From anaphoric pronoun to grammatical agreement marker: Why objects don't make it.

Abstract

The endpoint of the historical evolution of agreement marker from anaphoric person pronoun is the loss of referentiality on the part of the person marker and the obligatory presence of the nominal argument with which it agrees.Contrary to what might be supposed, such agreement, which I, inspired by Bresnan & Mchombo (1986, 1987), have termed grammatical, as opposed to anaphoric or ambiguous (grammatical and anaphoric) agreement, is cross-linguistically very rare.Moreover, among the attested instances of grammatical person agreement none involve object as compared to subject agreement.The present paper considers the distribution and formal realization of anaphoric, ambiguous and grammatical agreement markers in a sample of 272 languages and offers some tentative explanations for the existing asymmetry in regard to grammatical agreement.It is suggested that grammatical object agreement does not arise since ambiguous agreement, from which grammatical agreement evolves, is less common with objects than with subjects, and two of the potential sources of grammatical agreement, adherence to a verb-second constraint and phonological attrition are more likely to involve subjects rather than objects.`The woman saw the men.' b. nen-yau-a -?a 3PL-see -3SG-IND `She/he saw them.' (MacDonald 1990: 118) Dutch (3) a. Piet zie-t Kees elke dag.Piet see-2/3SG Kees every day `Piet sees Kees every day.' b. *(Hij) zie-t Kees elke dag. he see-2/3SG Kees every day `He sees Kees every day.'My cross-linguistic investigations of person agreement markers suggest that, whereas the markers of both anaphoric and ambiguous agreement may correspond to subjects as in the case of the suffix -i in the Macushi (1b) and the suffix -a in the Tauya (2a,b) and also objects as in the case of the prefix iin the Macushi (1b) and the prefix nenin the Tauya (2a,b), grammatical agreement markers are confined to subjects.In other words, there appear to be no languages with grammatical object agreement, as defined by the above tripartite agreement typology.Why languages should display grammatical subject but not grammatical object agreement is what this paper seeks to explore. 1 The discussion is structured as follows.Section 2 offers a more detailed account of the tripartite typology of agreement markers and relates it to the grammaticalization cline.Section 3 considers the cross-linguistic formal realizations of the three types of agreement markers.In section 4 the attested distribution patterns of bound anaphoric, ambiguous and grammatical agreement markers are presented.And finally in section 5 some tentative explanations are suggested for the occurring and non-occurring distributional patterns noted.