The mysterious case of em: Grammaticalization of a SVC in Fataluku

Fataluku, a Papuan language of East Timor with SOV word order, exhibits a morpheme em that occurs in transitive constructions with multiple objects. Em comes in three phonologically conditioned allomorphs, em (following a null pronoun), im (following a consonant), and m (following a vowel), and obligatorily marks the first object noun phrase (NP). It does not mark a natural class of semantic roles. Rather, it marks themes in certain cases as in (1) and instruments in others as in (2). The constituent order of these constructions is fixed; the two objects may not switch their position and em always attaches to the first object NP.

(1)   Jon                 a           ulur=im              Mari         aci-pura.
       Jon                  sbj        breadfruit=em    Mary        DIR-sell
       ‘John sells Mary a breadfruit.’

(2)   Jon                  a           hikar=im             pai        uca.
       Jon                  sbj         knife=em            pig        kill
       ‘John kills a pig with a knife.’

           This distribution of em is reminiscent of serial verb constructions (SVC) in the language, which has the form of [NP V1 (NP) V2], as shown in (3).

(3)   Tawa             [kafe      me         ma’o]
        3SG              [coffee   take       come]
        ‘He brings the coffee.’ (Literally, ‘He took coffee (and) come.’)

We argue that em originates from the verb eme ‘take’ as a result of grammaticalization from being a verb in a SVC to becoming a postposition. In fact, we show that this process is still ongoing: em is no longer a fully-fledged verb, nor is it fully functioning as a postposition.
           The semantic roles of the NPs to which em attaches suggests em originated from the verb eme of a SVC of the following patterns: [theme take][goal/location V2] and [instrument take][theme V2]. SVCs such as ‘I mango take Mary sell’ and ‘I knife take mango cut’ became ‘I mango-em Mary sell’ and ‘I knife-em mango cut’. This scenario also explains why em attaches to the instrument instead of the theme in sentences like (2). We argue that em is no longer a verb, as it is (a) semantically bleached, (b) limited in where it can occur, and (c) phonetically reduced, which are all characteristics of grammaticalized morphemes. Typologically common structures are more likely to grammaticalize. Thus, if Fataluku had a very frequent SVC using eme ‘take’ in ditransitive constructions, it would be susceptible to grammaticalization of a verb to a postposition.
           At the same time, em does not fully fit the profile of postpositions either. First, unlike em, postpositions are associated with a particular semantic role: hor comitative, hau beneficiary, and mucu locative.  Second, only em, but not postpositions, can be stranded if the object is a third person singular pronoun (and hence can be phonologically null) (4, 5).

(4)   ana                 Ø           em         tawa      ina
       1SG                3SG        em         3SG       give
        ‘I give (it) to him.’

(5)   *ana             Ø            hor         la’a
        1SG             3SG         with        go
       Intended: ‘I go with (him).’

           Fataluku em presents a unique opportunity to look into the transitional stage in grammaticalization of verbs into postpositions. By carefully studying this morpheme in a less documented language, this work contributes to the wider discussion of grammaticalization and the origin of postpositions.

This work is co-authored with Colleen O’Brien (University of Hawai‘i) and was presented at the 9th International Austronesian and Papuan Languages and Linguistics Conference held at Institut national des langues et civilisations orientales in Paris in June, 2017.