[Lingtyp] Homoorganic vs. heteroorganic assymetry in nasal-plosive onset clusters

Thu Aug 14 06:09:45 UTC 2025

Thank you, Alex and Larry,  for this discussion of Ian’s question and the wonderful data you present, with so many interesting segments, quite apart from the syllable structures they anter into. I would just like to put Larry’s generous references to Haike’s and my book in some perspective. 

To bring out the contrast between Nick Clements’ (1990) sonority profile of the syllable with the (still widely quoted) symmetrical profile, I referred to the latter as the Dinosaur theory of the syllable when teaching, after  the Monty Python sketch in which someone reveals their new theory: ”All brontosauruses are thin at one end, much, much thicker in the middle, and then thin again at the far end.” Nick’s profile retrospectively accounted for Theo Vennemann’s Syllable Contact Law, with the preferred sonority drop in the transition from one syllable to the next. 

Needless to say, I will send Larry a copy of the fifth edition as soon as I’m back home. It now mentions T.S Eliot’s poem ‘The Hollow Men’ as the source of the bang-whimper phrase, with thanks to a postcard sent by a Cambridgedon whose name we could not decipher.

The issue of the potentially loose [s] in sCC clusters was fruitfully addressed by Anne Hermes et al. (2013)  for Italian ‘impure s’, where they present data showing a loose temporal integration of [s] with the following C(C) of the onset (eg spina), unlike what they find for other consonants in the first position of CC onsets (eg prima). I take take that to be an indication that besides tones, onset C’s can float rather than being associated to an onset node (as I suspect is the case in English, but they don’t present comparative data).

And what’s with [ŋ] and [ʔ]? Excluded from the coda in Apinayé, but the only possible ones in Konjo (Friberg & Friberg 1991)?

Clements, G.N.  (1990). The role of the sonority cycle in core syllabification. In J. Kingston & M.E. Beckman
(eds.), Papers in laboratory phonology I: Between the grammar and physics of speech.
Cambridge: Cambridge University Press. 283– 333.
Friberg, Timothy; Friberg, Barbara (1991). Notes on Konjo phonology. In J.N. Sneddon (ed.), Studies
in Sulawesi linguistics II. Jakarta: NUSA (Universitas Katolik Indonesia Atma Jaya). 71– 115.
Hermes, Anne, Doris Mücke, & Martine Grice. 2013. Gestural coordination of Italian word-initial clusters: The case of Italian “impure s”. Phonology 30. 1–25.  
Venneman, Theo (1972). On the theory of syllabic phonology. Linguistische Berichte 18: 1– 18.

________________________________________
From: Lingtyp <lingtyp-bounces at listserv.linguistlist.org> on behalf of Alex Francois via Lingtyp <lingtyp at listserv.linguistlist.org>
Sent: Wednesday, August 13, 2025 01:25
To: Larry M Hyman <hyman at berkeley.edu>
Cc: LingTyp <lingtyp at listserv.linguistlist.org>
Subject: Re: [Lingtyp] Homoorganic vs. heteroorganic assymetry in nasal-plosive onset clusters

Dear Larry,Thanks for your reaction.  I like your CCCCVC examples from Apinayé!> I wonder if you have ever been tempted to say that the first consonant of the word-initial CC cluster is not in the onset?Interesting suggestion.  I would need to see what would be the advantages & disadvantages of such an analysis, for Dorig and for Hiw (which may warrant different conclusions). Now let me think..._____The language Dorig has interesting rules of syllabification, that seem to argue in favour of a CCVC underlying analysis. Indeed, Dorig has a handful of prefixes of the form C(V)-, i.e. a consonant followed by an elidable vowel:  e.g. Perfect /m(ɛ)-/, or Irrealis /s(ɔ)-/ (see this other paper, p.4 ). The elidable vowel will surface or not, depending on the shape of its lexical host:If the lexeme already has the shape CCVC  - like /tβiɣ/ 'bury' -, then a prefixed form *C-CCVC would be ill-formed (a word starting with 3 consonants), so the underlying vowel must surface:/m(ɛ)-/ + /tβiɣ/   → mɛ-tβiɣ/s(ɔ)-/ + /tβiɣ/    → sɔ-tβiɣBut if the lexeme has a shape CVC (e.g. /tur/ 'stand'), then the prefix vowel does elide, because the resulting combination C-CVC becomes a well-formed syllable:/m(ɛ)-/ + /tur/   → m-tur  /s(ɔ)-/ + /tur/    → s-tur I see this as an indication that the CCVC template is operational in the phonological system of Dorig.______In the case of Hiw, I can think of 2 arguments favouring, I guess, the CCVC template:1) CC- allows no schwa epenthesis.  In fact there are some minimal pairs with vs. without schwa:/təg͡ʟɔɣə/ 'dirty' (3 syllables)/təg͡ʟɔɣ/ 'peace' (2 syllables)/tg͡ʟɔɣ/ 'throw:Pl' (1 syllable)Schwa is thus an underlying phoneme in Hiw -- contrary to many languages (like Kalam or Nen) where the epenthesis is predictable ~ allophonic.This could suggest that a word like /tg͡ʟɔɣ/ is genuinely ~ underlyingly a monosyllable CCVC.2) Some Hiw compounds involve a first element ending in -C#, followed by a second element starting in #CC-, thus yielding a sequence of 3 consonants -VC.CCV-.  See these examples (from F2010, p. 398 ) :[X]It seems to me that these heterosyllabic C.CC clusters are best explained by establishing an underlying syllabic template of the form CCVC.____Finally, there is the special case of /w/ in Hiw, with unexpected clusters like /wte, wnɔt, wg͡ʟʉ/...I once considered explaining this odd behaviour of /w/ with respect to the SSP by proposing that /w/ is extra-syllabic ~ extra-templatic, in line with some analyses that have been proposed for Eng /s/, as you mentioned.  If I had followed that path, then at least that phoneme /w/ would be analysed as external to the syllable onset.However, this did not seem an economical solution to me, because /w/ otherwise patterns like a C in the CCVC template (Note also the example of 'shake hands' above, /g͡ʟaβ.wsɔɣ/, where /w/ is the first C of the second syllable.). If /w/ were extra-templatic in a language that otherwise has CCVC syllables, we would expect words starting in /wCCV-/, right? -- reminiscent of Eng. spl- & spr- words.  Instead, /w/ behaves just like any C in the phonotactics of the language... except for sonority expectations.My proposal ( pp.412 ff. ) to account for the special behaviour of /w/ was to note that this segment patterns not like a glide, but like an obstruent (cf. the table with gray cells in my earlier email).   It is as though /w/ were really an underlying */ɣʷ/, a rounded velar fricative. This hypothesis would make sense in the geometry of a system characterised by four plosives /p t k kʷ/ and four nasals /m n ŋ ŋʷ/, but only three fricatives /β s ɣ __/. If we accept the idea that all occurrences of [w] in Hiw are in fact the surface realisation of an underlying */ɣʷ/, then a word like [wnɔt] 'parcel' becomes analysable again as a well-formed CC onset {obstruent + nasal}, underlying */ɣʷnɔt/, parallel to /sŋi/ 'snout'. Likewise, a form like [wte] 'small' becomes a sonority plateau {obstruent + obstruent} */ɣʷte/, parallel to other well-formed onset plateaus like /ptɔɣ/ or even /ɣtiɣ/. _______For these reasons, I would propose that CCVC is the underlying syllabic template for both languages:with no sonority restrictions whatsoever in the case of Dorigwith sonority restrictions in the case of Hiw  (+ some language-specific adjustments).I hope I'm making sense  :-)bestAlexAlex FrançoisLaTTiCe — CNRS — ENS–PSL — Sorbonne nouvelleAustralian National UniversityPersonal homepage_________________________________________---------- Forwarded message ---------From: Larry M Hyman <hyman at berkeley.edu>Date: Tue, 12 Aug 2025 at 21:33Subject: Re: [Lingtyp] Homoorganic vs. heteroorganic assymetry in nasal-plosive onset clustersTo: Alex Francois <alex.francois.cnrs at gmail.com>Cc: JOO Ian <joo at res.otaru-uc.ac.jp>, LingTyp <lingtyp at listserv.linguistlist.org>Thanks, Alex, this is really great stuff! I wonder if you have ever been tempted to say that the first consonant of the word-initial CC cluster is not in the onset? I am reminded of the various treatments of initial sC and final Cs, which superficially violate the Sonority Sequencing Principle, where the /s/ has been analyzed as part of the onset, as within the syllable at a higher supersyllable level, outside the syllable, or by saying that /s/ is either not more sonorous than the adajcent stop, rather is just more strident. There are some languages that have even CCC and CCCC word initial clusters where there consonant releases (very short vocalic transitions?) among them. Here is a summary I once prepared of Apinayé [apn] (Macro-Ge, Brazil) from  Burgess, Eunice & Patricia Ham. 1968. Multilevel conditioning of phoneme variants in Apinayé. Linguistics 41.5-18.Maximal Syllable Structure: CCCCVC, which mostly follows the SSP:[X]On Tue, Aug 12, 2025 at 3:14 AM Alex Francois <alex.francois.cnrs at gmail.com> wrote:Dear Ian, dear Larry,Thanks for this discussion.  The Bantu data is fascinating.While Oceanic languages tend to comply with the Sonority Sequencing Principle (SSP) in their phonotactics, some languages have gone rogue in this respect.Two languages of northern Vanuatu in particular, namely Dorig and Hiw, have followed processes of syncope; they've ended up with a prototypical CCVC syllabic template, and many tautosyllabic (in addition to heterosyllabic) consonant clusters.NB: The data below comes from the following article [available here]:François, Alexandre. 2010. Phonotactics and the prestopped velar lateral in Hiw: Resolving the ambiguity of a complex segment. Phonology 27 (3): 393-434.In Dorig [Gaua island, Banks group], most combinations are attested, whether homorganic or heterorganic, and whether SSP-compliant or not:e.g.  /kmaːr/ '2dual', /tᵐbɪŋ/ 'to shut', /ᵐbtɔt/ 'canoe pegs', /nti/ 'child', /mkɛ/ 'above', /rk͡pʷa/ 'woman', /ɣtam/ 'door', /wⁿdɛ/ 'pig'...[X](Grayed cells represent sonority reversals)Note, in passing, the prenasalized phonemes in /ᵐbtɔt/ 'canoe pegs', /ⁿdŋ͡mʷuɣ/ 'mosquito', /wⁿdɛ/ 'pig', /ŋⁿdɪr/ 'coconut crab'.Dorig does not seem to show any signs, whether synchronic or diachronic, of preferring homorganic over heterorganic clusters. (If it has any preference, it would be for heterorganic.)________Hiw [Torres Is.] also has some interesting clusters:stop+nasal:  /tnɪɣ/ 'very', /pne/ 'to sling on shoulder', /kŋʷa/ 'today', /kʷne/ 'smell', others:   /tg͡ʟɵt/ 'sweet', /kʷg͡ʟɪ/ 'dolphin', /kʷg͡ʟɵɣ/ 'wooden club', /mg͡ʟe/ 'wrath', /ŋʷg͡ʟewon/ 'bush',              /βti/ 'star', /wte/ 'small', /wnɔt/ 'parcel', /wg͡ʟɵn/ 'fetch'However, as wild as some clusters may seem, most do comply with the SSP. Thus while /tn-/ is licensed, */nt-/ is ill-formed in Hiw (whereas it's fine in Dorig).The relevance of the SSP in Hiw is visible from the light-gray empty cells in this table:[X]Hiw does present some sonority reversals (dark gray cells), but these can be explained by Hiw-specific rules, e.g. regarding the odd behavior of /w/ (comparable to the odd behaviour of /s/ in English clusters).  In terms of sonority, the complex segment /g͡ʟ/ behaves not like a plosive (a laterally-released stop?) but like a liquid (a prestopped lateral) --- which was the main thread of my 2010 paper.To come back to Ian's question on homorganicity, we may note that, while Hiw treats sonority as (mostly) relevant in forming its syllables, it does not show obvious restrictions regarding heterorganicity. Compare /kg͡ʟe/ 'scraps' with /tg͡ʟɵt/ 'sweet', /tnɪɣ/ 'very' with /pne/ 'carry on shoulder'.The only signs that Hiw may have a slight preference for homorganicity are diachronic, as certain sound changes involved assimilation in point-of-articulation (at least for coronals assimilating to a velar):'belly':  *toᵐbʷa-  > *təkʷa- > *tkʷa- > /kkʷa/     (cf. /təkʷe/ in neighboring Lo-Toga)'today':  *ⁿdamʷai   > *ʈəŋʷa   > *tŋʷa    > /kŋʷa/     (cf. /ʈəŋʷe/ in Lo-Toga)Yet this slight preference of Hiw has not turned into a synchronic rule of avoiding heterorganic clusters in general.bestAlexAlex FrançoisLaTTiCe — CNRS — ENS–PSL — Sorbonne nouvelleAustralian National UniversityPersonal homepage_________________________________________---------- Forwarded message ---------From: Larry M Hyman via Lingtyp <lingtyp at listserv.linguistlist.org>Date: Tue, 12 Aug 2025 at 08:00Subject: Re: [Lingtyp] Homoorganic vs. heteroorganic assymetry in nasal-plosive onset clustersTo: JOO Ian <joo at res.otaru-uc.ac.jp>Cc: list, typology <lingtyp at listserv.linguistlist.org>Dear Ian, I’ve waited a week to see if others might answer in more explanatory phonetic terms, but since you got me thinking, I thought I’d offer a few comments about the nasal-stop asymmetries you noted. Since you only presented labials and dental-alveolars, here is a more complete table of what we need to consider (in fact, more). “Better” Onsets “Worse” Onsets pmpnpŋ mpnpŋptmtntŋ mtntŋtkmknkŋ mknkŋk While your numbers aren’t large and some of the differences are not so great (e.g. 11 mt- vs. 15 tm-), I think you are right that heterorganic CN is a much more common onset than NC. I would relate this to the Sonority Sequencing Constraint whereby a better CC onset would be one where the first C is less sonorous (e.g. a stop), than the second C (e.g. a sonorant). Thus, pl- is a good onset, while lp- is less so (and much rarer). The mirror image generally holds for codas: -lp is better than -pl. As Gussenhoven and Jacobs (2005: 138) put it: “syllables prefer to start with a bang and end with a whimper”. While the generalization covers more than nasal+stop or stop+nasal, nasals do present a special situation. The African languages I work on generally do not have complex onsets, but there are many which have either prenasalized consonants, syllabic nasal + consonant, or both. Nasally released CN is less attested, although importantly in Gwari [gbr], a Benue-Congo language of Nigeria (see below). Whether to analyze homorganic NC  and CN as one vs. two segments is a question that commonly arises. Heterorganic NC and CN are (always?) two segments. Coming back to the asymmetry you point out, I think it also could be useful to look at the origin of the complex clusters. I assume that consonant clusters mostly come from syncope, e.g. the yer phenomena in Slavic and vowel weakening in the minor syllable of sesquisyllabic languages in Southeast Asia. In Hyman (1972),  I showed, ignoring vowel nasalization, the following sequence of changes: *CVNV > CNV > CV᷉. Gwari shows the second step with nasal release. (In doing my fieldwork back in 1970, the [a] of Cŋa was perceptually oral to me. Although it may be slightly nasalized, it certainly does not sound like closely related Nupe [nup], which has fully nasalized vowels instead of nasally related consonants, e.g. gã̀ ‘speak’.) Gwari  cf. Proto-Grassfields Bantuò-kŋǎ‘monkey’ *kánákŋā‘to fry’ *káŋ-igŋà‘to speak’ *ɣàm So the question is whether syncope would be equally likely to produce such onsets as tm- and mt- from *tVmV and *mVtV, respectively? Siva has already pointed out that *m would tend to undergo homorganic nasal assimilation, i.e. mtV > ntV. Another possibility is that the nasal would become syllabic, as is often the case in African languages.  Again there is a parallel with /l/. In a number of West African languages earlier CVlV alternates with ClV, which ultimately wins out (see Sande 2024 for a recent statement about this phenomenon in Kru languages). Idoma [idu], a Benue-Congo language of Nigeria, actually has Cl̩V, where the /l/ (--> [r] after coronals) is syllabic and tone-bearing (Abraham 1951/1967: 108, Hyman 1985: 49), e.g. pĺ̩-à ‘deceived’, ú-dŕ̩-ō ‘navel’. Interestingly, syllabic /l/ doesn’t appear after nasals. Instead, the nasal is syllabic and the liquid a lone (non-syllabic) onset consonant: m̩̀lɛ̀ ‘swallowed’. Preconsonantal syllabic nasals are “better” than post-consonantal liquids!I assume that your asymmetry holds of onsets in general, not just in word-initial position, where longer (and more unusual) consonant sequences would be more likely. However, it does not hold for heterosyllabic consonant clusters. Instead we find the reverse asymmetry when consonants meet across syllables (cf. Vennemann 1988): “Worse” Contacts “Better” Contactsp.mp.np.ŋ m.pn.pŋ.pt.mt.nt.ŋ m.tn.tŋ.tk.mk.nk.ŋ m.kn.kŋ.k  This again has to do with sonority: with CVN.CV we get Gussenhoven & Jacob’s coda whimper followed by an onset bang. Thanks for getting me thinking about this. Cited works: Abraham, R.C. 1951/1967. The Idoma language. London: University of London Press. Gussenhoven, Carlos & Haike Jacobs. 2005. Understanding phonology. 2nd Edition. London: Hodder Arnold. (I haven’t checked more recent editions). Hyman, Larry M. 1972. Nasals and nasalization in Kwa. Studies in African Linguistics 3.167-206. Hyman, Larry M. 1985. A theory of phonological weight. Dordrecht: Foris. (Reprinted with a new introduction with William R. Leben. Stanford, CSLI, 2003.) Sande, Hannah. 2024. Insertion or deletion? CVCV/CCV alternations in Kru languages. In Ji Yea Kim, Veronica Miatto, Andrija Petrović & Lori Repetti (eds.), Epenthesis and beyond: Recent approaches to insertion in phonology and its interfaces, 21–55. Berlin: Language Science Pres Vennemann, Theo. 1988. Preference laws for syllable structure and the explanation of sound change: With special reference to German, Germanic, Italian, and Latin. Berlin: Mouton de Gruyter.On Tue, Aug 5, 2025 at 2:48 AM JOO Ian via Lingtyp <lingtyp at listserv.linguistlist.org> wrote:Dear typologists, It has occurred to me that homoorganic nasal-plosive onset clusters (e.g. /mp-/) seem to be more common than homoorganic plosive-nasal onset clusters (e.g. /pm-/), whereas heteroorganic plosive-nasal clusters (e.g. /pn-/) are more common than heteroorganic nasal-plosive clusters (e.g. /np/-).For example, based on Phonotacticon 1.0, which is limited to Eurasia, the following number of lects have the following onset clusters: l  /pm-/ : 2l  /mp-/ : 12l  /pn-/ : 14l  /np-/ : 4 When looking at, say, /tn- nt- tm- mt-/, the pattern is the sameː l  /tn-/ ː 8l  /nt-/ ː 13l  /tm- / ː 15l  /mt-/ ː 11  What could explain this assymetry? From Otaru,Ian - - - - - - - - - - - - - - - - - - - - - - -朱　易安 JOO, IAN 准教授 Associate Professor 小樽商科大学 Otaru University of Commerce 🌐 ianjoo.github.io- - - - - - - - - - - - - - - - - - - - - - - _______________________________________________Lingtyp mailing listLingtyp at listserv.linguistlist.orghttps://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp-- Larry M. Hyman, Distinguished Professor of the Graduate School& Director, France-Berkeley Fund, University of California, Berkeleyhttps://linguistics.berkeley.edu/~hyman_______________________________________________Lingtyp mailing listLingtyp at listserv.linguistlist.orghttps://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp-- Larry M. Hyman, Distinguished Professor of the Graduate School& Director, France-Berkeley Fund, University of California, Berkeleyhttps://linguistics.berkeley.edu/~hyman
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 119097 bytes
Desc: image.png
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250814/dd676757/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 100005 bytes
Desc: image.png
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250814/dd676757/attachment-0005.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 282779 bytes
Desc: image.png
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250814/dd676757/attachment-0006.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 112882 bytes
Desc: image.png
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250814/dd676757/attachment-0007.png>