Inversion, Part 1

The following three features are used to ensure the correct pattern of inversion:
$\langle$wh$\rangle$: possible values are +/-
$\langle$inv$\rangle$: possible values are +/-
$\langle$invlink$\rangle$: possible values are +/- Facts to be captured:
1. No inversion with topicalization
2. No inversion with matrix extracted subject wh-questions
3. Inversion with matrix extracted object wh-questions
4. Inversion with all matrix wh-questions involving extraction from an embedded clause
5. No inversion in embedded questions
6. No matrix subject topicalizations. Consider a tree with object extraction, where NP is extracted. The following feature equations are used:

(519)0(519

(520)

S_q.b:$\langle$wh$\rangle =$ NP.t:$\langle$wh$\rangle$   (520)0(520

(521)

S_q.b:$\langle$invlink$\rangle =$ S_q.b:$\langle$inv$\rangle$   (521)0(521

(522)

S_q.b:$\langle$inv$\rangle =$ S_r.t:$\langle$inv$\rangle$   (522)0(522

(523)

S_r.b:$\langle$inv $\rangle = -$  

Root restriction: A restriction is imposed on the final root node of any XTAG derivation of a tensed sentence which equates the $\langle$wh$\rangle$ feature and the $\langle$invlink$\rangle$ feature of the final root node. If the extracted NP is not a wh-word i.e. its $\langle$wh$\rangle$ feature has the value -, at the end of the derivation, S_q.b:$\langle$wh$\rangle$ will also have the value -. Because of the root constraint S_q.b:$\langle$wh$\rangle$ will be equated to S_q.b:$\langle$invlink$\rangle$ which will also come to have the value -. Then, by (), S_r.t:$\langle$inv$\rangle$ will acquire the value -. This will unify with S_r.b:$\langle$inv$\rangle$ which has the value - (cf. ). Consequently, no auxiliary verb adjunction will be forced. Hence, there will never be inversion in topicalization. If the extracted NP is a wh-word i.e. its $\langle$wh$\rangle$ feature has the value +, at the end of the derivation, S_q.b:$\langle$wh$\rangle$ will also have the value +. Because of the root constraint S_q.b:$\langle$wh$\rangle$ will be equated to S_q.b:$\langle$invlink$\rangle$ which will also come to have the value +. Then, by (), S_r.t:$\langle$inv$\rangle$ will acquire the value +. This will not unify with S_r.b:$\langle$inv$\rangle$ which has the value + (cf. ). Consequently, the adjunction of an inverted auxiliary verb is required for the derivation to succeed. Inversion will still take place even if the extraction is from an embedded clause.

(523)0(523

(524)

Who_i does Loida think [Miguel likes t_i] 

This is because the adjoined tree's root node will also have its S_r.b:$\langle$inv$\rangle$ set to -. Note that inversion is only forced upon us because S_q is the final root node and the Root restriction applies. In embedded environments, the root restriction would not apply and the feature clash that forces adjunction would not take place. The $\langle$invlink$\rangle$ feature is not present in subject extractions. Consequently there is no inversion in subject questions. Subject topicalizations are blocked by setting the $\langle$wh$\rangle$ feature of the extracted NP to + i.e. only wh-phrases can go in this location.