[Lingtyp] Universal constraints on lexicalisation

Sat Feb 1 21:47:15 UTC 2025

Dear all – First, a quick comment on kinship nomenclature, and then let me briefly discuss another hypothetical constraint on colexification.

I’ve been teaching Ling Anth for 20 years, and I always spend a class or two on Morgan’s classification and its influence on the history of cognitive anthropology (chiefly via Lounsbury 1964, from which Rosch indirectly got her notion of ‘prototype’, mediated via Berlin & Kay 1969).

But all this time, I’ve had my doubts about the validity of Morgan’s classification. Perfect exponents of the six types are fairly rare; I’m pretty sure (though can’t quantitatively verify this) that there are more imperfect than perfect instances, certainly across the six types, but maybe even for every single type.

Now, as a typologist, I wouldn’t worry about all those imperfect matches – if it wasn’t for the fact that Morgan’s six types aren’t actually based on a comprehensive classification of logically possible systems. It seems that he first realized the contrast between the ‘Eskimo’ (= modern European) and ‘Iroquoian’ types (he famously worked as a lawyer for members of the Seneca nation right around my adopted hometown of Rochester, NY) and then developed the other four types in contrast to those two.

Anyway, the tl;dr: I just recently came across Passmore et al. (2021)<https://www.researchgate.net/publication/352203219_Kin_Against_Kin_Internal_Co-selection_and_the_Coherence_of_Kinship_Typologies>, which seems to support my suspicion that Morgan’s typology is deficient. Passmore and colleagues show that Morgan’s classification doesn’t robustly support intergenerational inferences: that is, knowing a language’s classification in one generation (say, Ego’s generation, or the first ascending or descending generation), the predictive value for the same language’s classification in another generation is only moderate to fairly low.

Of course, this doesn’t necessarily mean that a superior typology is possible. Maybe that’s just how messy kinship nomenclatures are, especially considering that research on the relative diachronic stability of different systems seems to still be in its infancy today.

Now, a possible soft constraint on colexification that I’ve been interested in for a while concerns the phenomenon variously known as contronymy<https://en.wikipedia.org/wiki/Contronym>, enantiosemy, or autoantonymy. It basically involves          lexemes with antonymic senses. (Importantly, these senses should occur synchronically and be used by the same speakers. So we’re not talking about semantic change between antonymic senses, as in the case of peruse discussed in unpublished work by David Wilkins. But it could well be that incomplete semantic change is one source of contronymy.)

A well-worn example is cleave, for which the OED has both ‘To part or divide by a cutting blow; to hew asunder; to split’ and ‘To stick fast or adhere, as by a glutinous surface’. The former sense goes back to OE clíofan or cléofan, the latter to OE clífan/clifian/cleofian (all forms appear to be attested).

I would argue that these senses are merely imperfect antonyms (specifically, imperfect reversives). More importantly, the second sense occurs with much lower frequency than the first and is associated with a distinct argument structure, as it requires a to complement, which the first sense does not occur with.

A better case in point – and one that has long confused me as an L2 speaker – is arguably sanction. For this, the OED has on the one hand ‘To ratify or confirm by sanction or solemn enactment; to invest with legal or sovereign authority; to make valid or binding’, ‘To permit authoritatively; to authorize; in looser use, to countenance, encourage by express or implied approval’, but also ‘To enforce (a law, legal obligation, etc.) by attaching a penalty to transgression’ and ‘To impose sanctions upon (a person), to penalize’.

Here, the two senses are not differentiated by argument structure. But we’re also dealing with a different antonymy relation – presumably one of cohyponymy.

So there seems to be no question that contronymy really exists. But it’s equally clear that contronymy is rare. And it seems intuitively obvious why it’s rare: because it gives rise to confusion and misunderstandings. What is very much in question in my mind is under what conditions contronymy nevertheless occurs.

A possible boundary case we discussed here on Lingtyp not too long ago is colexification of ‘hot’ and ‘cold’. Is it possible for such a lexeme to be communicatively useful? As I pointed out in that thread, it might simply be used to mean effectively something like ‘noticeably distinct from body temperature.’

Best – Juergen

Juergen Bohnemeyer (He/Him)
Professor, Department of Linguistics
University at Buffalo

Office: 642 Baldy Hall, UB North Campus
Mailing address: 609 Baldy Hall, Buffalo, NY 14260
Phone: (716) 645 0127
Fax: (716) 645 3825
Email: jb77 at buffalo.edu<mailto:jb77 at buffalo.edu>
Web: http://www.acsu.buffalo.edu/~jb77/

Office hours Tu/Th 3:30-4:30pm in 642 Baldy or via Zoom (Meeting ID 585 520 2411; Passcode Hoorheh)

There’s A Crack In Everything - That’s How The Light Gets In
(Leonard Cohen)
--

From: Lingtyp <lingtyp-bounces at listserv.linguistlist.org> on behalf of Alex Francois via Lingtyp <lingtyp at listserv.linguistlist.org>
Date: Saturday, February 1, 2025 at 14:00
To: Maria Koptjevskaja Tamm <tamm at ling.su.se>
Cc: lingtyp at listserv.linguistlist.org <lingtyp at listserv.linguistlist.org>
Subject: Re: [Lingtyp] Universal constraints on lexicalisation
dear all,

I was going to make the same point as Guillaume.  Many languages in the world colexify Father with Father's brother (F=FB) - also known in English as paternal uncle.  However, the usual implication of such systems is that F=FB is dislexified from maternal uncle (Mother's brother, MB).
[Note:  I coined the term dislexify in my 2022 paper "Lexical tectonics: Mapping structural change in patterns of lexification<https://www.degruyter.com/document/doi/10.1515/zfs-2021-2041/html>"]

Vanuatu systems
The pattern {F=FB}≠{MB} is the most common one in Pacific societies. Thus in Vanuatu, I use different terms, and have different sorts of interactions, with my (classificatory) fathers vs. with my uncles [MB].
I here use Eng. father(s) to refer to the emic category F=FB, and uncle(s) for MB, i.e. whatever term is distinct from father.  The creole Bislama does the same: calquing the vernacular substrates,  ankel  exclusively refers to MB, whereas papa is both used for F and FB.

This system is known, in Morgan's 1871 kinship typology, as the "Iroquois system<https://en.wikipedia.org/wiki/Iroquois_kinship>" (or its Crow & Omaha variants):

[cid:ii_m6mjqsv45]

For comparison, European languages usually belong to Morgan's Eskimo system<https://en.wikipedia.org/wiki/Eskimo_kinship>:
[cid:ii_m6mjs29d6]

Systems where F=FB=MB are rare in the world, but they do exist. They correspond to Morgan's Hawaiian kinship system:
[cid:ii_m6mjmml84]

(Outside of Melanesia, Zygmunt mentioned that F=MB in some Chadic languages;  I assume that the term also includes FB, and so those systems are Hawaiian.)

I don't know any language in Vanuatu that has Hawaiian-like terminology, i.e. would colexify F=FB=MB.  The dislexification F(B)≠MB is usually perceived as essential:  while maternal uncles might be described in English figuratively as a kind of "social father", the point of the Iroquois / Crow system is precisely that they are distinct from actual fathers and their brothers.
In (at least some parts of) Vanuatu, this principle is linked to matrilineal transmission of land rights:  I own the same land as my mother and her brothers, but not the same land as my fathers.
________
Men vs. women referents
The Iroquois system (at least in Vanuatu societies) usually works symmetrically for women:
My mother's sisters are my (classificatory) mothers;  but my father's sisters are my "aunts" -- i.e. there's a special term dislexified from mother (sometimes derived from mother, but distinct from it).  Also, the word for aunt is also used for my maternal uncle's wife; but my paternal uncle's wife is simply my mother.  Using kinship abbreviations (where Z='sister'), we have {M=MZ=FBW} on the one hand, and {FZ=MBW} on the other.

We could say that the Iroquois system follows a principle of dislexification between 1/ the terms for parents and 2/ the term for parents' cross-sex siblings.

I have found some exceptions though, when the referents are women:
·         e.g. in Teanu (Solomons), the word <https://marama.huma-num.fr/Lex/Teanu/e.htm#%E2%93%94ete> ete<https://dictionaria.clld.org/units/teanu-ete_1> colexifies M=MZ=FZ=FBW=MBW;  whereas for males, aia<https://dictionaria.clld.org/units/teanu-aia_1> 'father' [F=FB=MZH] is still dislexified from gea 'uncle' [MB].
·         I also observed this unexpected colexification {M=MZ=FZ=FBW=MBW} in 3 languages of N Vanuatu, namely Hiw, Lo-Toga, Lakon.
In other terms, in the Pacific languages I've observed, the principle of dislexification between parents and their cross-sex siblings is absolute for male referents, but only a statistical trend for female referents.
This observation is confirmed by Fox (2021) for Austronesian languages more generally:
·         Fox, James J. 2021. A research note on laterality and lineality in Austronesian relationship terminologies. Oceania 91.3 (2021): 367-374. [doi<https://doi.org/10.1002/ocea.5317>]
________

What I don't think exists, though, are languages that colexify F with MB and not with FB.  Like Guillaume and Masha, I would be very intrigued if these were found.  (For women, the same surprise would occur if M colexified with FZ, but not with MZ.)

Kinship systems, due to their inherent constraints, can surely provide other examples of impossible patterns of lexification.

best
Alex
________________________________

Alex François
LaTTiCe<http://www.lattice.cnrs.fr/en/alexandre-francois/> — CNRS–<https://www.cnrs.fr/en>ENS<https://www.ens.fr/laboratoire/lattice-langues-textes-traitements-informatiques-et-cognition-umr-8094>–PSL<https://www.psl.eu/en>–Sorbonne nouvelle<http://www.univ-paris3.fr/lattice-langues-textes-traitements-informatiques-cognition-umr-8094-3458.kjsp>
Australian National University<https://researchers.anu.edu.au/researchers/francois-a>
Personal homepage<http://alex.francois.online.fr/>
_________________________________________

---------- Forwarded message ---------
From: Maria Koptjevskaja Tamm via Lingtyp <lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>>
Date: Sat, 1 Feb 2025 at 18:19
Subject: Re: [Lingtyp] Universal constraints on lexicalisation
To: Guillaume Jacques <rgyalrongskad at gmail.com<mailto:rgyalrongskad at gmail.com>>
Cc: lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org> <lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>>

Dear all, dear Guillaume,

Thanks for your input on the alleged kin term universal. Guillaume is completely right in what he writes about its claim – mea culpa, I should have made this clearer from the start.

I would be interested in getting more information on whether it holds – or whether there are examples going in the direction of Östen’s “guess”.

Best,
Masha

On Feb 1, 2025, at 17:43, Guillaume Jacques via Lingtyp <lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>> wrote:

Dear all,

Concerning the universal on kinship terms that Masha was mentioning, the claim is not that no language can colexify Father (F) and Mother's Brother (MB), but rather that if F=MB then F=MB=FB (Father's Brother), in other words you don't have a language colexifying F and MB and dislexifying FB from them (F=MB≠FB). I think that this is a very robust universal, which brings important evidence for the general principles of the evolution of kinship systems.

Best wishes,

Guillaume

Le sam. 1 févr. 2025 à 17:19, Östen Dahl via Lingtyp <lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>> a écrit :
Dear all,

With regard to the claim that 'father' and 'mother's brother' cannot be colexified, consider the following quotation from the Wikipedia article on "Matrilineality":

"While a mother normally takes care of her own children in all cultures, in some matrilineal cultures an "uncle-father" will take care of his nieces and nephews instead: in other words *social fathers* here are uncles."

That is, fathers and maternal uncles are similar in that they can both play the role of "social fathers"; it is not unthinkable that a language spoken in a society on the borderline between patrilineality and matrilineality will lexify the concept "social father". What this shows is that the criterion of cognitive complexity can lead you in the wrong direction. In fact, kinship terms sometimes unite relationships which are tricky to give a common definition, such as "brother-in-law" in English.

·         Östen

Från: Lingtyp <lingtyp-bounces at listserv.linguistlist.org<mailto:lingtyp-bounces at listserv.linguistlist.org>> För Martin Haspelmath via Lingtyp
Skickat: den 1 februari 2025 16:40
Till: lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>
Ämne: Re: [Lingtyp] Universal constraints on lexicalisation

Dear Masha and others,

In addition to "cognitive complexity", one may also consider frequency of use as constraining lexification.

For example, 'female wolf' is not more cognitively complex than 'female horse' (English mare, contrasting with stallion), but gender/sex is less commonly mentioned in connection with wild animals than with domestic animals, so English does not dislexify 'male wolf' and 'female wolf'.

In my 2023 Frontiers paper, I suggested that some important lexification tendencies can be explained with reference to root length possibilities: Roots are typically 1-2 syllables long, so when a meaning is not frequent enough, it needs more syllables and hence multiple morphs:
Haspelmath, Martin. 2023. Coexpression and synexpression patterns across languages: Comparative concepts and possible explanations. Frontiers in Psychology 14. (doi:https://doi.org/10.3389/fpsyg.2023.1236853)

(The paper also cites David Gil's 1992 paper.)

Incidentally, it seems that "lexification" is clearer than "lexicalization", because the latter is used in multiple meanings (see my 2024 paper, §7: https://www.peren-revues.fr/lexique/1737).

Best,

Martin

On 01.02.25 12:40, David Gil via Lingtyp wrote:
Hi Masha,

Some examples from the semantic domain of quantification can be found here:

Gil, David (1992) "Scopal Quantifiers: Some Universals of Lexical Effability", in M. Kefer and J. van der Auwera eds., Meaning and Grammar, Cross-Linguistic Perspectives, Mouton de Gruyter, Berlin, 303-345.

Best wishes,

David

On Sat, Feb 1, 2025 at 5:29 PM Maria Koptjevskaja Tamm via Lingtyp <lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>> wrote:
Dear all,

I am involved in a handbook chapter in which I would like to give a few examples of suggested universal constraints on lexicalisation, e.g., those primarily concerning meanings that should not be expressible in a word (a stem, root or whatever), preferably not from the domain of colour terms. To give an example, Rappaport Hovav and Levin (2010) argue that no verb encodes both manner and result simultaneously, which has been contested by Beavers and Koontz-Garbodens.

Or,  a definition of a term covering both ‘father’ and ‘mother’s brother’ would be cognitively very complex since it will require disjunction (‘father’ or ‘mother’s brother’, cf. ‘male relative of one’s patriline’ for ‘father’ and ‘father’s brother’) (Evans 2001) – I don’t know if this constraint still holds.

Many thanks and all the best,
Masha

Prof. Maria Koptjevskaja Tamm
Dept. of linguistics, Stockholm university, 106 91, Stockholm, Sweden
Editor-in-chief of “Linguistic Typology”
President-Elect of Societas Linguistic Europaea
www.ling.su.se/tamm<http://www.ling.su.se/tamm>
tamm at ling.su.se<mailto:tamm at ling.su.se>

_______________________________________________
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp

--

David Gil

Senior Scientist (Associate)

Department of Linguistic and Cultural Evolution

Max Planck Institute for Evolutionary Anthropology

Deutscher Platz 6, Leipzig, 04103, Germany

Email: dapiiiiit at gmail.com<mailto:dapiiiiit at gmail.com>

Mobile Phone (Israel): +972-526117713

Mobile Phone (Indonesia): +62-082113720302

_______________________________________________

Lingtyp mailing list

Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>

https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp

--

Martin Haspelmath

Max Planck Institute for Evolutionary Anthropology

Deutscher Platz 6

D-04103 Leipzig

https://www.eva.mpg.de/linguistic-and-cultural-evolution/staff/martin-haspelmath/
_______________________________________________
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp

--
Guillaume Jacques

Directeur de recherches
CNRS (CRLAO) - EPHE- INALCO
https://scholar.google.fr/citations?user=1XCp2-oAAAAJ&hl=fr
https://langsci-press.org/catalog/book/295<http://cnrs.academia.edu/GuillaumeJacques>
http://panchr.hypotheses.org/
_______________________________________________
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp

Prof. Maria Koptjevskaja Tamm
Dept. of linguistics, Stockholm university, 106 91, Stockholm, Sweden
Editor-in-chief of “Linguistic Typology”
President-Elect of Societas Linguistic Europaea
www.ling.su.se/tamm<http://www.ling.su.se/tamm>
tamm at ling.su.se<mailto:tamm at ling.su.se>

_______________________________________________
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
https://listserv.linguistlist.org/cgi-bin/mailman/listinfo/lingtyp
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250201/cf7e8cd9/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 11087 bytes
Desc: image.png
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250201/cf7e8cd9/attachment-0003.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 73023 bytes
Desc: image.png
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250201/cf7e8cd9/attachment-0004.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 23013 bytes
Desc: image.png
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20250201/cf7e8cd9/attachment-0005.png>