On Thursday I went over to Berkeley to hear a summer lecture by Nick Evans of the University of Melbourne. Lots of fun stuff--though I'll admit it was easier to understanding everything at the beginning than it was to follow-along in parts of the middle. One of Berkeley’s newest faculty members, Alice Gaby, is a student of Evans. Evans will be heading up the Summer Institute of Linguistics that's at Berkeley in 2009.

 

Among all the countless things and classes that there are, most are miscellaneous, gerrymandered, ill-demarcated. Only an elite minority are carved at the joints, so that their boundaries are established by objective sameness and difference in nature. (Lewis 1984: 227)

 

Evans points out that if you have two participants in a reciprocal event, you've got predicate calculus, but if you go to multiple participants, you have to start using lambda notation. No, I don't really know what that means. But part of the point was that he'd talk about two people in reciprocal events rather than three or more.

 

Towards the end of his talk, he had a phonological aside to show how autosegmental analyses “overcomes the tyranny of the linear.” Linearity is a problem in reciprocals if you’re mapping thematic roles to syntactic arguments since there get to be a lot of criss-crossing lines. See the Amele example, below.

 

Evans is currently analyzing/publishing findings based on a bunch of Western Pacific languages—he and his students have look at existing corpora, recorded gossip (a good source of reciprocals) and had people watch videos and explain who-was-doing-what.

 

All examples, unless otherwise noted, come from Evans. I’ve preserved his numbering scheme for my own convenience and cross-reference to the hand-out he had.

Typical ways to represent reciprocals

Simultaneous reciprocal:

(1a)      John and Mary love each other / one another.

love (j, m) & love (m,j)

 

Sequential reciprocal:

(1b)      John and Mary groomed / massaged / deloused each other /one another.

[delouse (j, m)]_t1 & delouse [m, j)]_t2

 

You end up having one clause to represent two (or more) events. The “thematic roles” (if you believe in using ‘em, cf. Levin) of the participants are permuted. That means that there’s a double-linking of thematic roles to argument position.

The Amele way of doing reciprocals

On the other hand, the Amele adds some element representing coordinated activity, mutual causation.

 

            (3a)      Age     qet-u-do-co-b              qet-u-do-co-b              eig-a

            AME   3pl       cut-pred-3sg-DS-3sg   cut-pred-3sg-DS-3sg   3pl-tod.pst

                        ‘They cut each other.’ (Roberts 1987: 132)

 

“Essentially, ‘They, he₁ cuts him₂, he₂ cuts him₁ they did,’ says Evans. (Also, ‘We he₁ despise he₂ he₂ despise he₁ we did’.)

Theoretical assumptions

Here are the theoretical assumptions of the talk, which fit in with the leit motif of the talk, typology and diversity.

 

Standard semantic representations need to apply to all lexical forms that can occur in the construction

but

The form of constructions may be motivated by prototypical semantic components, which need not hold for all lexical items occurring in the construction. (Evans 2006 handout: 3)

 

Fun ways of denoting complex events in Kalam

The English verb ‘massage’ is pk wyk d ap tan d ap yap g-, literally, ‘strike rub hold come ascend hold come descend do’.

How do deal with overcrowding

With complex events there are too many semantic roles to put onto syntactic argument positions.

 

Evans notes three possible solutions:

            (a)        keeping the clauses distinct

            (b)        finding a common (macro-)role that can underly integration of two verbs

            (c)        having a role hierarchy that linearizes complex roles into one dimension

Monoclausal vs. multiclausal approaches to reciprocals

Evans says the majority of the world’s languages probably adopt a monoclausal approach to reciprocals, where ‘each other’ is a pronoun (cf some problems in Modern Greek with case).

 

(11a)    A new report on the state of language teaching in France and Germany shows that the grasp of each other’s languages is in decline.

            Guardian Weekly, Feb 19-25 2004, TEFL Supplement, p. 1.

 

That’s an example of “reciprocal anaphora”, where you fill the “lower” syntactic argument position with a shorthand that isn’t exactly a repetition.

 

There are also monoclausal reciprocal strategies that use verbal derivation. For example Kaydo (sp?) from Australia or Mundari (sp?) from India. Note that body parts tend to take the same case as the people they belong to. See some exceptions in Evans 2006: 8.

 

Compare this to the multiclausal approach. You could chain the events together as Golin does (Chimbu family, Papuan). (Amele chains events together, too, but problematically—see (3a) above.)

 

            (30)      Abal                su         alemile                         abal                  ta         gibl-in     

           GOL      woman            two      be.standing:SEQ         woman              one      head-3

                       

igin                  aato-n-g-w-e                abal                  ta         kwi                  abal     

                        hair-3               touch-3-Ass-3-Dist     woman              two      in-return          woman

                       

                        ta         gibl-in              igin      aato-n-g-w-e                eri-n-g-w-e

                        one      head-3             hair-3   touch-3-Ass-3-Dist     go-3-Ass-3-Dist

 

                        ‘Two women are touch each other’s hair.’

 

Lit., ‘two women are standing and one woman touches her hair and then in return one woman touches her back.’

 

You can also merge cluases together, as in Mandarin.

 

            (31)      Tamen             da-lai-da-qu

            MAN   they                 hit-come-hit-go

                        ‘They hit each other.’ (Liu 1999: 124)

 

One of the really fun things that happens in Iwaidja/Mawng is that you do something basically like ‘he saw her and she in turn’, where ‘she’ is a special free-form version of that pronoun and where you indicate that the original verb is getting repeated but you just use a short-cut phrase. (If you have an event like give, the thing that is given stands outside the clause.)

 

            (40c)    anb-uku-n                                lda       wamin             a-ngurnaj

                        3plA>3plO-give-NPST           and      3plCONTR     3pl-name

                        ‘They used to give each other their (clan) names.’

 

Each Iwaidja clan has its own set of names, but sometimes clans would trade names with each other. The sentence here is basically, ‘They₁ gave them₂ and they₂ in turn, names.’

What’s the conclusion?

 

Each people leaves some things unaid in order to be able to say others. Because everything would be unsayable. (Ortega y Gasset 1957)

 

There are two ways to map semantics to grammar, says Evans:

 

(a)        compositional – every semantic component maps, compositionally, onto a grammatical component (rule-for-rule hypothesis etc)

(b)        constructional – construction is a gestalt, with a meaning directly linked to it; the form of the construction may be motivated by only a subset of the semantic components of the meaning it expresses. (Evans 2006: 11)

 

The first is basically Montague, the second says you don’t have to motivate every piece—some forces win out over others—not all semantics get in compositionally.

 

Cross-linguistically: the greater the variation in how the same meaning is expressed, the harder it is to maintain a compositional approach for all languages involved (i.e. if a structure in language A has construction α, then why does the same meaning get expressed as β in language B, etc.?) (Evans 2006: 11)

 

The full abstract and details are below:

 

Complex Events, Propositional Overlay, and the Typological Diversity of Reciprocal Strategies

 

Nicholas Evans

University of Melbourne

http://www.linguistics.unimelb.edu.au/about/staff/profiles/evans/

 

ABSTRACT

The lexicalization of event-denoting expressions has emerged as the area of semantics with the most extreme cross-linguistic variation. This makes the mapping between the semantic level of 'event' and the syntactic level of 'clause' one of the greatest challenges to semantics, syntax and typology. One dimension of this problem is working out how much complexity, in terms of event structure, can be accommodated within a single clause. Cross-linguistic studies of causatives, benefactive and instrumental constructions, and motion events have showed us that single-clause English expressions like 'I dropped the cup', 'I baked Mary a cake', 'I cut the bread with a knife' and 'The ball rolled down the hill' are all syntactically decomposable in some languages into multi-clause expressions of the type 'I made/let the cup fall', 'I baked cake gave Mary' and 'The ball descended the hill, rolling'. A wide range of languages employ such strategies, through methods like syntactic causatives, serial verb constructions or manner-framing through participial or other means; typological work here has taught us a great deal about how to motivate complex syntactic behaviour from the semantics of event decomposition.

 

Reciprocal constructions are a further type of complex event that maps to a single clause in English, as in many other languages. Their semantic characterisation requires at least two propositions, with arguably a third in prototypical cases: the meaning of 'John and Mary kissed each other' will be represented, logically, by the conjoined propositions 'John kissed Mary, and Mary kissed John', with symmetric exchange of referents between argument roles, and there are strong reasons to postulate a third proposition along the lines of 'John and Mary did this together / interacted'.  Yet little typological work on the syntax of reciprocals has sought to ground the unusual syntactic features of these clauses in their complex event structure,  in a way that relates these to other typological work on causatives, benefactives and motion-event-decomposition in  'event-atomizing languages', even though a large number of languages employ serial verb constructions for reciprocals just as they do for other types of complex event.  For example in Golin, a Chimbu language of the Papuan highlands, the same overall strategy that produces multi-verb representations for benefactives or motion events does the same for reciprocal situations, by reporting them as complex symmetrical pairings of subevents.

 

In the cross-linguistic study of reciprocal constructions that I report on in this paper, I will discuss widespread perturbations of clause structure found in reciprocal clauses, essentially resulting from an 'overlay' of symmetrically mirrored events, that  include apparently contradictory signs of transitivity,  apparent violations of principles producing a one-to-one mapping of NPs to thematic role, 'sesquiclausal' constructions that appear to hover between one and two clauses in size, and violations of binding conditions through the use of non-anaphor pronouns in the same clause as their antecedent.

 

Effects like these raise the question of how many features claimed to be universals of clause structure (such as biunique mapping between thematic roles and syntactic argument positions) are really syntactically motivated. A fuller typological stocktake of reciprocal constructions suggests that,  instead, they may be epiphenomena of limits to how much event complexity may be mapped onto a single clause; when, as happens with reciprocals in some languages,  the clause exceeds a certain level of complexity, these syntactic constraints no longer hold.