The aim of this paper is to cast doubt on the so called 'Lexical Integrity Hypothesis', which claims that syntax is blind to morphological information. We provide data concerning the behaviour of prefixes in the Phonological and Sematic Interfaces that is analysed as evidence for the necessity of allowing syntactic rules to look inside the internal structure of the word. The analysis proposes that words should be considered as phases, as outlined in Chomsky (2001). It also advances the possibility of defining phases at the morphological level.

Aspectual composition occurs when the eventuality type and the grammatical aspect are closely related and together they contribute meaning to a phrase. In Modern Greek (MG) the grammatical aspect is represented in the morphology of the verbal lexeme whereas the eventuality type is simple inferred by the semantics. Hence, I argue that in MG there is no straightforward mapping between elements of the surface morphology and the semantics, following Filip (2000) who reaches the same conclusion regarding the Slavic languages. The interrelation, though, of the grammatical aspect and the eventuality type is important for the meaning of the verbal lexeme. The grammatical aspect is taken to be a function which selects an eventuality type as argument. When the eventuality is not the appropriate input for the grammatical aspect, then the meaning of the eventuality type changes and aspect shifts occur. Following de Swart (1998), I provide explicit formalisation of these observations adapted in the Head-driven Phrase Structure Grammar framework which depicts the aspectual distinction in MG. The semantic representations are shown using Minimal Recursion Semantics, where the scopal relation between grammatical aspect and eventualities can be explicitly stated.

I examine class features as means of defining what motivates the application of morphological operations in Distributed Morphology Halle and Marantz (1993). I am interested in determining not only the form but also the function of class features based on evidence drawn from non-suppletive verbal forms in Modern Greek. I suggest that inflection class features are abstract, binary features and not privative ones (like Class I and Class II). Moreover, I claim that roots enter the syntactic component fully specified. This does not violate the Feature Disjointness Principle Embick (2000), as I propose that class features are invisible at the syntactic component which does not have the mechanisms to interpret features which are not syntacticosemantic. At the morphological component, class features trigger morphological operations, such as Lowering Merger (Embick and Noyer 1999), resulting in the formation of non-suppletive stems, as well as agree operations, merging the stem with the inflection markers (voice and tense/agreement).

This paper advocates two points: how to attain periphrastic inflections of grammatical features and how to formalise the realisation of dependency information. As an interaction of those two issues, I illustrate an analysis of English prepositional phrases. The frameworks I adopt in this paper are Lexical Functional Grammar (LFG) and Paradigm Function Morphology (PFM). LFG's functional projections are incorporated into PFM with a revision of the view 'morphology competes with syntax'. The marriage allows us to capture the periphrastic realisations of inflectional features. Further, following the insight of recent morphological and syntactic literature, I argue that adpositions are periphrastic realisations of grammatical functions of nouns. By associating an underspecified grammatical function with a noun lexeme, I show how the proposed realisation model of LFG and PFM naturally accounts for the multi-word realisation of dependency information.

This paper describes the morphological properties of Bulgarian verbal aspect with focus on those features that make it difficult to deal with this phenomenon in a straightforward fashion within the existing model of Paradigm Function Morphology (PFM). The paper argues that aspect is part of the paradigm of Bulgarian verbs, but only one of the morphological processes traditionally associated with it, that of va- suffixation, should be treated as a process related to the exponence of this category proper. The addition of the suffix va-, however, in some cases leads to a change in the conjugational class of a verb. This is taken as one argument why stems should play a significant role in the realization of aspect. A proposal is made for some tentative extensions to the existing stem-formation mechanisms of PFM which make it possible to add aspectual features to the paradigm space.

I describe a variant of Stump's (2001) Paradigm Function Morphology (PFM), Generalised PFM (GPFM). In the realization rules (RRs) of GPFM, exponence and linear placement are stated separately. The RRs define an affix set and the paradigm function (PF) aligns the affixes with respect to each other and a stem. The PF is defined over the whole lexical entry. This permits reference to any property of the lexeme, not just its root. This means that we can use a single architecture to define all those aspects of morphology which have a paradigmatic character, from (transparent, productive) derivation to 'pure' inflection without having to make decisions about whether a process is 'really' derivational or 'really' inflectional.

This paper explores the notion that in some cases CASE features are interpretable. From this general premise I attempt to derive a set of facts such as free word order, multiple case realisation and certain extraction patterns in Korean. Finally I try to derive Case interpretability from the uniformity of the morphological paradigm.