Tree-local MCTAG with Shared Nodes: An Analysis ofWord Order Variation in German and Korean

TALN 2004, Fs, 19–21 avril 2004 Tree-local MCTAG with Shared Nodes: An Analysis of Word Order Variation in German and Korean Laura Kallmeyer, SinWon Yoon UFRL, University Paris 7 2 place Jussieu, Case 7003, 75251 Paris Cedex 05 {laura.kallmeyer,swyoon}@linguist.jussieu.fr Mots-clefs – Keywords Grammaires d’Arbres Adjoints, brouillage d’arguments, ordre des mots, allemand, cor´een Tree Adjoining Grammars, scrambling, word order, German, Korean R´esum´e - Abstract Les Grammaires d’Arbres Adjoints (TAG) sont connues pour ne pas ˆetre assez puissantes pour traiter le brouillage d’arguments dans des langues `a ordre des mots libre. Les variantes TAG pro- pos´ees jusqu’`a maintenant pour expliquer le brouillage ne sont pas enti`erement satisfaisantes. Nous pr´esentons ici une extension alternative de TAG, bas´ee sur la notion du partage de noeuds. En considerant des donn´ees de l’allemand et du cor´een, on montre que cette extension de TAG peut en juste proportion analyser des donn´ees de brouillage d’arguments, ´egalement en combi- naison avec l’extraposition et la topicalisation. Tree Adjoining Grammars (TAG) are known not to be powerful enough to deal with scrambling in free word order languages. The TAG-variants proposed so far in order to account for scram- bling are not entirely satisfying. Therefore, an alternative extension of TAG is introduced based on the notion of node sharing. Considering data from German and Korean, it is shown that this TAG-extension can adequately analyse scrambling data, also in combination with extraposition and topicalization. Kallmeyer, Yoon 1 LTAG and scrambling Lexicalized Tree Adjoining Grammars (LTAG, (Joshi & Schabes, 1997)) is a tree-rewriting formalism. An LTAG consists of a finite set of trees (elementary trees) associated with lexical items. Larger trees are derived by substitution (replacing a leaf with a new tree) and adjunction (replacing an internal node with a new tree). In case of an adjunction, the new elementary tree has a special leaf node, the foot node (marked with an asterisk). When adjoining such a tree (a so-called auxiliary tree) to a node µ, in the resulting tree, the subtree with root node µ from the old tree is put below the foot node of the new auxiliary tree. Non-auxiliary elementary trees are called initial trees. LTAG elementary trees represent extended projections of lexical items and encapsulate all syntactic arguments of the lexical anchor. They are minimal in the sense that only the arguments of the anchor are encapsulated, all recursion is factored away. Roughly, scrambling is the permutation of elements (arguments and adjuncts) of a sentence (we use the term scrambling in a purely descriptive sense without implying any theory of move- ment). A special case is long-distance scrambling where arguments or adjuncts of an embedded infinitive are ‘moved’ out of the embedded VP. This occurs for instance in languages such as German, Hindi, Japanese and Korean. These languages are therefore often said to have a free word order. Consider for example the German sentence (1). In (1), the accusative NP es is an argument of the embedded infinitive zu reparieren but it precedes der Mechaniker, the subject of the main …

Content truncated to 500 words