LINGUIST List 5.931

Mon 29 Aug 1994

Sum: [tT] affricates, Russian "nine"

Editor for this issue: <>


  • Daniel Radzinski, Sum: [tT] affricates
  • , Russian "nine" -- summary of responses.

    Message 1: Sum: [tT] affricates

    Date: Sun, 28 Aug 1994 10:39:43 Sum: [tT] affricates
    From: Daniel Radzinski <>
    Subject: Sum: [tT] affricates

    Some time ago I posted the following query:

    >Hebrew, German and Italian have the stop [t], the fricative [s] and the >affricate [ts]. Greek has all of these and also the fricative [T] (i.e. >'theta'), but it does not have an affricate [tT]. (English, Castillian and >Arabic have [t], [s] and [T], but neither [ts] nor [tT]). Does anyone know of >languages having a [tT] affricate? More generally, [pT], [tT], [kT], etc. >affricates? (To clarify, by affricates I mean real affricates that pass tests >of single-segmenthood, not accidental adjacencies such as [p] [T] in the >English word "depth".) If there is such a language, does it distinguish between >such affricates and [ps], [ts], [ks], respectively?

    I have received kind responses from: (in alphabetical order)

    Brad Coon, Lance Eccles, Dirk Elzinga, Blaine Erickson, Alice Faber, Johannes Heinecke, John Kingston, John Koontz, John Lawler, Wen-Chao Li, Stavros Macrakis, Geoff Nathan, Marc Picard, Don Ringe and David Stampe

    I am grateful to all of these individuals.

    A number of them indicated that [ps], [ks], [pT] or [kT] are not affricates at all as their stops are not homorganic with their fricatives. Once this point has been clarified, we can focus only on [tT]. The answer is a clear "yes":

    1. The Athapaskan language Chipewyan (Lawler, Nathan, Picard, Ringe, Stampe) has all of [tT], [ts], [tS] and [tl] as single segments, each with ejective and aspirated variants. Other Athapaskan languages having (at least) [tT] include Tahltan (Kingston) and Kutchin (Picard). Some relevant references: Krauss & Golla (1981) Northern Athapaskan Languages (in Handbook of North American Indians) John Laver (1994): Principles of Phonetics. Fang-Kuei Li (1946): Chipewyan (in Linguistic Structures of Native America). Ian Maddieson: Sounds of the World's Languages.

    2. The Nuorese dialect of the Romance Sardinian (Eccles) has (or at least had) [tT]. E.g. [pratTa] - small square vs. [pratta] - money. References: Ugo Pellis (1933-4): 50 inchieste linguistiche in Sardegna (in BALI). Massimo Pittau (1972): Grammatica del sardo-nuorese: il piu conservativo dei parlari neolatini.

    3. The Montasik dialect of the Austronesian Acehnese (Lawler) has a phonemic [tT] ("s" in standard orthography). References: M. Durie (1985): A Grammar of Acehnese on the Basis of a Dialect of North Aceh. J. Lawler (1977): A Agrees with B in Achenese: A Problem for RG (in S&S 8).

    4. Many Chinese dialects (Li) have [tT]. Most of these are rural ones which have been documented only since the early eighties in specialist dialectology journals written in Chinese and not widely circulated.

    5. Several Salishan languages, particularly Saanich (Coon -- "reasonably certain") have [tT].

    6. Western dialects of the Uzo-Aztecan language Shoshoni (Elzinga) have the affricate [tT]. It is a single phonological unit and patterns with the other stops (p,t,k,kw) in lenition and mutation processes.

    7. Modern Greek (Macrakis) has [tT] and [pP], e.g. /matTeos/ (Matthew), /sapPo/ (Sappho), but these are accidental adjacencies, i.e. not single segments.

    I have found an answer to my question. Thanks again to all contributors.

    Daniel Radzinski Tovna Translation Machines Jerusalem, Israel

    Message 2: Russian "nine" -- summary of responses.

    Date: Sun, 28 Aug 1994 18:02:21 Russian "nine" -- summary of responses.
    From: <>
    Subject: Russian "nine" -- summary of responses.

    I asked after the etymology of the Russian word for "nine", which is (in transcription) {dev'at'}. I was curious about how this could be derived from the Indoeuropean, more faithfully preserved in Latin /nowem/. Many thanks to all who responded. Almost all respondents gave some version of the same answer, which I will summarize for the curious.

    {dev'at'} is indeed from the IE, but the initial n- has been levelled to d- in analogy with the word for "ten", {des'at'}. (In the Germanic languages, a similar thing happened to "four" and "five"; if "four" had evolved "naturally", it would now be *"whour". One respondent also noted similar "crosstalk" in some Japanese speakers between the words for 7 and 8.) The analogy with "ten" might have occurred in Proto-Balto-Slavic, since Lithuanian has {devyni}, {deSimt}. But here the evidence is mixed, since Old Prussian (Baltic) retains n.

    The IE for "nine" was perhaps *newm (although one respondent says there is evidence for *Hnewn, where the H is the first laryngeal). The -tI suffix that appears on 9 and 10 in Old Slavic is apparently an ordinal suffix plus a declension marker that turns the word back into a noun. (The result might be glossed as something like "ninthness".)

    A couple of repondents corrected my misimpression about "eight" {vos'em'}. It too may be derived in regular fashion from IE.

    One informant referred me to Jadranka Gvozdanovic, ed., The Indo-European Numerals, Mouton de Gruyter, Berlin, 1992.

    I have not associated respondents with their individual contributions, to avoid inadvertently ascribing errors in my own paraphrases to them. Thanks again to all of them. (I fear I have deleted one name irrecoverably with an editing error; my apologies to the uncredited respondent.)

    "Larry Trask" <> "Evan S. Smith" <> Mike Hammond <> (Nick Ostler) (Schneider-Zioga) robert westmoreland <> "Margaret E. Winters" <GA3704SIUCVMB.SIU.EDU> (Johanna Nichols) Don Ringe <> Ivan A Derzhanski <> Bernard Comrie <>