Translation of sentences featuring clitics often poses a problem to machine translation systems. In this chapter, we illustrate, on the material from a Serbian ~ Catalan parallel corpus, a rule-based approach to solving translational structural mismatches between linguistic representations that underlie source- and target language sentences containing clitics. Unlike most studies in this field, which make use of phrase structure formalisms, ours has been conducted within the dependency framework of the Meaning-Text linguistic theory. We start by providing a brief description of Catalan and Serbian clitic systems, then introduce the basics of our framework to finally illustrate Serbian ~ Catalan translational mismatches involving the operations of clitic doubling, clitic climbing, and clitic possessor raising.
TopIntroduction
This chapter focuses on translational mismatches between Serbian and Catalan sentences featuring clitics.
A clitic is a deficient wordform in that it lacks inherent stress and has to lean prosodically on a stressed wordform (or phrase) in the clause, called the host of this clitic. Prototypical examples of clitics are clitic pronouns, but there are also clitic auxiliaries, conjunctions, particles, etc. In addition to having outstanding phonological behavior, clitics can be set apart form “normal” wordforms by their morphonology (special inflection and/or external sandhis) and syntax (clustering, rigid linear placement even in so called free word-order languages). The literature on clitics is huge; for a general and typological perspective on clitics, see in particular Zwicky (1977), Klavans (1995), Halpern (1995) and Spencer and Louis (2012); for work on Serbian and Catalan clitics, as well as for illustrative examples, see below.
By translational mismatches, or divergences, we mean non-trivial correspondences between linguistic representations that underlie a source language sentence and its equivalent in the target language. More specifically, we are interested in structural mismatches, which represent cases of violation of isomorphism between linguistic representations (Mel’čuk 2006, pp. 105-106). Two linguistic representations, say two syntactic dependency trees, are isomorphic if 1) all their nodes are in one-to-one correspondence based on the identity or semantic equivalence of their lexical labels, and 2) the dependency inside any two corresponding pairs of nodes has the same direction (i.e., there is no inversion of subordination, or head-switching). An influential typology of structural mismatches was proposed in Dorr (1993, 1994) and further developed in Mel’čuk and Wanner (2006); more on this will be said in due course.
For example, the (surface) syntactic structures of the Serbian sentence (1a) and its Catalan equivalent (1b) feature a structural mismatch involving clitic pronouns (boldfaced); for the structures themselves, see Figure 1:1
| (1) | a. | Ponudiše | ga | votkom | ‘They treated him with.vodka’. |
| | they.treated | he-ACC.SG | vodka-INSTR.SG | |
| b. | Li | van | oferir vodka | ‘They offered him vodka’. |
| | he-DAT.SG | they.go | offer vodka | |