summary: associative plurals via noun-verb disagreement

[David Gil]

Dear all,

On November 13 I posted a query which began as follows ...

"I am interested in the cross-linguistic distribution of a construction 
type in which an associative plural meaning, eg. 'John and his 
associates', results from a singular noun triggering plural number 
agreement on the verb, as illustrated in the following examples from 
Roon (an Austronesian language spoken in the Cenderawasih bay of New 
Guinea):

(1) Amos-i i-berif
Amos-PERS 3SG:ANIM-laugh
'Amos is laughing'

(2) Amos-i su-berif
Amos-PERS 3DU:ANIM-laugh
'Amos and his friend are laughing'

(3) Amos-i si-berif
Amos-PERS 3PL:ANIM-laugh
'Amos and his friends are laughing'

Example (1) shows ordinary agreement, with a singular subject triggering 
singular verb agreement. However, examples (2) and (3) illustrate how an 
associative plural interpretation is derived via disagreement, with the 
still-singular subject occurring in construction with dual- and 
plural-subject marked verbs respectively.We might therefore call the 
construction in (2) and (3) an Associative Plural via Disagreement, or 
ASPD.

My question is: how common is this ASPD construction in the languages of 
the world? ..."

In response to the above query, I received lots of information on other 
languages; a summary of this information can be found in Section 1 
below. In addition, the query triggered an extensive discussion between 
myself and a number of colleagues concerning the distinction between 
associative plurals and inclusory constructions, a summary of which is 
presented in Section 2 below. Embedded in Section 2 you'll also find two 
additional questions (of the 'can you think of any examples of' kind) 
with respect to which I would be grateful for any further information.


SECTION 1:

*Languages with a construction resembling (2) or (3):*
Russian, Talitsk -- [ Corbett (2000), citing Bogdanov (1968) ]
Maltese -- [ Corbett (2000), citing Fabri (pc, 1993) ]
Mori Bawah -- [ David Mead (pc) citing Esser (1927-1933) ]
Haruai -- [ Corbett (2000), citing Comrie (pc) ]
Bininj Gun-Wok -- [ Evans (2003) ]
Plains Cree -- [ Daniel and Moravcsik (2005) ]
Diuxi-Tilantongo Mixtec -- [ Kaius Sinnemäki (pc), citing Kuiper and 
Oram (1991:238) ]

*Languages with a construction corresponding to (2) or (3) but where the 
number marking on the verbal complex is periphrastic:*
French ("very substandard") -- [ Alex François (pc) ]
Jakarta Indonesian ("only some speakers")-- [ Tom Conners (pc) ]
Papuan Malay -- [ own field work ]

*Languages with verbal number marking but without a construction 
resembling (2) or (3):*
English -- [ own knowledge ]
Irish -- [ Elisa Roma (pc) ]
Basque -- [ Miren Lourdes Oñederra Olaizola (pc) ]
German, Allemanic -- [ Frans Plank (pc) ]
Czech -- [ Viktor Elšik (pc) ]
Romany (Slovak) -- [ Viktor Elšik (pc) ]
Russian -- [ Michael Daniel (pc) ]
Hungarian -- [ Edith Moravcsik (pc) ]
Finnish -- [ Matti Miestamo (pc) ]
Hebrew -- [ own knowledge ]
Chechen -- [ Johanna Nichols (pc) ]
Ingush -- [ Johanna Nichols (pc) ]
Azeri -- [ Don Stilo (pc) ]
Vafsi -- [ Don Stilo (pc) ]
Persian -- [ Don Stilo (pc) ]
Urdu -- [ Elena Bashir (pc) ]
Belhare -- [ Balthasar Bickel (pc) ]
Tagalog -- [ Schachter and Otanes (1972:335) ]
Agutaynen -- [ Steve Quackenbush (pc) ]
Javanese -- [ Tom Conners (pc) ]
Mwotlap -- [ Alex François (pc) ]

My hope was to identify typological correlates, or predicators, for the 
construction in question; but even the limited sample of languages above 
suggests that this may be a futile quest: at least the following three 
pairs of languages can be identified which are typologically similar (as 
well as genealogically related), but in which one member of the pair has 
the construction while the other hasn't:

Talitsk Russian - Standard Russian
Maltese - Hebrew
Jakarta Indonesian - Javanese


SECTION 2:

In their responses, a number of colleagues suggested that the 
construction in question is an "inclusory" construction. In a follow-up 
posting, on November 15, I poured cold water on the idea, arguing that, 
while in other cases there may be overlap between associative plurals 
and inclusories, there is no evidence that constructions such as those 
in (2) and (3) are inclusory. Well, apologies to those whose suggestions 
I dismissed: Alex François has since provided me with the obvious 
knock-down argument that (2), at least, is inclusory; since (2) is 
understood as referring to Amos plus one other person, the dual prefix 
/su-/ can only be construed as including /Amos/ in its reference. In 
fact, Alex's dual argument extends also to ordinary associative plurals 
formed with a 3PL pronoun, such as the following from Papuan Malay:

(4) Amos dong
Amos 3PL
'Amos and his friend(s)'

Crucially, (4) may be understood as referring to two persons; and in 
this case an inclusory analysis for the plural pronoun /dong/ is forced. 
At this point, an empirical cross-linguistic question arises:

QUESTION 1: In other languages with an associative plural of the form 
"[proper noun] 3PL", can the construction also refer to exactly two 
persons (as in (4) above)? (I would be particularly interested in 
languages in which the construction contains a (synchronic or 
diachronic) coordinator, "Amos 'n them", as in such cases one might 
perhaps expect the construction as a whole to refer to a minimum of 
three people, Amos plus the referents of the pronoun, which would then 
rule out an inclusory analysis.)

Over the last couple of weeks, Alex François, Misha Daniel, Edith 
Moravcsik and myself have been staying up late at night debating the 
relationship between associative plurals and inclusories. Following is a 
summary of what I have learned from our discussion. (The following 
reflects my own perspective on the discussion; its other participants 
may not necessarily agree with everything below, or feel that these are 
the most important points to emerge from the discussion.)

An /associative plural/ construction consists of two formal elements, a 
referring expression /A/ plus a grammatical marker /m, /with the 
resulting meaning 'A and A's associates'. The crucial semantic property 
of the construction is that its plural reference is *constituted* from 
the single designated member A, that is to say, they are conceptualized 
as being associated with A.

An /inclusory/ construction consists of at least two formal elements, a 
referring expression /A/ plus a pronoun or pronoun-like element /p/, 
where the reference of A is wholly included in the reference of p. (In 
addition, there may or may not be a third formal element /c/ marking the 
construction as inclusory.)

Associative plural and inclusory constructions may overlap, yielding the 
following three types (with examples of each):

Type A:
non-associative-plural inclusory
Russian /
my s Davidom
'/1PL with David-INSTR'
'David and me/us'

Type B:
associative-plural inclusory
Papuan Malay (4) above
Amos dong
Amos 3PL
'Amos and his friend(s)'

Type C:
associative-plural non-inclusory
Hungarian
Edith-ek
Edith-ASS.PL
'Edith and her friend(s)

Now for each of the above types, one may further distinguish between two 
morphosyntactic types: (a) /nominal/, such as the above three examples, 
where all of the constituent parts of the construction form a nominal 
group, or NP, or (b) /discontinuous/, where the grammatical marker 
and/or the pronoun occur in construction with the verb, either 
periphrastically, or as bound morphology on the verb. Examples of the 
latter discontinuous case:

Type A:
non-associative-plural inclusory
Bininj Gun-wok (Evans 2003:240)
... daluk ngal-mekbe bene-re-y ...
...woman FE-DEM 3uaP-go-PI
'...he and his wife were going ...'

Type B
associative-plural inclusory
Roon (3) above
Amos-i si-berif
Amos-PERS 3PL:ANIM-laugh
'Amos and his friends are laughing'

Type C
associative-plural non-inclusory
???

The Bininj Gun-wok case is the only one so far that I am familiar with 
of a discontinuous inclusory that is not also associative. As Nick Evans 
explains in his grammar, this example is taken from a story in which the 
topic of discourse is a particular man, and "the fact that he is living 
as a married man is introduced with" the inclusory construction, 
specifically the dual verbal prefix /bene-./ Since /bene-/ is dual, the 
construction is clearly inclusory; however, the discourse context 
clearly indicates that the semantics is not associative -- the dual 
reference of /bene-/ is not constituted by the woman, as would be the 
case for an associative plural.

In the absence of consistently detailed descriptions, it is hard to tell 
whether, in the languages listed in Section 1 above (Talytsk Russian, 
Maltese, etc.), the construction in question is non-associative Type A, 
like the Bininj Gun-wok, or associative Type B, like the Roon. In fact, 
it may turn out to be the case that in some or all of these languages, 
the same sentence may allow either non-associative or associative 
interpretations, depending on context. But what about Type C, so far 
unattested? ...

QUESTION 2: Are there any examples of a construction in which a singular 
argument A can occur in construction with a specific non-pronominal 
verbal marker (periphrastic or bound) to yield the interpretation 'A and 
A's associate(s)?

Answers to the above two questions, and any other comments, would be 
greatly appreciated. Again, thanks for all your comments and suggestions 
to date.

David

-- 
David Gil

Department of Linguistics
Max Planck Institute for Evolutionary Anthropology
Deutscher Platz 6, D-04103 Leipzig, Germany

Telephone: 49-341-3550321 Fax: 49-341-3550119
Email: gil at eva.mpg.de
Webpage:  http://www.eva.mpg.de/~gil/