Right and wrong in typology: Chappell & Creissels versus Stassen on predpossessive constructions

In an excellent recent paper in Linguistic Typology that I recomment to all linguists, Hilary Chappell and Denis Creissels argue against Leon Stassen’s (2005; 2009) classification of Mandarin Chinese predpossessive constructions as in (1).

(1) 她有书 Tā yŏu shū. [3SG have/exist book] ‘She has a book.’

Chappell & Creissels want to say that this is a transitive construction, like its English translation, and that the verb 有yŏu is used in its ‘have’ meaning here. They contrast this with Stassen’s classification as a “Topic Possessive”, both in his (2005[2013]) WALS chapter, and in his 2009 book.

I found the argument persuasive, and the paper is really interesting and reads well. But do the authors show that Stassen was wrong? What does it mean to say that a typologist’s classification is right and another typologist’s classification is wrong, or at least that one classification is “better” than another one? This is not immediately obvious, because we do not make claims about innate grammatical knowledge (which might be said to be uniquely determined, so that certain claims about our genetic predispositions are wrong and others are right).

I think there is a clear sense in which universal claims can be wrong (or right). For example, Kayne’s Generalization (that clitic doubling is found only when the object is preceded by a preposition) is now generally agreed to be wrong because there are languages, such as Greek, which do not conform to it (e.g. Tsakali & Anagnostopoulou 2008: §3.1). And Dryer (1988) disproved the earlier claim that the order of adjective and noun correlates with the order of adpossessor and noun.

But can a typological classification system be wrong? Can the comparative concepts used in such classifications be wrong, or at least better or worse? Gilbert Lazard famously said that comparative concepts are cadres conceptuels arbitraires (“arbitrary conceptual scaffolds”) and I noted the similarities between his view of what typology is like and mine (Haspelmath 2010: 678). Lazard and I have been criticized for this, and people keep asking me whether comparative concepts are not sometimes right or wrong, or at least better or worse (rather than “arbitrary”). My answer is: Comparative concepts cannot be wrong (because they are concepts, not claims), but of course they can be better than others in the sense that they are easier to match with language-particular phenomena than others, or in the sense that they give rise to better cross-linguistic generalizations. As Lazard (2005: 8) put it: they are “tools for research, not hypotheses susceptible of being verified or falsified”. Tools can be good or less good, but only claims can be right or wrong.

But of course, the assignment of a language to a type (its typological classification) can be wrong. It would be wrong, for example, to say that Yoruba has the dominant word order S-O-V, or that Lezgian has dominant accusative alignment. So maybe there is nothing wrong with Stassen’s “Topic Possessive” type (i.e. this comparative concept), but he misclassified Chinese (and the other Southeast Asian languages that Chappell & Creissels discuss)?

Chappell & Creissels give very good arguments for regarding the predpossessive construction with 有 yŏu as a transitive construction, and for treating the possessor as a subject – in particular the fact that it can be questioned (谁有书 shéi yŏu shū? ‘Who has a book?’). One would not expect an element that occupies a special “dangling topic” construction to be questionable. And they give reasons for thinking that the original meaning of 有yŏu was ‘have’, not ‘exist’, so that its existential use (as in 有书yŏu shū ‘There is a book’) is a secondary development, much like in French, where il y a ‘there is’ is based on a ‘has’.

So did Stassen misclassify Chinese? If so, which languages did he classify correctly? It turns out that according to Chappell & Creissels (2019: §9), there is no language that has the Topic Possessive type, because such languages would not be able to express interrogation. In other words, the authors say that this type, as proposed by Stassen, is not possible and exists in no language. At this point, I asked myself: How would Leon Stassen react to this critique? I suspect that he would say that he simply defined “Topic Possessive” differently from the way Chappell & Creissels understood the term. In his WALS chapter, he says nothing about dangling topics, but states:

The Topic Possessive shares with the Locational and the Genitive Possessive the characteristic that the possessed NP is construed as the grammatical subject of the existential predicate. The distinguishing feature of the Topic Possessive lies in the encoding of the possessor NP, which is construed as the topic of the sentence. As such, the possessor NP indicates the “setting” or “background” of the sentence, that is, the discourse frame which restricts the truth value of the sentence that follows it. Its function can thus be paraphrased by English phrases such as given X, with regard to X, speaking about X, as far as X is concerned, and the like.

So Stassen would say that in Tā yŏu shū [3SG have/exist book] ‘She has a book’, the nominal shū ‘book’ is the subject, while is the topic. This is different from the definitions adopted by Chappell & Creissels (and discussed by them in some detail), but it does not mean that there is necessarily a problem. Stassen might say that whenever the ‘have’ predicator has the same shape as the existential predicator (i.e. can also mean ‘there is’), he regards the possessed nominal as a subject (by definition!), and the possessor preceding the predicator as the topic (again, by definition). There is no universally accepted definition of “subject” and “topic”, so he might say that he adopts these definitions, leaving the “dangling topics” aside. Then the Mandarin (and Southeast Asian) predpossessive construction would be a Have-Possessive according to Chappell & Creissels, and a Topic Possessive according to Stassen – there is no incompatibility (merely different terminology). To my taste, Chappell & Creissels’s terminology is more transparent, so I will follow them (rather than Stassen) in any future work, but I do not think that they have shown that Stassen was wrong.

(It is true that Stassen’s more detailed characterization of the Topic Possessive in (2009: §2.5) may be taken to imply a “dangling topic” view, but it is not very clear, and I think that a retreat to a position as sketched in the preceding paragraph would still be open to Stassen.)

So far, I have been assuming that Chappell & Creissels (like Stassen) follow the usual procedure of Greenbergian typology and classify languages by comparative categories that are not necessarily identical to the categories of particular languages, because there is no hint that they assume innate categories of a universal grammar blueprint. Thus, when they talk about how to classify Mandarin (and the other languages), this does not prejudge anything about the description or analysis of these languages. Maybe in some of the languages, the situation is different from Mandarin, and indefinite subjects behave differently from questioned subjects? It is a frequent experience in typology that test batteries lead to different results in different languages (cf. Croft 2009). And one cannot presuppose that the difference between dangling topics and clause-integrated elements is the same in all languages – these are comparative concepts themselves, and while they may be easy to identify, this needs to be studied carefully. Moreover, are we so sure that predicators like yŏu are polysemous? in other words, that the existential use (as in 有书yŏu shū ‘There is a book’) represents a different meaning from the transitive possessive use? It seems that a more cautious characterization in terms of coexpression is more appropriate in this situation (because it makes no claim of whether there are two meanings in the language or just a single broad meaning). And indeed, these items are (quite felicitously) glossed “EPP” (= existential/possessive predicator) in this paper.

So because of this indeterminacy of the description/analysis, I think it is a bit confusing that Chappell & Creissels sometimes use the verbs “classify” and “analyze” in a seemingly interchangeable way. But typological classification and language-particular description/analysis are different things. By classifying a construction, we have not described/analyzed it sufficiently. Description/analysis is generally based on other language-particular categories, not on the kinds of general concepts that our comparative notions must be based on. Just as we can misclassify a language (but nobody suggested that Stassen misclassified Mandarin), we can misdescribe a language – but again, Stassen did not misdescribe anything (his purpose was classification, not description). Maybe his choice of comparative concept was not very felicitous, and his description of the concept was not really complete.

To sum up, I would say that Chappell & Creissels’s contribution lies primarily in spelling out their reasons for classifying their languages in great detail (clearly superior to what we find in Stassen’s work). This gives us a much better sense of how languages differ, and their diachronic speculations are intriguing. I also find their terminology more felicitous than Stassen, so there is definite progress here. But “right and wrong” in typology concerns universal claims (of which we have none here), not the comparative concepts or the terminology.


Chappell, Hilary & Denis Creissels. 2019. Topicality and the typology of predicative possession. Linguistic Typology 0(0). doi:10.1515/lingty-2019-0016.

Croft, William. 2009. Methods for finding universals in syntax. In Sergio Scalise, Elisabetta Magni & Antonietta Bisetto (eds.), Universals of language today, 145–164. Dordrecht: Springer.

Dryer, Matthew S. 1988. Object-verb order and adjective-noun order: Dispelling a myth. Lingua 74(2–3). 185–217.

Haspelmath, Martin. 2010. Comparative concepts and descriptive categories in crosslinguistic studies. Language 86(3). 663–687. doi:10.1353/lan.2010.0021.

Lazard, Gilbert. 2005. What are we typologists doing? In Zygmunt Frajzyngier, Adam Hodges & David S. Rood (eds.), Linguistic diversity and language theories, 1–23. Amsterdam: Benjamins.

Stassen, Leon. 2005. Predicative possession. In Martin Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie (eds.), The world atlas of language structure, 474–477. Oxford: Oxford University Press. (2013 online version: http://wals.info/chapter/117).

Stassen, Leon. 2009. Predicative possession. Oxford: Oxford University Press.

Tsakali, Vina & Elena Anagnostopoulou. 2008. Rethinking the Clitic Doubling parameter: The inverse correlation between clitic doubling and participle agreement. In Dalina Kallulli & Liliane Tasmowski (eds.), Clitic doubling in the Balkan languages, 321–357. Amsterdam: Benjamins.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.