[Lingtyp] Query re anaphoric object pronouns

Geoffrey Haig geoffrey.haig at uni-bamberg.de
Tue Dec 17 13:26:35 UTC 2019

Dear All.

I sense that Martin's original question has been sidetracked in the 
ensuing discussion, so I'll contribute some thoughts on that, with some 

The original query concerned a possible unidirectional correlation 
between (a) obligatoriness of object pronouns, and (b) bound versus free 
form of expression:

(1) "In almost all languages, if the anaphoric object pronoun is 
obligatory, it is a bound form (= a form that cannot occur on its own, 
i.e. an affix or a clitic)."

As several people have mentioned, there are a couple of problems with 
this, most notably the notion of 'obligatory' and the notion of 'bound'. 
In particular 'obligatory' is problematic, because in fact there are 
very few languages where object pronouns are always required (even in 
English, objects that are propositional following verbs such as KNOW, 
HEAR etc. are not obligatory; perhaps these sentential objects could be 
taken as the bottom line in an animacy scale); most object indexing is 
conditioned (Haig 2018), e.g. via animacy, topicality etc., though if 
you want to consider rule-governed conditioning as a special case of 
obligatory, that's ok with me.

An alternative approach to this is a corpus or token-based approach, 
where we might re-phrase the tendency as follows:

(2) Zero expresssion of objects is overall less frequent in languages 
with bound object pronouns than with free object pronouns.

There has been quite a lot of corpus-based work on zero anaphora with 
objects recently. Schnell & Haig (2014) looked at object pronouns, bound 
and free, in a small sample of language corpora - too small to draw any 
definitive conclusions, but we noted among other things a contrast 
between two closely related languages (Central Kurdish and Northern 
Kurdish), which differ in that the former has clitic object pronouns, 
while the latter lacks them. The rate of zero objects (3rd person only) 
was indeed quite different. The relevant figures are (Tables 6 & 7, 

Central Kurdish (with bound object indexing): 3% zero object pronouns 

Northern Kurdish (no bound object indexing): 17% zero object pronouns 

This is obviously very tentative, but I suspect that the general 
principle suggested in (2) would merit closer investigation.

A more detailed and methodologically more sophisticated case study of 
zero pronouns in discourse is Schnell & Barth (2018) on Vera'a (Oceanic).

More recently we have compared rates of zeroes in various functions, 
including objects, across the language corpora in Multi-CAST (the raw 
data are all available in Multi-CAST, Mandarin is on its way). 
Interestingly enough, Mandarin turns out to be not an extreme case - we 
had higher rates on Sanzhi (Nakh-Daghestanian, Tondano (Phillippine 
type) and Northern Kurdish (Iranian, Indo-European). The real outlier in 
our sample was English with its exceedingly low level of zeroes (see 
Vollmer 2017 and ongoing work in the Multi-CAST framework).

Season's greetings


Haig, Geoffrey. 2018. The grammaticalization of object pronouns: Why 
differential object indexing is an attractor state, /Linguistics /56(4): 
781–818, DOI: https://doi.org/10.1515/ling-2018-0011

Multi-CAST online resource,  citable as: Haig, Geoffrey & Schnell, 
Stefan (eds.), /Multi-CAST: Multilingual corpus of annotated spoken 
texts./ (multicast.aspra.uni-bamberg.de/) (date accessed)

Schnell, Stefan & Danielle Barth. 2018. Discourse motivations for 
pronominal and zero objects across genres in Vera’a/. Language Variation 
& Change /30.1, 51-81.
Schnell, Stefan & Geoffrey Haig. 2014. Assessing the relationship 
between object topicalisation and the grammaticalisation of object 
agreement. /Selected Papers from the 44th Conference of the Australian 
Linguistic Society, /2013, edited by Lauren Gawne and Jill Vaughan, 

Vollmer, Maria. 2019. How radical is pro-drop in Mandarin? A 
quantitative corpus study on referential choice in Mandarin Chinese. 
MA-Thesis, University of Bamberg.

I think a couple of terms in this

Am 17.12.2019 um 05:51 schrieb Randy LaPolla:
> Hi All,
> Can I make an appeal for using natural data in discussing this (and 
> everything else)? Even for those who distinguish morphosyntax and 
> pragmatics, this is clearly a situation where one has to take into 
> account the communicative situation and the intention of the speaker.
> It is very easy to find natural examples. Here is one from the first 
> Chinese web page I looked at 
> (https://m.ppzuowen.com/book/xiaomaoriji/lansedetuerduocao/53393.html)
> No anaphor:
> “哪里有死神?我怎么没看见?”
> “Where is there (a) Death Spirit? How is it I never saw (one)?
> with anaphor (same story):
> 整整一个晚上,我就这么守护着雨樱,决不让死神再靠近她。
> ’The whole night I watched over Yu Ying this way, no way would I allow 
> the Death Spirit to come near her. ‘
> In this case it would have been possible for the writer to not use the 
> anaphoric pronoun, but the meaning would be different: with the 
> pronoun the writer restricts the interpretation of who the Death 
> Spirit would not be allowed to come near to Yu Ying, whereas without 
> the pronoun the interpretation would not be constrained in this way, 
> and so could be understood as ‘near us’.
> Randy
> Sent from my phone
>> On 17 Dec 2019, at 6:19 AM, Alex Francois <francois at vjf.cnrs.fr> wrote:
>> dear Chao,
>> I realise that the examples you cite for animate pro-drop all follow 
>> a particular structure, namely, they are responses to polar 
>> questions, in which the verb is simply repeated, often dropping its 
>> subject and/or its object.  Yet I agree with David's point, that this 
>> is a specific pragmatic context, which may have its own rules, in 
>> languages like Hebrew and Mandarin.
>> So it's possible that we are not actually carrying out the same test 
>> in all languages.
>> If we come back to Martin's question, I believe we should first agree 
>> on a particular syntactic context to be tested.  This would make our 
>> data comparable across languages, and give stronger value to our 
>> generalisations ("structure X is allowed in language A but not in 
>> language B").
>> In this case, the test could be defined as follows:
>>   * The test sentence must have a transitive verb, which is _not_ the
>>     mere repetition of a previous verb (as in a reply to a polar
>>     question).
>>   * Its grammatical object is a participant that is already activated
>>     in discourse (topical), and is retrieved anaphorically.
>>   * Can this object be zero-expressed?
>>   * In each language, the test could be carried out with
>>         (1) a speech act participant
>>         (2) a human referent
>>         (3) a non-human, animate referent
>>         (4) an inanimate referent.
>> Here would be possible questionnaire sentences:
>>     (*1*) My sister knows you already. She saw [[you]] last month at
>>     the party.
>>     (*2*) You know my sister already.  You saw [[her]] last month at
>>     the party.
>>     (*3*) You know my cat already.  You saw [[it]] last month in my home.
>>     (*4*) You do know that song.  You sang [[it]] last year in class.
>> Mwotlap (Vanuatu) has obligatory expression of the object for 
>> sentences (1)–(2)–(3), using free pronouns;
>> it has obligatory dropping (zero expression) of the object in (4).
>> English and French have obligatory expression of the object in all 
>> four sentences.
>> What about Mandarin? Hebrew? other languages?
>> According to Jürgen's message, Mayan would have segmental realisation 
>> of the object in (1), but zero in (2)–(3)–(4).
>> I propose the following hypothesis (which needs to be tested):
>> Languages will locate the boundary between overt and zero expression 
>> of the object  somewhere along that scale {1>2>3>4}, with overt 
>> expression to the left and zero to the right.
>> best
>> Alex
>> ------------------------------------------------------------------------
>> Alex François
>> LaTTiCe <http://www.lattice.cnrs.fr/en/alexandre-francois/> — CNRS– 
>> <http://www.cnrs.fr/index.html>ENS 
>> <https://www.ens.fr/laboratoire/lattice-langues-textes-traitements-informatiques-et-cognition-umr-8094>–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>
>> Academia page <https://cnrs.academia.edu/AlexFran%C3%A7ois> – 
>> Personal homepage <http://alex.francois.online.fr/>
>> ------------------------------------------------------------------------
>> On Mon, 16 Dec 2019 at 22:29, Chao Li <chao.li at aya.yale.edu 
>> <mailto:chao.li at aya.yale.edu>> wrote:
>>     Dear Alex and Martin,
>>     I agree with Alex, Eitan, and Volker that animacy is an important
>>     factor when all the contexts of object-drop are examined. At the
>>     same time, I would like to add that, with respect to Mandarin,
>>     animate objects may also be naturally dropped, and this is
>>     particularly true of conversational contexts, as shown in (1-2).
>>     (1)A: 你喜欢他(/她)吗?
>>     Nǐ xǐhuan   tā (/tā)         ma?
>>     you  like        him(/her) Question.Particle
>>     ‘Do you like him(/her)?'
>>     B: 当然喜欢啦。
>>     Dāngrán     xǐhuan   la.
>>     of.course    like Sentence.Final.Particle
>>     '(I) of course like (him/her).'
>>     (2) A: (holding a picture)
>>     你见过这个人吗?
>>     Nǐ      jiàn-guo zhè-gè              rén         ma?
>>     you   see-Experiential    this-Classifier person   Question.Particle
>>     'Did you see this person before?'
>>     B:  我没见过。
>>     Wǒ   méi    jiàn-guo.
>>     I       not     see-Experiential
>>     'I didn't.'
>>     Best regards,
>>     Chao
>>     On Mon, Dec 16, 2019 at 10:30 AM Alex Francois
>>     <francois at vjf.cnrs.fr <mailto:francois at vjf.cnrs.fr>> wrote:
>>         dear Martin,
>>         One dimension you forgot to mention, and which appears
>>         crucial to me for any study of that sort, is animacy.
>>         I can think of many languages where zero-anaphora is allowed
>>         (or even the rule) for inanimate patients, while it would be
>>         less possible for animate ~ human ones.
>>         I think this is true, for example, for the *Mandarin *example
>>         you cite:
>>             今天早上我见到了*她*。
>>             Jīntiān zǎoshang  wǒ   jiàndào le *tā*.
>>             today morning   1sg  see     PFT  3sg:(Fem)
>>             'I saw _her_ this morning.'
>>             今天早上我见到了。
>>             Jīntiān zǎoshang  wǒ  jiàndào le ∅.
>>             today  morning   1sg see     PFT ∅
>>             'I saw _it_ this morning.'
>>         (Chinese speakers and experts, please correct me.)
>>         *Mwotlap *(and other northern Vanuatu languages) would be
>>         similar:   Zero anaphora is the norm for non-human objects,
>>         but not expected for human objects:
>>             No  m-eksas *kē * aqyig        lemtap
>>             1sg PFT-see   3sg  today:Past morning
>>             'I saw _her_ this morning.'
>>             No m-eksas *∅*** aqyig  lemtap
>>             1sg PFT-see ∅  today:Past morning
>>             'I saw _it_ this morning.'
>>         I don't think that Mandarin /tā/ qualifies as a clitic;  nor
>>         does Mwotlap /kē/.
>>         Insofar as they are obligatorily expressed for animate
>>         patients, then these cases would constitute, like English,
>>         exceptions to the general principle you're proposing.
>>         best
>>         Alex
>>         ------------------------------------------------------------------------
>>         Alex François
>>         LaTTiCe <http://www.lattice.cnrs.fr/en/alexandre-francois/> —
>>         CNRS– <http://www.cnrs.fr/index.html>ENS
>>         <https://www.ens.fr/laboratoire/lattice-langues-textes-traitements-informatiques-et-cognition-umr-8094>–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>
>>         Academia page <https://cnrs.academia.edu/AlexFran%C3%A7ois> –
>>         Personal homepage <http://alex.francois.online.fr/>
>>         ------------------------------------------------------------------------
>>         On Mon, 16 Dec 2019 at 15:13, Haspelmath, Martin
>>         <haspelmath at shh.mpg.de <mailto:haspelmath at shh.mpg.de>> wrote:
>>             I have a question on 3rd person anaphoric pronouns in the
>>             world’s languages. In many languages, these are optional
>>             when they refer to a continuous topic, not only in
>>             subject (S/A) role, but also in object (P) role. So we
>>             get patterns like the following:
>>             – Have you seen Lee today?
>>             – Yes, I met (her) in the cafeteria.
>>             I’m wondering if the following universal tendency is true:
>>             (U) In almost all languages, if the anaphoric object
>>             pronoun is obligatory, it is a bound form (= a form that
>>             cannot occur on its own, i.e. an affix or a clitic).
>>             Spanish and Arabic are examples of languages where the
>>             obligatory anaphoric object forms are bound (clitic or
>>             affix). English and German are exceptions to this
>>             generalization (and perhaps a few other European
>>             languages as well).
>>             But are there many exceptions? According to Siewierska
>>             (2004: 43), about two thirds of all languages (223 out of
>>             378 in her WALS chapter <https://wals.info/feature/102A>)
>>             have bound object person forms (= object indexes), so the
>>             hypothesized universal tendency is a question about those
>>             languages that lack object indexes, and have only
>>             independent personal pronouns or demonstratives for
>>             object function. Are there many among them which (like
>>             English) obligatorily require an overt form in this function?
>>             Or are most of them like Mandarin Chinese, which
>>             according to Wiedenhof (2015: §5.2.2) happily allows
>>             zero-anaphora sentences like /Nǐ yào ma?/ [you want Q]
>>             ‘Do you want it?’
>>             I’m interested in all reports of languages outside of
>>             Europe which are unlike Mandarin, and like English, in
>>             this respect.
>>             Many thanks,
>>             Martin
>>             -- 
>>             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
>>             _______________________________________________
>>             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
>> _______________________________________________
>> Lingtyp mailing list
>> Lingtyp at listserv.linguistlist.org
>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp

Prof. Dr. Geoffrey Haig
Lehrstuhl Allgemeine Sprachwissenschaft
Universität Bamberg
Schillerplatz 17
96047 Bamberg

Tel. ++49 (0)951 863 2490
Admin. ++49 (0)951 863 2491


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20191217/3c14bf7c/attachment.htm>

More information about the Lingtyp mailing list