[Lingtyp] Innovation of functional categories

Wed Jun 17 21:00:48 UTC 2020

Dear Bernhard and all,

The idea that Jürgen Bohnemeyer cited (viz. that grammatical items are those that are conventionalized as carriers of background information) certainly has predecessors, but I believe it presents a rather new take on the problem of defining grammatical items.

Firstly, the auto- vs. synsemantic (or categorematic vs. syncategorematic) distinction, which can indeed be traced back at least to Aristotle, captures only the fact that grammatical items depend on host items, not the fact that they contribute to prioritizing the parts of complex messages.

Secondly, while frequency unquestionably plays a role in grammaticalization, grammar and grammaticalization cannot be reduced to frequency (perhaps this is also not what you were saying, Bernhard). For one thing, individuals with agrammatic aphasia have problems producing grammatical items, but no problems producing frequent items (e.g. Hachard 2015; Martìnez-Ferreiro et al. 2019).

Best wishes,
Kasper

References
Hatchard, R. (2015). A construction-based approach to spoken language in aphasia. University of Sheffield thesis.
Martínez-Ferreiro, S., R. Bastiaanse & K. Boye. 2019. "Functional and usage-based approaches to aphasia: the grammatical-lexical distinction and the role of frequency". Aphasiology, DOI: 10.1080/02687038.2019.1615335.

Fra: Bernhard Wälchli <bernhard at ling.su.se>
Sendt: 17. juni 2020 22:10
Til: Kasper Boye <boye at hum.ku.dk>; David Gil <gil at shh.mpg.de>; lingtyp at listserv.linguistlist.org
Cc: Peter Harder <harder at hum.ku.dk>
Emne: Re: [Lingtyp] Innovation of functional categories

Dear all,

Several of the ideas mentioned have in some or other form been around for a very long time. Consider, for instance, Anton Marty's distinction between autosemantic and synsemantic expressions. https://stanford.library.sydney.edu.au/entries/marty/<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fstanford.library.sydney.edu.au%2Fentries%2Fmarty%2F&data=02%7C01%7Cboye%40hum.ku.dk%7C1b5d06f727bd49927cfa08d812fa6c25%7Ca3927f91cda14696af898c9f1ceffa91%7C0%7C0%7C637280215066374787&sdata=fjONy50iT4Ww7Gf7sfHbucUnWWyDXfyo6pauUSBNgGg%3D&reserved=0> and G. K. Zipf's findings that frequency correlates with the number of meanings of an expression. [And, of course, also Marty and Zipf have their predecessors.]

Zipf, George Kingsley. 1945. The meaning-frequency relationship of words. The Journal of General Psychology 33:2, 251-256

"If we postulate (A) that man is always economical in his speech, and if we assume (B) that there are M number of different meanings to be verbalized by a given vocabulary, then we see the emergence of two conflicting economies in allocating meanings to words. One economy (a) will be the speaker's economy which would favor a single word of 100 per cent frequency
with M different meanings; although such an arrangement would cause the auditor extreme work in understanding, it would save the speaker the work of selecting particular words with particular meanings. On the other hand (b) there will be an auditor's economy which, for the auditor's greater ease of comprehension and to the disadvantage of the speaker, would
favor a different word for each different meaning, and therefore favor a vocabulary of M different words with a much lower average relative frequency.
            Hence there are two opposing drives: the one (a) making for a vocabulary of a single word with M meanings and 100 per cent frequency, and the other (b) making for a vocabulary of M different words with one meaning per word and with a lower average relative frequency.
            As a result (C) of the above opposing drives we may expect to find in a sizeable sample of running speech some sort of balance between the n-number of different words on the one hand and their frequency of occurrence on the other." (1945: 255)

[Alas, linguistic terms are subject to Zipf's meaning-frequency relationship as much as expressions of natural languages. "Fore-/background(ing)" are good examples of terms that are used in quite different senses in the literature.]

Given that linguistic items vastly differ in frequency (incidentally, another of Zipf's findings), it is not particularly surprising that some items will always be more synsemantic than others (to different extents in different languages or proto-proto-languages).

It seems to me that it does not speak against these and similar ideas that these wheels are constantly reinvented in linguistics - with slight variations.

Bernhard

________________________________
From: Lingtyp <lingtyp-bounces at listserv.linguistlist.org<mailto:lingtyp-bounces at listserv.linguistlist.org>> on behalf of Kasper Boye <boye at hum.ku.dk<mailto:boye at hum.ku.dk>>
Sent: Wednesday, June 17, 2020 7:10:39 PM
To: David Gil; lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>
Cc: Peter Harder
Subject: Re: [Lingtyp] Innovation of functional categories

Dear all,

I would like to first respond to David Gil's comment on the problems of defining grammatical, then to follow up on Jürgen Bohnemeyer's ideas about the job grammatical items do.

There are a couple of solutions to the problem of defining "grammatical" on the market. One is Christian Lehmann's (2015) structural definition in terms of autonomy; another one is Peter Harder's and my own functional and usage-based definition (to which I think Bohnemeyer was referring in his initial message) in terms of conventionalized secondary discourse prominence (or backgroundedness) (Boye & Harder 2012; references below). Both definitions bring together morphosyntactic and semantic properties of being grammatical. For instance, being by convention backgrounded (that is, being conventionally associated with background meaning only) entails being dependent on a host expression: a background requires a foreground.

Based on our 2012 definition, Peter and I are currently working in the same direction (book in prep.) as Bohnemeyer. As we see it, the raison d'être for grammatical items is that they offer a shortcut to meaning, which may be inferable from the context: knowing by conventions is faster than inferring from context, even if inferencing is guided by expectations (cf. Lehmann's remark earlier in this thread). In fact, they offer a cheap shortcut: representing background information they can be processed rather superficially and thus quickly by the hearer.

Looking now at grammatical items from the speaker perspective, it is clear that the production of grammatical items requires an extra effort. Not only does it require something extra to produce a grammatical item than not to produce anything, recent studies suggest that grammatical items are in fact a little harder to produce than lexical ones (e.g. Michel Lange et al. 2015; Nielsen et al. 2019). This may be seen as a consequence of their dependency: when you produce a grammatical item, you also have to consider which host expression to attach it to.

If we now look at grammatical items from the speaker and hearer perspective at the same time, a rather sympathetic picture of human communication emerges, a picture that can be characterized in terms of the notion of 'audience design': the speaker invests a little extra production resources so that the hearer can save a little processing resources.

Why then do languages differ when it comes to the amount of grammatical morphemes? Well, firstly evolution is not optimizing, but satisficing, and we can easily live without grammatical morphemes (our forefathers probably did; cf. Hurford 2012) - it puts an extra burden on inferencing, but we are quite good at inferencing. Secondly, phonologically concrete morphemes are not the only kind of grammatical items we have. Also construction grammar's schematic constructions qualify as grammatical items: they are conventionalized as carriers of background info.

With best wishes,
Kasper

References
Boye, K. & P. Harder. (2012). A usage-based theory of grammatical status and grammaticalization. Language 88.1. 1-44.
Link: https://www.jstor.org/stable/pdf/41348882.pdf<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.jstor.org%2Fstable%2Fpdf%2F41348882.pdf&data=02%7C01%7Cboye%40hum.ku.dk%7C1b5d06f727bd49927cfa08d812fa6c25%7Ca3927f91cda14696af898c9f1ceffa91%7C0%7C0%7C637280215066384783&sdata=l3gP4beA6ScPnvCVbrqXr0hmE%2BYdZcu5BpnNgSKdOyQ%3D&reserved=0>

Hurford, J.R. (2012). The origins of grammar: Language in the light of evolution. Oxford: Oxford University Press.

Lehmann, Christian. 2015. Thoughts on grammaticalization, 3rd ed. Berlin, Germany: Language Science Press.

Michel Lange, V., M. Messerschmidt, P. Harder, H.R. Siebner & K. Boye. (2017). Planning and production of grammatical and lexical verbs in multi-word messages. PLoS ONE 12.11. e0186685. https://doi.org/10.1371/journal.pone.0186685<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.1371%2Fjournal.pone.0186685&data=02%7C01%7Cboye%40hum.ku.dk%7C1b5d06f727bd49927cfa08d812fa6c25%7Ca3927f91cda14696af898c9f1ceffa91%7C0%7C0%7C637280215066384783&sdata=2%2FKna4UlgjTmhqb6faRxz0xEMjl9lbkNcKk7ndArKlY%3D&reserved=0>

Nielsen, S.R., K. Boye, R. Bastiaanse & V. Michel Lange. 2019. The production of grammatical and lexical determiners in Broca's aphasia. Language, Cognition and Neuroscience, DOI: 10.1080/23273798.2019.1616104.
Link: https://doi.org/10.1080/23273798.2019.1616104<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdoi.org%2F10.1080%2F23273798.2019.1616104&data=02%7C01%7Cboye%40hum.ku.dk%7C1b5d06f727bd49927cfa08d812fa6c25%7Ca3927f91cda14696af898c9f1ceffa91%7C0%7C0%7C637280215066394782&sdata=%2FGzNtw2iiI9Wbb714cM%2FQvRTipcYbUHwYPypxDDQrzQ%3D&reserved=0>

Fra: Lingtyp <lingtyp-bounces at listserv.linguistlist.org<mailto:lingtyp-bounces at listserv.linguistlist.org>> På vegne af David Gil
Sendt: 17. juni 2020 14:14
Til: lingtyp at listserv.linguistlist.org<mailto:lingtyp at listserv.linguistlist.org>
Emne: Re: [Lingtyp] Innovation of functional categories

Dear Juergen and all,

My favourite example of an innovation of functional categories comes from some Malayic dialects of central Sumatra - see references below.  For the most part, Malayic languages are completely devoid of functional categories; however, in some parts of central Sumatra, culminating in Kerinci, a system has developed whereby almost every word in the language has two forms, absolute and oblique, formally distinguished by complex rules of ablaut.  The functions of the absolute/oblique alternation are also complex, but I'll mention just one of them, since it ties in to earlier discussion in this thread about the development of articles: in phrase final position, a noun will typically occur in the absolute, however, if it takes the oblique form it is interpreted as definite.  McKinnon et al trace the historical development of the absolute/oblique alternation in terms of a coalescence of two separate developments: (a) the grammaticalization of erstwhile phrase-final phonological alternations; and (b) the phonological attraction of an erstwhile free form (cognate to Standard Malay nya).  Comparative evidence suggests that these developments are very recent, and since there are no non-Malayic languages in the vicinity that would form the basis for a contact explanation, it seems pretty clear that the absolute/oblique alternation constitutes an internally-motivated development of a functional category.

Best,

David

McKinnon, Timothy (2011) The morphophonology and morphosyntax of Kerinci Word-shape alternations, PhD Dissertation, University of Delaware, Newark.
McKinnon, Timothy, Peter Cole and Gabriella Hermon (2011) Object agreement and 'pro-drop' in Kerinci Malay, Language 87.4:715-750.
McKinnon, Timothy, Gabriella Hermon, Yanti and Peter Cole (2018) "From Phonology to Syntax: Insights from Malay", in H. Bartos, M. den Dikken, Z. Bánreti and T. Varadi eds., Boundaries Crossed, at the Interfaces of Morphosyntax, Phonology, Pragmatics and Semantics, Studies in Natural Language and Linguistic Theory 94, Springer, Berlin, 349-371.  DOI: 10.1007/978-3-319-90710-9_22<https://eur02.safelinks.protection.outlook.com/?url=http%3A%2F%2Fdx.doi.org%2F10.1007%2F978-3-319-90710-9_22&data=02%7C01%7Cboye%40hum.ku.dk%7C1b5d06f727bd49927cfa08d812fa6c25%7Ca3927f91cda14696af898c9f1ceffa91%7C0%7C0%7C637280215066394782&sdata=r2JCcMxooCnxJlOaHnbYbHj8XHpVQC6N0G1JNpm2hLs%3D&reserved=0>
McKinnon, Timothy, Yanti, Peter Cole and Gabriella Hermon (2015) "Infixation and Apophony in Malay: Description and Developmental Stages", Linguistik Indonesia 33.1:1-19.

On 16/06/2020 04:48, Bohnemeyer, Juergen wrote:

Dear colleagues - I'm looking for examples of innovations of functional categories. By 'functional categories', I mean the 'grammatical categories' of traditional grammar, such as tense, mood, person, gender, case, etc. I propose a more technical definition below.

Here is what I mean by 'innovation': language families or genera in which the functional expression in question is (i) grammaticalized in one or more members or branches while (ii) being absent in others, with (iii) the balance of evidence  pointing to acquisition in the former languages/branches rather than loss in the latter, and (iv) there being no obvious contact-based explanation for the emergence of the expression in question. (Of course one could define innovation to include contact-based innovation, but I happen to be specifically interested in innovation of functional categories in the absence of contact models.)

I realize of course that certainty about (iii) and (iv) is in many if not most cases not to be had. Consider for illustration the emergence of definite articles in Western Europe (Celtic, Romance, Germanic) during the "Dark Ages". In this case, we can be certain that this was an innovation event due to the presence of historical records both from ancestors of some of the Indo-European languages that developed articles and from ancestors of those that didn't. But when and where this innovation started, and what role (if any) contact with languages from outside Western Europe, such as Ancient Greek, Hebrew, and Arabic, may have played, appears to continue to be unclear.

It is possible if not likely that some of the clearest examples of innovations of functional categories arise in creole languages. Of interest here would be creoles that have grammaticalized a functional category not present in either the lexifier or any substrate or adstrate language.

As a working definition, functional expressions in general (a superordinate category of functional categories in the narrow sense) might be defined as (i) morphemes that (ii) do not belong to any major lexical category, but (iii) enter into fully productive and compositional combinations with (projections of) members of lexical categories. This very broad and general characterization would encompass a host of subtypes. Of great interest to me is the observation that these subtypes are not uniform in how commonly they are grammaticalized vs. missing in the languages of the world. Some functional expressions, such as negation, occur in every single human language. Some, such as adnominal or adverbial expressions of quantification, apparently are present in all languages except for languages that rely on complex predicative workarounds (existential predication for existential quantification, conditional-like structures for universal quantification).

Contrast this with the subtype of functional expressions I'm particularly interested in here, such as tense, viewpoint aspect, definiteness, number, and gender, which are typically present in only between a third and two thirds of the samples of the WALS chapters that report on them. My hypothesis is that this difference in variability correlates with the communicative function of the expressions: expressions such as tense, number, and gender are typically (in the great majority of utterances in which they occur) not needed to express part of the speaker's communicative intention, as the information they contribute is predictable in context. The grammaticalization of such largely redundant expressions apparently serves to reduce the hearer's inference load.

This gradual pragmatic redundancy is from my perspective a defining feature of the class of expressions in question. Obviously, this doesn't translate into a simple diagnostic. However, it aligns with relatively advanced degrees of grammaticalization (compared to things such as negation, demonstratives, or modals), and advanced grammaticalization in turn jibes with the primarily metalinguistic function of the expressions in question: they are always backgrounded, never express "at issue" content, and as a result can never be focalized except metalinguistically.

I hope that wasn't too convoluted ;-)

Thank you in advance for your help! I will post a summary if I receive a sufficient number of responses. - Best - Juergen

--

David Gil

Senior Scientist (Associate)

Department of Linguistic and Cultural Evolution

Max Planck Institute for the Science of Human History

Kahlaische Strasse 10, 07745 Jena, Germany

Email: gil at shh.mpg.de<mailto:gil at shh.mpg.de>

Mobile Phone (Israel): +972-556825895

Mobile Phone (Indonesia): +62-81344082091
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20200617/1305e068/attachment.htm>