WORKSHOP ON MORPHOLOGY AND LFG

Motivation for Workshop

Contemporary lexicalist theories share several guiding principles including the strict separation of syntax and morphology and the rejection of derivational operations in the structural syntax: these leading ideas find expression as lexical integrity and the principle of direct syntactic encoding/monotonicity in LFG. Crucially, in LFG it is recognised that this separation of syntax and morphology applies only to the structural domain: syntax and morphology are equal, interacting and competing contributors in the functional domain.

These architectural assumptions have led to a considerable amount of exciting research on the morphology-syntax interface within LFG and the theory has been shown to accommodate insightful analyses of phenomena such as complex predicates, split inflection, head mobility, noun incorporation. In particular, much recent work has focussed precisely on the morphological expression of functional information itself in languages with rich morphology (and less rich syntax), raising many interesting new issues. Andrews (1996) and Nordlinger (1998) explore the way in which Case markers specify GFs and associate f-structure information with morphemes using inside out equations. Such an approach presupposes the existence of non-semantic Case features such as ERG at f-structure, essentially reflecting the choice of means of exponence in the f-structure. Phenomena such as case stacking motivate a Principle of Morphological Composition to define longer paths within the f-structure. However if there are no morphemes, interesting questions arise as to how such a principle should be defined.

Given the central role rightly ascribed to morphological means of expression in the theory, it is a striking fact that there has been very little discussion of what a morphological component for LFG would look like. Despite the central importance of this question, the strict structural separation of syntax and morphology itself has permitted and encouraged simplifying assumptions to be made about the nature of the lexical/morphological component. Indeed, most linguistic work has tended to ascribe functional properties to individual morphemes by means of equational statements associated with nodes/terminals elements in the familiar fashion.

However, there are several reasons why this approach is inadequate.

First, at least as far as inflectional morphology is concerned, there is wide acceptance among morphologists ofthe superiority of the 'realizational-inferential' or 'Word-and-Paradigm' approach (over tree-based, 'word-syntax' approaches). Realizational (or WP) morphology denies that morphemes are signs (form-content pairs). Instead, inflections are stem modifications which serve as exponents of morphological feature sets.

Second, it can be shown that the vocabulary of features needed for morphology is distinct from that needed for morphosyntax, and that there is a quite complex relationship between these sets of features. This means that one cannot simply assume one (f-structural) vocabulary of features such as TNS, PER, NUM, VOICE, etc, and associate them with elements or operations of the morphology. Synthetically expressed tenses such as those involving the combination of a functional category with a lexical category provide a simple example of one aspect of the difficulty here: past tense may be expressed by a combination of a morphologically present auxiliary with a morphologically tenseless participle: these vocabularies are conceptually distinct. It is not desirable (and often not possible) to conflate the two, reflecting the morphological feature content of inflected word forms in f-structure. Butt et al. (1996) and Frank and Zaenen (1998), discussing auxiliary constructions, have argued for a separation of c-structure, f-structure and m-structure. Ackerman and Webelhuth (1998) argue that our definition of the lexicon should be extended to associate lexical representations with such syntactically transparent constructions and express this within an HPSG formalism. It can also be shown that a periphrastic construction can be an exponent of purely morphological features (i.e. syntax realizes pure morphology) (Spencer and Sadler (1999), Stump (in press)). (Such phenomena pose severe problems for current theories based on underspecified morphemes: Steele's Articulated Morphology, Halle \& Marantz's Distributed Morphology, Wunderlich's Minimalist Morphology.)

Issues

Against this background, this workshop addressed a number of issues relating to the role and form of the morphological component and the morphology-syntax interface in LFG (and generated much heated discussion in doing so). Taking as our starting point the requirements placed by the nature of morphological operations and entities themselves, we planned to meet the following objectives in the workshop:

Workshop Outline

Contributors: Farrell Ackerman, Kersti Borjars, Anette Frank, John Payne, Louisa Sadler, Andrew Spencer.

Session One Architectural Issues

Session Two Nominal Structure

Session Three Case Studies in WP and LFG

Topics for Discussion