From: Steve Parker <steve_parkergial.edu>
Subject: Responses for Consonant Cluster Typology
E-mail this message to a friend
Query for this summary posted in LINGUIST Issue: 20.3119
This is a follow up to a query I submitted on September 15, 2009, summarizing the responses that I have received to date. The topic is consonant cluster typology. Specifically, I asked for information and resources to help answer the question, within complex tautosyllabic onset clusters, which type of consonant sequence is cross-linguistically unmarked or preferred: obstruent plus liquid, such as /pl/, or obstruent plus glide, such as /tw/?
(1) Herb Stahlke of Ball State University wrote:
Interesting project. As you note, you will find complexities with the OL clusters. Some of the Idomoid languages appear to have such clusters, but they in fact count as disyllabic. The liquid bears a tone, high or low. Stem vowels in these languages can typically be high, mid, or low. With /l/ the tone shows up on the liquid. With /r/ there is a very short epenthetic vowel between O and L, and the vowel bears the tone.
(2) Shigeto Kawahara of Rutgers University wrote:
One case that I can offer is Japanese, although there are some complications. One thing that is clear is that Japanese never allows OL clusters. The pronunciation of English loanwords shows that Japanese speakers split up the cluster, like: try=> [torai]; cluster =>[kurasutaa]. Japanese on the other hand also has so-called 'yoo-on', which can either be analyzed as CG clusters or palatalized C. I do not know any good argument for one way or another. Bloch's original analysis (1950, Language) treats them as CG clusters. Vance (1987) 'An introduction to Japanese phonology' seems to agree with this position, when he introduces Hattori's theory (p. 64). Other theories treat yoo-on as palatalized consonants, but usually without arguments. At any rate, Japanese does not allow Cw clusters (or 'labialized Cs').
Another set of references that may be useful is the debate between Broselow and Finer (1992) and Eckman and Iverson (1992), both of which appeared in Second Language Research. They observe patterns of second language acquisition in terms of different types of clusters. If I remember correctly, Eckman and Iverson take the Clementian position and Broselow and Finer take the Steriadian position.
I realize that this Japanese case is not super clear-cut, but I hope it nevertheless helps.
[Later Shigeto wrote back again, with the following:]
OK, now about 'yoo-on' in Japanese--being curious myself, I've asked Osamu Fujimura whether he knows some arguments against one position or the other. He was of the opinion that yoo-on should be treated as clusters for the following reason. [j] cannot appear before front vowels in Japanese ([ja, ju, jo] vs. *[ji, je]). Neither can the yoo-on contrast: we know that before [i] the consonants are (perhaps phonetically) palatalized in the first place so that we do not have a contrast between [kja] and [ka]--we only have [kja]. If we treat yoo-on as clusters, then we can treat these two restrictions as one generalization 'the [j]-zero contrast cannot appear before front vowels'.
I am not sure if this argument is terribly persuasive because we can consider yoo-on as palatalized consonants ([+palatal]) and posit a constraint *[+palatal]-[-back], assuming that [j] is also [+palatal].
(3) Daniela Muller of the Université de Toulouse wrote:
In response to your query on LINGUIST List, I would like to draw your attention to sound change, since the direction it takes may help answer your questions. For example, /kla, gla, pla, bla/ change into /kja, gja, pja, bja/ in most Romance dialects and some Tai dialects. In Proto-Slavic /pja, bja/ change into /pla, bla/.
For a language with a wide variety of onset clusters, I suggest Greek (both Modern and Ancient). Any dictionary should give an impression of the permissible clusters in this language.
Lastly, the sonority scale itself, as you probably know, is a highly debated concept. The following two articles critically examine this issue:
Ohala, John J. & Haruko Kawasaki (1984), Prosodic phonology and phonetics. In Phonology Yearbook 1; 113-127.
Ohala, John J. & Haruko Kawasaki-Fukumori (1997), Alternatives to the sonority hierarchy for explaining segmental sequential constraints. In Stig Eliasson & Ernst Håkon Jahr (Eds.), Language And Its Ecology: Essays In Memory Of Einar Haugen. Trends in Linguistics. Studies and Monographs, Vol. 100. Berlin: Mouton de Gruyter; 343-365.
(4) Caren Brinkmann in Germany wrote:
The lexical database CELEX contains phonetic transcriptions and frequency information for very entry in English, Dutch, and German: http://www.ldc.upenn.edu/Catalog/CatalogEntry.jsp?catalogId=LDC96L14 Its coding can be quite daunting, so let me know if you need any help.
(5) Boyan Nikolaev of the University of Sofia wrote:
If you are also interested in cases where a sonorant, as r, l or m, is involved, I think I can provide a handful of examples from Bulgaria and other South Slav languages. By the way -and that I know purely by chance- in Swahili the word for 'a child' is mtoto.
Steve Parker again: many thanks to these colleagues for their helpful replies.
Linguistic Field(s): Phonology Typology