see for further details

The "morphonosyntax" of short and long prepositions in Polish

Jesse Tseng (CNRS, Loria)

A handful of prepositions in Polish exhibit an alternation between a consonant-final short form and a long form ending in the support vowel _e_. This talk focuses primarily on the non-syllabic prepositions _w_ `in(to)' and _z_ `with, from' and the conditions under which they are realized as, respectively, _we_ and _ze_. Although this phenomenon is mentioned in all grammars of Polish, it is rarely described satisfactorily (two exceptions are Decaux 1978, p. 4.98-99, and Swan 2002, p. 33-34). In fact, the data turn out to be more complex than generally recognized.

The short/long alternation is determined by the immediately following word (though there is marginal evidence of triggering at a distance). The phenomenon is usually characterized in phonological terms: The long form _we_ appears before consonant clusters of the form [f/v] + C, and the long form _ze_ before clusters of the form ([v]) + [dental/palatal spirant] + C. This condition is sufficient (with minor exceptions) but not necessary, as we find the long forms in many other contexts. First, _ze_ appears to be generalizing for some speakers to simpler [v]-initial clusters (e.g., ze wzoru `from the model'). Second, it is well known that there are idiosyn-cratic cases where the long forms are required or preferred (we/*w mnie `in me', we/w krwi `in the blood', ze/*z mna `with me', ze/z wsi `from the country').

There is an additional, often overlooked morphological trigger of long forms. _We_ (resp. _ze_) is obligatory before words beginning with the related prefix _w_ (resp. _z_), even if there is no resulting consonant cluster (we/*w wjezdzie `in the entrance', ze/*z zewaluowanym projektem `with the evaluated project'). There is a clear contrast (i) with cases where _w_ and _z_ are simply root consonants (w/*we wierszu `in the poem', z/*ze zebra `with a zebra') and (ii) with cases where the prefixes themselves are realized as _we_ and _ze_ before a cluster-initial root (w/*we wewnetrznym pokoju `in the inner room', z/*ze zestresowanym szefem `with the stressed-out boss').

In phonological approaches, the _e_ of the long forms is the surface realization of an abstract vowel (Rubach 1984) or an epenthetic filler of an empty nucleus (Scheer, in press), conditioned by the properties of the following vowel/nucleus. Purely phonological accounts are unable to handle idiosyncratic lexical or collocational triggers, have difficulty dealing with contexts where the short and long forms are both possible, and offer no easy explanation of the prefix data.

We propose a lexicalist analysis of the observations presented, formulated in the framework of HPSG. Various alternatives are considered, including an extension of the phonology-based approach of Asudeh and Klein (2002), but a full treatment of the data requires the interaction of phonology, morphology, and lexical information at the word level, and syntactic mechanisms to achieve the correct structure at the phrasal level. The analysis features low attachment of the preposition, to the first word of the NP, with syntactic and semantic interpretation at the NP level. We incorporate the notion of "edge features", also required for phenomena like French consonant liaison and phrase marking in many languages. The Polish data lead us to address a crucial question: What information can and must be shared at the interface between phonology, morphology, and syntax?

Asudeh, A., and Klein, E. (2002) "Shape Conditions and Phonological Context." In F. Van Eynde et al. (eds) Proceedings of the 8th International Conference on Head-Driven Phrase Structure Grammar, p. 20-30. Stanford, CA: CSLI Publications.
Decaux, E. (1978) Lecons de grammaire polonaise. Paris: Institut d'Etudes Slaves.
Pullum, G., and Zwicky, A. (1988) The syntax-phonology interface. In F. Newmeyer (ed) Linguistics: The Cambridge Survey; Volume I, Linguistic Theory: Foundations, p. 255-280. Cambridge: Cambridge University Press.
Rubach, J. (1984) Cyclic and Lexical Phonology: The Structure of Polish. Dordrecht: Foris.
Rubach, J. (1976) "The Concept of an Underlying Representation." Biuletyn Polskiego Towarzystwa Językoznawczego 34:101-109.
Scheer, T. (in press) “How yers made Lightner, Rubach, Gussmann, Spencer & Co invent CVCV”. To appear in the proceedings of Generative Linguistics in Poland (GLiP) 3.
Swan, O. (2002) A Grammar of Contemporary Polish. Bloomington, IN: Slavica.