LINGUIST List 6.176

Thu 09 Feb 1995

Sum: C gemination (syntactic)

Editor for this issue: <>


  1. Claude Boisson, Sum: C gemination (syntactic)

Message 1: Sum: C gemination (syntactic)

Date: Wed, 8 Feb 1995 21:20:52 +Sum: C gemination (syntactic)
From: Claude Boisson <>
Subject: Sum: C gemination (syntactic)

Content-Length: 10885

Summary of data on syntactic gemination of consonants

A couple of weeks ago I posted a query on what I termed "syntactic
 gemination", for which I got information from no fewer than 15 respondents.
 I am very grateful to them all. Here they are, listed in alphabetical order

List of the 15 respondents:
Prathima Christdas (
Vincent DeCaen (
Lance Eccles (
Maik Gibson (
Ralf Grosserhode (
Jacques Guy (j.guytrl.OZ.AU)
Marcia Haag (
Mark Robert Hale (
Bruce Nevin (bnevinLightStream.COM)
John Phillips (
Mari Siiroinen (SIIROINENcc.Helsinki.FI)
Norbert Strade (
Mark Verhijde (
Caroline R. Wiltshire (

The term "syntactic gemination" was not specific enough, as I had in mind
 only gemination at word boundaries, and not word-internal gemination at
 morpheme boundaries. Nevertheless I will mention such cases as have been
 pointed out by respondents. Besides, in my haste, I had forgotten to
 mention Classical Greek, which I had also taken into account in the
 preliminary version of my paper.

Here are the data:

A Syntactic gemination at word-boundary

A.1 - Italian
This is the "raddoppiamento sintattico" of Central and Southern Italian.
 Nevertheless the conditions under which this appears seem to vary
 considerably among the dialects (and the speakers ?). For instance, the
 example "a casa" [a 'kkasa], taken from Lepschy & Lepschy (1981) is not
 accepted in the dialect of an Italian colleague at my university.

A.2 & A.3 - Biblical Hebrew and Phenician
In these languages the definite article /ha/ triggers gemination of the
 initial consonant of the following word (except for certain consonants).
 For instance, in BH, we have /su:s/ (horse), but /ha ssu:s/ (the horse).
The same would seem to apply to Phenician (see the grammars by Segert, van
 den Branden), although we have only one epigraphic attestation in Punic.
The explanation is that a proto-article is reconstructed as */han/ or
 */hal/, so that */hal su:s/) */has su:s/, then reinterpreted as /ha
 ssu:s/. This hypothesis has been connected with the Arabic data : the final
 C of the article /?al/ is assimilated to the initial C of the following
 word (at least for the socalled "sun" Cs). For instance: /?as samak/ (the f

46or Arabic, Gibson adds this:
In Tunisian Arabic there is a clearer case ( is schwa):
shaaf-t 'saw-3SF' "She saw"
shaaf-tt-u 'saw-3SF-3SM' "She saw him"
The doubling is done to maintain syllable structure, but this is not the20
normal way. We would normally expect the elision of , to shaaf-t-u, but20
this does not happen in the 3S Feminine past.

A.4 - Classical Greek
In Attic Greek word-initial /r-/ goes to /rr-/ under certain conditions
 after a word ending in a short final V. In epic texts, or in other
 dialects, this gemination is extended to other sonants: /l-, m-, n-/. We
 even find /pp-/ attested in Boeotian

These four languages were the only instances known to me when I posted the
 query. Incidentally, note that they are different from the cases of Latin
 "hic" and "hoc", which were pronounced /hikk/ and /hokk/ before a word
 beginning with a V. The form /hokk/ is original (and /hikk/ analogically
 modelled after it), so that diachrony forces us to say that /hokk/ is
 simplified into /hok/ before a C-initial word.

Here are now the additional data kindly supplied by my respondents, which I
 paste freely.

A.5 - Finnish (Eccles, Siiroinen, Strade)
There is consonant gemination at word boundaries in certain cases in
 Finnish. It is morphologically conditioned though it is "syntactic".
 Several morphemes or forms trigger it: imperative 2nd person singular (ota
 'take'/ otas se 'take it'), allative case (annan sinulle 'I-give to-you' /
 annan sinullek kirjan 'I-give to-you a book'), most of the nouns ending in
 -e (kirje 'a letter'/kirjet tuli 'a letter arrived') and so on.
Other examples:
In some negative forms: *en mene sinne* (I don't go there), pronounced20
 *en menes sinne*
 negation +V-stem +there
 verb (go) 20
The same in imperative:* a"la" mene sinne* (don't go there!),20
pronounced: 20
 *a"la" menes sinne*
 imp. 2. sg.
Also in the so called "1. infinitive":
 *Ha"nen pita"isi tulla ta"nne* (he/she ought to20
come here), pron: *................tullat ta"nne* 20
 inf.1 here
The background for this gemination is the historical loss of a final
 consonant in the suffix-less verbal stem and in the mentioned infinitive
 form. This consonant was assimilated to a following consonant. While it
 disappeared without any trace in an end position or in front of a vowel,
 the gemination of the following consonant was retained. This feature isn't
 marked in orthography.

A.6 & A.7 - Tamil and Malayalam (Christdas, Wiltshire)
Tamil has gemination of word initial stops following words with some case
 markings (accusative at least). A brief account can be found in Christdas,
 Prathima (1987) "On constraining the power of lexical phonology: evidence
 from Tamil" in McDonough, J. and Plunkett (eds) Proceedings of NELS 17,
 volume 1:122-146.20
Syntactic gemination is also found in Malayalam, a closely related language.

A.8 - Celtic languages (Phillips)
The Celtic languages have "mutations", changes to the beginnings of words
 due to their syntactic environment. Types of mutations include prefixing of
 h or n to a vowel and voicing, devoicing, nasalising, etc., of consonants.
 One of the mutations in Old Irish was gemination. The Welsh spirant
 mutation is historically cognate with Irish gemination, e.g. ci "dog", but
 tri chi "three dogs", cath a chi "a cat and a dog", though gemination in
 Old Irish occurred in a much wider range of environments.
Some of the mutations in Breton are realised phonetically as gemination,
 though spelt otherwise.
On other cases of C-mutation, see the mention by Verhijede below (Fulla,
 Southern Paiute).

A.9 - The Kelantan dialect of Malay (Gil)
In the Kelantan dialect of Malay, agents of passive clauses (ie.
 "by"-phrases) are marked not with a preposition (as in standard Malay),
 but, rather, by gemination of the initial consonant.
CB: If I understand this correctly, the preposition has been ellipted?

B - Word-internal syntactic gemination

Word-internal syntactic gemination seems to be widespread, probably more
 than gemination at word-boundary. As pointed out by Nevin, in many
 languages, some phonotactic effects apply only in certain syntactically
 defined domains, for example, in roots or in verb stems, but not in
 affixes. Hale adds that the number of languages which show this process is
 quite large. For instance there are some Oceanic examples. At any rate,
 both Sanskrit and (if the meter is to be believed) and preClassical
 (Homeric) Greek show such processes.

Here are a few specific instances:

B.1 - Classical Greek
Word-internal /r/ is geminated to /rr/ after the augment or after a vowel in
 compounds, variably.

B.2 - Choctaw (Haag)
Choctaw (a Muskogean lg of N. America)has an inflectional form (for aspect
 marking) that involves deformation of the stem such that a medial consonant
 is geminated, or /y/ is inserted an geminated if there are not the
 requisite number of syllables. So we have falama `return' becoming
 fallaama `finally return' while ala `arrive' becomes ayyaala `finally arriv

B.2 - Sakao (Guy)
Sakao is a language spoken at Espiritu Santo,, in Vanuatu (formerly
New-Hebrides). When the direct object is incorporated in the verb, the
 initial C of the verb is geminated.
Examples, with sOn 3D to hunt/shoot with a bow, EnEs 3D fish:
mOsOn EnEs 3D he is fishing fish with a bow (now)
mOssOnEs 3D he fishes fish with a bow (generally), il peche a l'arc
 mV- 3eme p. sg. realis
 nEs (( EnEs with disappearance of the compulsory article V-,
 then simplification of the two n's inro one n)
 NB: O 3D open o (IPA "open o" in Pullum & Ladusaw, p. 117)
 E 3D open e (IPA epsilon)

B.3 - Biblical Hebrew
Gil adds this on Hebrew:
In Biblical Hebrew the 2nd binyan ("pi99el") is formed by reduplication of
 the 2nd root consonant, together with the appropriate choice of vowels.
 Now in most part, the binyan system is considered "derivational" and hence
 not, strictly speaking, syntactic; however, in some cases, the 2nd binyan
 is the "transitive" or "causitive" of the first, in which case gemination
 (plus vowel pattern) does have a syntactic function.

Let me end by a more theoretical note, quoting Verhijde:
Your question touches upon the fields of interaction between
 morphology-syntax and phonology (prosody). Now as far as I know, all
 C-gemination is in itself strictly phonological. Thus if I understand your
 question correctly, you wish to find out whether there are languages that
 appear have morphological/syntactic triggers for C-gemination. There is a
 huge bulk of material on Sandhi-effects, as for example the Italian RS case
 you mentioned in your query. Perhaps Ellen Kaisse (1985), _Connected
 Speech_ may be of some help. Now with respect to your query: I was thinking
 (from a phonological point of view) that C-gemination is really: share
 melody. Or in more abstract terms: SHARE [X]. If this is correct, then for
 example C-mutation (like in Fula, Southern Paiute and Celtic languages)
 under syntactic considerations may become very interesting for you.

Thans again to all of you who generously answered my question.

Claude Boisson
Universite Lumiere,
Lyon, France
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue