subj: vowel clusters
 From: SPSCOBmain.queen-margaret-college.ac.uk
 From: SPSCOBmain.queen-margaret-college.ac.uk
X-From: SPSCOBmain.queen-margaret-college.ac.uk (Dr James M Scobbie)

Some people might think that /r/ sandhi in English is of the type you are
looking for, but really /r/ is just the consonantal glide counterpart of
non-high vowels, so it's just like finding /j/ near /i/ or /w/ near /u/.

Note, however, that to avoid inserting /r/ in certain socially
stigmatised environments, many speakers insert glottal stop:

law[?]and order

instead of

law[r]and order

========================================================

 Subj: C-insertion
 From: SEEGMILLERapollo.montclair.edu
 (STEVE SEEGMILLER)

Karachay, a Turkic language, has a clear case of a consonant appearing under
certain circumstances involving what would otherwise be V-V sequences. The
plural suffix is -le or -la (depending on vowel harmony), so that at 'horse'
has the plural atla. However, if the plural suffix is itself followed by
another suffix beginning with a vowel, an -r- appears: atlari"m 'my horses'.
(the 1sg possessive suffix is -im/i"m/um/u"m). the -r- is clearly historical,
since the Karachay forms with -r- are similar to the ordinary Turkish form.
That is, the Turkish forms corresponding to the
Karachay ones given above are at, atlar, atlari"m. The thing I don't know
is whether there is motivation for postulating this -r- in the underlying
form of the plural suffix or not. There are some reasons to say yes and some
to say no, so I'm not sure.

==================================================================

 From: GLADNEYVMD.CSO.UIUC.EDU


I don't have references, but here are three E examples I have noticed:
 freebie, Sukie (a diminutive of Sue), and the L.A. Lakers.
Best wishes, Frank Y. Gladney.

========================================================

 From: KarlReinaol.com
 Subj: Intrusive glides

I don't know if this is what you are looking for, but Portuguese has [y]s and
[w]s that come between more open final and initial vowels. In some dialects
(continental and insular) 'na agua' is [n(schwa)+y+agwa]. When nasal vowels
are on both sides, the [y] is nasal. In many varieties (all?) 'no ano' is
[nu+w+(mid-central a)nu]. Sorry about the lack of symbols. Karl Reinhardt,
Dept. of Modern and Classical Languages, University of Houston.

========================================================

 From: ferrycimrs1.mnhn.fr
 To: JPKIRCHNERaol.com

For glide insertion in Eskimo language you can see : Jorgen Rischel, Topics
in West Greenlandic phonology, Akademisk forlag, Copenhagen, 1974. Insertion
of nasal velar or nasal uvular to prevent the fusion of two long vowels.
Best regards
Philippe Mennecier, Muse'e de l'Homme, Paris

========================================================

 Subj: Consonant insertion
 PICARDVAX2.CONCORDIA.CA (MARC PICARD)

I think you'll probably find find that most cases of consonant insertionare
the result of analogy. In French, for example, what should be CHANTE-IL 'is
he singing' is actually CHANTE-T-IL, the epenthetic /t/ having been
introduced in all first conjugation verbs through the influence of a final 3
sg final /t/ in the other conjugations, e.g. FINIT-IL, FAIT-IL, RECOIT-IL,
etc.

Also, because of final consonant deletion, words like ABRI 'shelter' and
DEBIT came to have an identical ending. However, because so many such forms
alternated with morphologically complex words ending in /t/, some original
vowel-final words acquired this consonant in composition so that, for
example, ABRITER came to replace ABRIER (which is still used in Canadian
French in the sense of 'to cover'). You'll also find cases like CLOU 'nail'
with derived forms either with or without /t/, e.g. CLOUER 'to nail' vs.
CLOUTE, CLOUTERIE, CLOUTIER. Although /t/ is the 'default' consonant, as it
were, you'll also find cases like JOLI, JOLIE 'pretty' yielding ENJOLIVER 'to
embellish'; you'll also find family names like JOLIET and JOLIVET.

I'm sure you can find out more about all this stuff in Mildred Pope's FROM
LATIN TO MODERN FRENCH.

Marc Picard

========================================================

(In this one I have edited out those of Mr. Alvarez's examples containing
characters that did not survive transmission. He also included an extensive
Spanish outline of the Guajiro language and people, which I'll make available
to those who request it. JK)

 Subj: C-insertion in Guajiro
 X-From: jalvarconicit.ve (Jose R. Alvarez)

In Guajiro, an Arawakan language spoken in Colombia and
Venezuela, all the six vowels show a contrast between short and long. Long
 vowels (written as double vowels in the practical orthography) and
diphthongs (written as two vowels) behave alike for the process of consonant
insertion to be described shortly. Whenever the situation arises where
 due to morphological concatenation we have a theme (complex or
monomorphemic) ending with a long vowel and a suffix beginning with a long
 vowel, an epenthetic [h] (written as j) or [w] (written as w) is inserted
to break up the impermissible vowel cluster. The choice of either [h] or [w]
is simply a matter of dialect differences, [h] being preferred by the
 Arribero speakers and [w] preferred by the Abajero speakers. The
 insertion of [h/w] to break up vowel clusters applies with no
conditioning by the surrounding vowels, that is, it is a general and very
productive process.

atpanaa+ee+chi -) atpanaaJeechi OR atpanaaWeechi
"it will be rabbit"

ke+kii+ee-shi -) kekiiJeeshi OR kekiiWeeshi
"he wants to have a (good) head"

a+ko'ojoo+ee+shi -) ako'oJeeshi OR ako'oWeeshi
"he wants/wanted to embrace"

Contrary to what these few examples may lead one to suspect, this epenthesis
in NOT conditioned by the presence of a long ee in the second half of the
cluster. It is simply an accident of Guajiro morphology that very common
suffixes have this long vowel (-ee FUTURE, -ee DESIDERATIVE, -eema
APPARENTATIVE, etc.). However, only [w] is used in epenthesis in certain
cases. This last case is particularly important in infinitive formation.
 A Guajiro infinitive may end with one of the six long vowels aa ee ii oo
uu uu [this last vowel is /uu/ with "umlaut" --JK] or with the sequence waa.
This duality in infinitive formation is directly related with the
phonological structure of the stem from which the infinitive is
constructed: if the final syllable of the stem is light, that is, if it
ends with a short vowel, then the final vowel is lengthened; if the final
syllable of the stem is heavy (that is, if it ends with a long vowel
 or a diphthong), then -aa is suffixed, an epenthetic [w] being added. The
infinitives ashakataa /a=shaka-ta-aa/ (0=get.off-TT-INF) "to get off" y
cheecheewaa /cheechee-aa/ (soft-INF) "to be soft" illustrate this
contrast.

========================================================

Thanks again to all who responded. I'd welcome more examples if anyone knows
of any.

James Kirchner
jpkirchneraol.com