[Lingtyp] Plural markers on (already) plural pronouns

Haspelmath, Martin haspelmath at shh.mpg.de
Mon Nov 18 06:19:06 UTC 2019


It seems to me that we do not need special mechanisms such as "cryptanalysis" (Croft 2000) or "multiple exponence" (Matthews 1972; Harris 2017) to explain phenomena like Sri Lanka Malay kitham-pada [we-PL], or Latin no-s [we-PL]. As shown by Daniel (2005/2013) in his WALS chapter<https://wals.info/chapter/35>, such cases are very common in the world's languages.

All languages have very few suppletive roots, but when they occur, we find them in high-frequency domains such as pronouns, verbs meaning 'be' or 'go', adjectives meaning 'good' or 'bad', and nouns denoting people. This is a very strong and well-understood tendency (due to the greater memory strength of high-frequency experiences).

Now if we distinguish informally between "root meanings" and "affix meanings", we find that a suppletive root may either express only a root meaning (as in "go/wen-t", "good/bett-er", "ego/no-s")), or it may "fuse" or "compress" both a root meaning and an affix meaning (as in "is/was", "bad/worse", "I/we").

Corbett (2007: 15) says that the latter case ("fused-exponence suppletion") is more canonical, but this does not imply that it is somehow more usual or more preferred.

Bybee (1985) showed that "more relevant" affix meanings are more likely to occur close to the root, and also to be compressed together with the root meaning in a suppletive root. So we expect implications like the following:

case > number (i.e. if a pronoun suppletes for case, it suppletes for number)
person > tense (i.e. if a verb suppletes for person, it suppletes for tense)

This is indeed what we seem to find (though systematic worldwide studies are rare in the recent literature). In Indo-European, we often find case suppletion (e.g. Greek eghó/eme-na [I.NOM/I-ACC]), but we also find number suppletion. In a few cases we find person suppletion (e.g. Latin sum/es-t [be.1SG/be-3SG]), and again tense suppletion is also found in these verbs (sum/fu-i [be.PRS.1SG/be.PST-1SG]).

So apart from high-frequency irregularity and Bybeean relevance, we do not seem to need any explanatory mechanisms. Nothing favours Sri Lanka Malay kithang [we] over kitham-pada [we-PL], as far as I can see, so the [we-PL] type can happily occur in many languages, often alongside the compressed [we] type.

Best wishes,
Martin

P.S. As for Bernhard Wälchli's question: "why is it just morphologically marked categories that tend to be reinforced?" The answer seems quite simple: "Morphologically marked categories" really means "less frequently used grammatical meanings", and for well-understood Zipfian (least-effort) reasons, "marking/reinforcing" occurs when the hearer cannot so easily guess what the speaker means (and this is orthogonal to "transparency").

On 17.11.19 17:22, Peter Arkadiev wrote:
Dear Bernhard, dear all,

an insightful cross-linguistic and diachronic discussion of reinforcement is given in "Multiple Exponence" by Alice Harris (OUP, 2017). As far as I remember, she does not furmulate universal hypotheses, but her book if surely highly relevant for the points raised by Bernhard.

Best regards,

Peter

--
Peter Arkadiev, PhD
Institute of Slavic Studies
Russian Academy of Sciences
Leninsky prospekt 32-A 119991 Moscow
peterarkadiev at yandex.ru<mailto:peterarkadiev at yandex.ru>
http://inslav.ru/people/arkadev-petr-mihaylovich-peter-arkadiev



17.11.2019, 00:33, "Bernhard Wälchli" <bernhard at ling.su.se><mailto:bernhard at ling.su.se>:



Dear all,



Bill Croft treats the phenomena in question in his seminal book “Explaining Language Change” (2000: 134) under the heading of “cryptanalysis” (§5.5): “In cryptanalysis, the listener analyses a covert semantic/functional property of a syntactic unit as not grammatically marked, and inserts {obviously it is meant when turning into a speaker again, BW} an overt marker expressing its semantic value.” Some of the examples discussed on this list are given in Bill’s subsection §5.5.3 “Pleonastic double marking / reinforcement”.

I am actually not certain whether this is the only possible interpretation. The overt marker could be inserted because the speaker considers the semantic property not sufficiently marked. I do not think all speakers using reinforcement are unable to analyze the non-reinforced expression (despite frequent claims by purists that using pleonasms is stupid). A more general mechanism that can account for reinforcement - although not the sublte differences in politeness in Turkish - is Lüdtke’s model of quantitative language change. (Croft 2000: 159 discusses it as the “periphrasis-fusion-erosion” cycle.) According to Lüdtke, speech has to be redundant since it occurs in a noisy environment, and this entails in the long run three types of changes for saying the same thing which trigger each other: fewer phonemes (Croft’s term “erosion”), more morphemes (Croft’s “periphrasis”) and fewer morpheme boundaries (Croft’s “fusion”) for conveying the same message. So, aren’t the phenomena under discussion here just instances of this? As in the well-known development of French demonstratives from Latin:

ista N > (ecce)ista N > (c)este N > cette N > cette N(-ci)

hoc > (ecce)hoc > (ç)o > ço > ce > ce(la) > cela > ça

However, probably there is more to it if we take phonology into account. It seems to me that it happens very rarely that a structure Y-x (with -x being the marker) is reinforced pleonastically with a phonologically identical marker: Y-x-x. This seems to be due to avoidance of subsequent identical sequences (there is also a diachronic process doing away with them: haplology, funnily except in the term itself). Indeed, in no example mentioned so far, the new reinforced structure is Y-x-x (with x being the same allomorph).

Against such phonological considerations one might argue that the reinforcing element is usually (has to be ??) more transparent than the earlier exponent of the semantic category (otherwise the reinforced marker is not “crypto-”). However, more transparent exponents also tend to be more productive, and it is more likely that reinforcement is done by a productive strategy than by a non-productive strategy. After all, if the point of reinforcement is more clarity for the listener, it might be strange to use non-transparent markers for that purpose. Another interesting question is: why is it just morphologically marked categories that tend to be reinforced? (This might be relevant for the issue of typological markedness.) Does anybody have an example where a singular personal pronoun is reinforced by a singular marker? (There are certainly many cases of dual pronouns being reinforced by ‘two’ and the like). Note also that reinforcement may entail more complexity as when gender is introduced in pronouns: Spanish nos-otros/nos-otras ‘we’, Lithuanian ju-du ‘you(dual)-two[M]’, ju-dvi ‘you(dual)-two[F]’.



So here are some claims (I’d love to see counter-examples)

1. Reinforcing morphemes are more transparent and productive than reinforced morphemes.

2. Reinforced and reinforcing morphemes are not the same allomorph.

3. Reinforcing (except purely phonetic reinforcing e.g. by stress) occurs at the periphery (by affixes, adjacent words, not by infixes, ablaut etc.).

4. Typologically unmarked categories are not reinforced. (This does not hold. For instance, Indo-European present -i is a counter example, so no counter-examples for this, please)

4b. Much weaker: singular personal pronouns are not reinforced by singular.

4c. Typologically unmarked categories are reinforced less often than typologically marked categories [possibly a tautology when considered from a pan-chronic perspective]

(Note that 4b can be accounted for by frequency, whereas 4 cannot.)

Hence 4d: Typologically unmarked categories are not reinforced in uses where they are relatively more frequent than their marked counterpart.

5. Short forms are reinforced more frequently than longer forms.

6. Transparently marked forms are less frequently reinforced than opaquely marked forms.

7. Reinforcement is more likely to happen in phrases consisting of one word than in phrases consisting of several words (thus we typologists is less prone to be reinforced than we).



Lüdtke, Helmut (1980). Kommunikationstheoretische Grundlagen des Sprachwandels. ‘Sprachwandel als universales Phänomen’, 1-19; ‘Auf dem Weg zu einer Theorie des Sprachwandels’, 182-252. Berlin: de Gruyter.



Best,

Bernhard W.






________________________________
From: Lingtyp <lingtyp-bounces at listserv.linguistlist.org<mailto:lingtyp-bounces at listserv.linguistlist.org>> on behalf of Ponrawee Prasertsom <ponrawee.pra at gmail.com<mailto:ponrawee.pra at gmail.com>>
Sent: Friday, November 15, 2019 5:01 PM
To: lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>
Subject: [Lingtyp] Plural markers on (already) plural pronouns

Dear all,

I have been looking at a number of Tai languages and found that in some of these languages, plural pronouns can optionally take a plural marker normally used on common nouns. For instance, in Shan (Southwestern Tai), the third person plural pronoun khau can optionally take the plural marker cɯ(nai), viz. khau cɯ(nai)--at least according to Cushing 1871.

Assuming this analysis is correct (if it's not please kindly inform me), I'm wondering how rare this is for pronouns? A quick lookup revealed that a similar phenomenon called "double plural marking" is found in some languages, but seems to be restricted to common nouns only. Does anyone know of any other instances like this for pronouns in other languages?

Sources: Cushing, Josiah Nelson. Grammar of the Shan Language. Rangoon: American Mission Press, 1871.

Best regards,

--
Ponrawee Prasertsom

Graduate Student
Department of Linguistics
Faculty of Arts, Chulalongkorn University
Bangkok, Thailand
,

_______________________________________________
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
http://listserv.linguistlist.org/mailman/listinfo/lingtyp



_______________________________________________
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
http://listserv.linguistlist.org/mailman/listinfo/lingtyp




--
Martin Haspelmath (haspelmath at shh.mpg.de<mailto:haspelmath at shh.mpg.de>)
Max Planck Institute for the Science of Human History
Kahlaische Strasse 10
D-07745 Jena
&
Leipzig University
Institut fuer Anglistik
IPF 141199
D-04081 Leipzig
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20191118/8c6a849a/attachment.htm>


More information about the Lingtyp mailing list