Syntax of the World’s Languages 5 in Dubrovnik, a conference report

Eight years ago we decided to organize a new kind of conference in Leipzig, a syntax conference for linguists with a serious interest in the whole range of diversity of the world’s languages. In other words, a conference for fieldworkers and typologists, where the comparativists can meet the linguists with first-hand data experience, and where (unlike at typology conferences) fieldworkers who focus on a single language do not need to feel out of place (Syntax of the World’s Languages). This proved to be a success, and after Lancaster, Berlin and Lyon, the fifth conference in this biennial series was organized by the University of Zagreb’s Ranko Matasović, in Dubrovnik on the Adriatic coast. There were about 70 papers, of which about 15 were typological papers, and the rest on individual languages from around the world. Well, Eurasian languages were overrepresented as usual, with only one Papuan paper, and only seven papers on languages of the Americas. This was probably also due to the fact that the country with the biggest number of representatives was Russia, where many syntacticians work on Caucasian, Uralic and Altaic languages.
Some of the highlights from my perspective:

Bernard Comrie‘s plenary talk on differential object marking (DOM), where he discussed the claim by Bickel & Witzlack-Makarevich (2008) that DOM is not robustly attested in the world’s languages. Comrie noted that in contrast to what these authors do in their paper, DOM has usually been formulated in fairly narrow terms (involving only flagging, not indexing, and involving only animacy and referentiality, not other parameters). And even though he did not give statistical data of his own, he noted that under a reasonable definition of counterevidence and a reasonable definition of what is an independent case, there could be little doubt that DOM is a strong general tendency of human language. The fact that it is not equally represented in all areas is not damaging for this claim, as long as we find it again and again even in the other areas (e.g. in some corners of the grammar of Tagalog, in Sereer in West Africa, and in a Japanese dialect). Bickel and his associates (see now also Bickel et al. 2012+) seem to assume that when two languages are spoken in the same macro-area, they cannot constitute independent cases, but it’s very difficult to imagine, for example, how Japanese and Tagalog should have influenced each other. Thus, they seem to greatly underestimate the number of independent cases.

Sonia Cristofaro‘s talk on alienable/inalienable contrasts, where she argued both against a Haiman-style iconicity account and against my frequency account (Haspelmath 2008). Instead, she claimed, alienability contrasts are due to the sources from which possessive constructions grammaticalize. For example, when possessive constructions come from a noun meaning ‘possession’ (as in Arabic vernaculars), one does not expect this to be used for kinship terms and body-part terms because relatives and body-parts are not thought of as possessions. But here I would still ask why one doesn’t find constructions with an over noun meaning ‘part’ being grammaticalized as a possessive marker: ‘my-part the hand’ would be semantically as plausible as ‘my-possession the book’, but such developments never take place. So I feel that a source-determination account will eventually have to be combined with a frequency-based account.

Robert D. Van Valin‘s talk on information structure in Lakhota gave a detailed account of various properties of Lakhota that are relevant to information structure, such as the rich system of (especially) indefinite articles, of emphatic personal pronouns, of word order and prosody, and of clefting by means of an internally-headed relative clause (“internally-headed cleft”). This talk was remarkable because Van Valin is better known as the leading figure of Role and Reference Grammar, but here he showed that he has not lost interest in in-depth language-particular studies which make no direct reference to a formal framework, i.e. that he has not lost touch with the foundations of all theory. He did ask a general question, namely whether Lakhota’s head-marking structure is related to any of its information-structural properties, but his answer was negative: None of its peculiarities seem in any way related to its head-marking nature. I was not so surprised by this, because it seems to me that the deeper significance of the head-marking/dependent-marking typology has often been overestimated. For information structure, what seems more relevant is whether highly topical referents can be completely dropped (as in Japanese and Thai), or whether they need to be expressed obligatorily, as in all languages with bound person forms and in languages like English (where one cannot say “Do you like it? Yes, Ø like it.”) . Languages with obligatory topic expression are overwhelmingly of the Lakhota type, with bound person forms expressing the topic, so Lakhota is actually the most usual type of language (it’s English that is the unusual case, with an obligatory set of person forms that are not restricted to topical use and that can be used contrastively).

Alexander Letuchiy and Peter Arkadiev, who talked about indirect antipassives in Circassian (Kabardian and Adyghe). While a standard antipassive downgrades (demotes or removes) a direct object (or P-argument), an oblique (or “indirect”) object can also be removed by a kind of antipassive construction in these two closely related languages. Such cases have not been reported previously in the literature, but Leutuchiy & Arkadiev note that they are not entirely unexpected: Similarly, applicatives sometimes add oblique arguments rather than P-arguments (e.g. in Kartvelian, where they are called “version”). And Circassian languages have an unusually large number of two-argument verbs with an oblique second argument (e.g. ‘hit’, ‘bite’, ‘kiss’, ‘scold’, ‘catch’), so apparently the need for such a construction is higher here than elsewhere.

This is just a very subjective selection from a wonderful conference in one of Europe’s most splendid tourist destinations. The University of Zagreb has a conference building here, just a few minutes from the old town and the beach, which was the beautiful backdrop for a very productive encounter.


Bickel, Balthasar & Alena Witzlack-Makarevich. 2008. Referential scales and case alignment: reviewing the typological evidence. In Marc Richards & Andrej L Malchukov (eds.), Scales, 1–37. (Linguistische Arbeitsberichte 86). Leipzig: University of Leipzig.

Bickel, Balthasar, Alena Witzlack-Makarevich & Taras Zakharko. 2012+. Typological evidence against universal effects of referential scales on case alignment. Ms., University of Zurich.

Haspelmath, Martin. 2008. Frequency vs. iconicity in explaining grammatical asymmetries. Cognitive Linguistics 19(1). 1–33.

One thought on “Syntax of the World’s Languages 5 in Dubrovnik, a conference report

  1. Just two quick notes on the report from Bernard Comrie’s talk (which I haven’t heard myself): (a) in our papers we clearly distinguish between case and other possible loci of DOM effects; (b) no one knows how many languages located between the ancestors of Japanese and Tagalog had DOM 10,000 years ago, and so we cannot rule out areal spread over time lightly. This is why one needs to control for deep-time areal confounds when testing universals (as Matthew Dryer has made clear long ago in his 1989 classic Studies in Language paper).

    What we do know (from our 2012+ paper) is that the estimated probabilities of developing DOM that fits a universal scale is in the same statistical ballpark as the estimated probability of developing DOM that does not fit a universal scale. This holds for most versions of how one defines the universal scale. The only exception we note in the paper is part-of-speech and definiteness/topicality-based scales: here developments towards DOM fitting the scale outrank developments against DOM fitting the scale, but the overall number of cases is too small for establishing universals in the same class of robustness as, say, word order universals.

Comments are closed.