LFG2004 Workshop - Coordination and Agreement

[Peter Peterson]

Dear all

As part of the LFG 2004 conference in Christchurch, there will be a Workshop on "Coordination and Agreement". The following paragraphs outline the theme of the workshop and call for anyone interested in these issues to consider active participation. For further information, contact the Workshop convenor:
Peter.Peterson at newcastle.edu.au

Workshop on "Coordination and Agreement"

Recent LFG accounts of the syntax of coordination (Kaplan & Maxwell 1988, Dalrymple & Kaplan 2000, Peterson 2004) support an analysis in which coordinate structures are not headed but rather constitute sets.  At the level of f-structure, the set structure blocks the flow of information between constituents.  Lexical properties of individual conjuncts have no pathway by which they can percolate up to the node dominating the whole coordination structure. Conversely, external syntactic requirements cannot percolate down to individual nodes within the coordination.  This analysis clearly gives desirable results in some instances. For instance, it ensures that John is not SUBJ in John and Mary are sleeping; and it ensures that number features of individual NPs within a coordinate Subject are irrelevant to the number value of the finite verb  - are is PLU; John, Mary are both SING.

However, the analysis (at least apparently) raises several general problems with respect to agreement.  For instance:
1.  If lexical features within a coordinate structure are insulated from the external syntax by the set structure, what determines the PLU feature on are in John and Mary are sleeping? What determines Case features on individual NPs within a coordinate NP?  Two different approaches to this question are outlined in Dalrymple & Kaplan 2000 and Peterson 2004 respectively.  Dalrymple & Kaplan's account, based on feature resolution, provides for an extra 'shell' of f-structure for resolved agreement features.  Peterson's account, based on feature distribution, assumes that no such 'escape hatch' is provided for agreement within the syntax, and that lexical features remain 'hidden' inside a coordination set.

2.  There are many examples in the literature of 'single conjunct agreement', where only one of the conjuncts carries the 'expected' agreement feature value; e.g. only the first of two conjoined NPs has NOM Case,  or only the second of two conjoined adjectives has the same gender value as the modified noun.  Single conjunct agreement poses problems for any current treatment of coordination.  If agreement can reach inside coordinate structures, it should affect all conjuncts equally; if agreement is blocked from applying inside coordinate structures, why is it not blocked from all conjuncts?

The aim of the Workshop is to make progress towards a better understanding of coordination and ideally to establish an algorithm for coordinating f-structures.  We therefore welcome input to the Workshop in the form of informal contributions (5 or 10 minutes in length) addressing theoretical issues or presenting problematic data on gender/number/person/case agreement patterns in relation to coordinate structures. Please contact Peter Peterson (address below) if you are interested in contributing.

References:
Dalrymple, M. & Kaplan, R. 2000. Feature indeterminacy and feature resolution. Language 76: 759-798.
Kaplan, R. & Maxwell, R. 1988. Constituent coordination in Lexical-Functional Grammar. Reprinted in Dalrymple et al (eds) 1995. Formal issues in Lexical-Functional Grammar, Stanford: CSLI.
Peterson, P. 2004. Coordination: consequences of a lexical-functional account. To appear in NLLT.

Peter Peterson, Coordination Workshop Convenor
email: Peter.Peterson at newcastle.edu.au