Topicalization and the value of the <inv> feature

Our analysis of topicalization uses the same trees as wh-extraction. For any NP complement position a single tree is used for both wh-questions and for topicalization from that position. Wh-questions have subject-auxiliary inversion and topicalizations do not. This difference between the constructions is captured by equating the values of the S_r's <inv> feature and the extracted NP's <wh> feature. This means that if the extracted item is a wh-expression, as in wh-questions, the value of <inv> will be + and an inverted auxiliary will be forced to adjoin. If the extracted item is a non-wh, <inv> will be - and no auxiliary adjunction will occur. An additional complication is that inversion only occurs in matrix clauses, so the values of <inv> and <wh> should only be equated in matrix clauses and not in embedded clauses. In the English XTAG grammar, appropriate equating of the <inv> and <wh> features is accomplished using the <invlink> feature and a restriction imposed on the root S of a derivation. In particular, in extraction trees that are used for both wh-questions and topicalizations, the value of the <inv> feature for the top of the S_r node is co-indexed to the value of the <inv> feature on the bottom of the S_q node. On the bottom of the S_q node the <inv> feature is co-indexed to the <invlink> feature. The <wh> feature of the extracted NP node is co-indexed to the value of the <wh> feature on the bottom of S_q. The linking between the value of the S_q <wh> and the <invlink> features is imposed by a condition on the final root node of a derivation (i.e. the top S node of a matrix clause) requires that <invlink>=<wh>. For example, the tree in Figure 13.1 is used to derive both ((138)) and ((139)).

(137)0(137

(138)

John, I like.  (138)0(138

(139)

Who do you like? 

For the question in ((139)), the extracted item who has the feature value <wh>=+, so the value of the <inv> feature on VP is also + and an auxiliary, in this case do, is forced to adjoin. For the topicalization ((138)) the values for John's <wh> feature and for S_q's <inv> feature are both - and no auxiliary adjoins.