Theiling Online    Sitemap    Conlang Mailing List HQ    Attic   

Re: Most common consonant cluster types cross-linguistically

From:Eldin Raigmore <eldin_raigmore@...>
Date:Sunday, August 17, 2008, 17:37
On Sat, 16 Aug 2008 19:25:19 -0400, J. 'Mach' Wust
<j_mach_wust@...> wrote:
>On Sat, 16 Aug 2008 13:47:02 -0400, Eldin Raigmore wrote: >>If you count affricates as clusters, then homorganic stop+fricative (/ts/) >>and homorganic fricative+stop (/st/) are almost surely the most common. > >My impression is that these are not common at all except for the two samples >you just mentioned. I don't know any language that'd feature homorganic /fp/ >or /xk/. Not that I'd know a large number of languages, but my knowledge is >mostely restricted to Indogermanic languages, and these are rich in >consonant clusters. On the other hand, affricates such as /ts/ or /tS/ may >be fairly common, but I've learned that affricates such as /pf/ or /kx/ are >quite "exotic". > >In the clusters composed of a stop and a fricative, there might be a >preference for clusters that involve dental or alveolar sounds. That is to >say, my impression is that clusters such as /st/ /ft/ /xt/ /sp/ /sk/ /ts/ >/ps/ /ks/ are more common than for instance /xk/ /fp/ /fk/ /pf/ /px/. >-- >grüess >mach
Maybe you're right. Do you have a reference for any hard data? For instance, would the UPSID database allow one to find that information? On Sun, 17 Aug 2008 16:48:19 +0100, David McCann <david@...> wrote:
>The answers to this so far seem to be muddling several different >phenomena. You are obviously interested in clusters within a syllable, >while some people are talking about groups at syllable boundaries.
I'm not sure how obvious that was; but at least clearly we want to differentiate the places the clusters occur. I was able to find information -- or at least professional guesses -- about word-initial or word-final clusters, but not just "clusters" in general. The answers will be different for word-initial vs word-final vs word-internal clusters; and for onset-clusters vs coda-clusters; and, possibly, if there are enough languages allowing three-or-more consecutive consonants in a syllable, for onset-initial vs onset-final vs onset-internal, or coda-initial vs coda-final vs coda-internal, clusters.
>As for affricates like /ts/, these are not clusters
Maybe not.
>I don't know of any statistical studies,
Does anyone? I will try to find some.
> but my own observations suggest >1. The commonest initial clusters are stop+semivowel, followed by stop >+liquid. >2. The commonest final clusters are nasal+stop, followed by liquid+stop.
Laver (I am referring to "Principles of Phonetics" by John Laver, Professor of Phonetics at University of Edinburgh, one of the "Cambridge Textbooks in Linguistics", copyrighted and published 1994 by Cambridge University Press. Its ISBN is 0-521-45644-X and its Library of Congress number is P221.L293 1994) seems to suggest that nonhomorganic approximant+fricative, nonhomorganic approximant+stop, nonhomorganic fricative+approximant, and nonhomorganic stop+approximant, would be quite frequent. Here "approximant" could be read as "semivowel or liquid". I don't remember any differences he might have mentioned between initial clusters and final clusters. WALS.info, in the "syllable structure" chapter, suggest that there are many languages with (C)(C)V(C) syllable-structures in which the second consonant of the onset-cluster must come from a restricted set of fairly sonorous consonants, such as semivowels, liquids, or nasals; those languages would drive up the number of initial clusters that were stop+semivowel, fricative+semivowel, stop+liquid, fricative+liquid, fricative+nasal, and nasal+semivowel.
>Many languages have extrasyllabic consonants: initial and final clusters >that can only occur word- initially or finally. English can have a word >ending in /ŋθs/, but not an internal syllable; Dutch can have a word >beginning in /str/ but not an internal syllable. The unnerving initial >clusters in Polish and Georgian mostly include extrasyllabics, and are >word-initial only.
Interesting and relevant. Thanks.