This is the abstract of a TULCON14 presentation.
Kyle Maycock (Ohio State University)
The Albanian noun phrase marks four distinct morphosyntactic properties: case, definiteness, number, and gender. While number and gender are found on every word in the phrase, only the first lexical word—either the noun (1) or an adjective (2)—marks case and definiteness.
(1) vajz-at e mir-a
girl-NOM.DEF.PL.F NOM.DEF.PL.F good-PL.F
‘the good girls’
(2) e mir-at vajz-a
NOM.DEF.PL.F good-NOM.DEF.PL.F girl-PL.F
‘the good girls’
The placement of number and gender is straightforwardly morphological, but the placement of case and definiteness is dependent on the syntax, giving these morphosyntactic properties clitic-like properties. Clitics lie at the boundary of morphology and syntax and are thus informative for studying the morphology-syntax interface. In this presentation, I develop an analysis of the Albanian data that defines the roles of syntax and morphology into roles of placement and insertion of form, respectively.
Canonical clitics (Spencer and Luís 2012) lie somewhere between words and affixes; they contain grammatical content and require a host, like inflectional affixes, but are placed relative to the phrase, like words. The Albanian clitics are unlike canonical clitics due to their cumulative expression; they express the affixal properties of number and gender and the clitic-like properties of case and definiteness in one suffix. Their placement is also “special” (Zwicky 1977) because they are placed in a particular location: second position within the noun phrase. An analysis must deal with these two major issues.
Cumulative expression presents a unique problem for models of second-position clitic placement, as most previous models, such as Halpern (1995) and Anderson (2005), assume that phonological form must be inserted prior to final placement. Applied to the Albanian data, this results in two suffixes (one with case and definiteness and one with number and gender) rather than the requisite single suffix. Instead, a model that does not make this assumption is preferred. I develop criteria for a model to account for the Albanian data as edge inflection and offer a possible analysis using Head-Driven Phrase Structure Grammar (HSPG, Pollard and Sag 1999). In dealing with cumulative expression, a particularly useful tool for a model of morphology is specificity-based ordering; when multiple suffixes are applicable, specificity-based ordering selects the most specific of the applicable suffixes and applies it to the stem. The suffix in (2) on vajz-a and the suffix in (1) on vajz-at are both applicable to the stem in (1) because they both contain PL.F. An analysis using specificity-based ordering, from a model such as Paradigm Function Morphology (PFM, Stump 2001), results in the selection and application of -at, the most specific applicable suffix. In this presentation, I develop an analysis that solves the issues of placement and cumulative expression using HPSG and PFM and demonstrate how this delineates the roles of syntax and morphology into placement and insertion of form.