Complementizer Selection

The following features are used to ensure the appropriate distribution of complementizers:
$\langle$comp$\rangle$, possible values: that, if, whether, for, rel, inf_nil, ind_nil, nil
$\langle$assign-comp$\rangle$, possible values: that, if, whether, for, ecm, rel, ind_nil, inf_nil, none
$\langle$mode$\rangle$, possible values: ind, inf, sbjnct, ger, base, ppart, nom, prep
$\langle$wh$\rangle$, possible values: +, - The value of the $\langle$comp$\rangle$ feature tells us what complementizer we are dealing with. The trees which introduce complementizers come specified from the lexicon with their $\langle$comp$\rangle$ feature and $\langle$assign-comp$\rangle$ feature. The $\langle$comp$\rangle$ of the Comp tree regulates what kind of tree goes above the Comp tree, while the $\langle$assign-comp$\rangle$ feature regulates what kind of tree goes below. e.g. the following equations are used for that:

(540)0(540

(541)

S_c.b:$\langle$comp$\rangle =$ Comp.t:$\langle$comp$\rangle$   (541)0(541

(542)

S_c.b:$\langle$wh$\rangle =$ Comp.t:$\langle$wh$\rangle$  (542)0(542

(543)

S_c.b:$\langle$mode$\rangle =$ ind/sbjnct  (543)0(543

(544)

S_r.t:$\langle$assign-comp$\rangle =$ Comp.t:$\langle$comp$\rangle$  (544)0(544

(545)

S_r.b:$\langle$comp$\rangle =$ nil 

By specifying S_r.b:$\langle$comp$\rangle =$ nil, we ensure that complementizers do not adjoin onto other complementizers. The root node of a complementizer tree always has its $\langle$comp$\rangle$ feature set to a value other than nil. Trees that take clausal complements specify with the $\langle$comp$\rangle$ feature on their foot node what kind of complementizer(s) they can take. The $\langle$assign-comp$\rangle$ feature of an S node is determined by the highest VP below the S node and the syntactic configuration the S node is in.