Advances in Role and Reference Grammar

A SYNOPSIS OF ROLE AND REFERENCE GRAMMAR 147

This is not possible when there is no single lexical verb to paraphrase the V

• NP, as in What did Mary dispute the claim that Fred bought?. In the case
  of make the claim, lexical semantic factors override (111) and allow the sub­
  ordinate clause to be in the PFD, meeting (130). This phenomenon man­
  ifests itself in a number of constructions, e.g. Who did you read a book
  about? = Who did you read about? but not Who did you deface a book
  about?. Thus lexical semantic factors play a crucial role in determining the
  possibility of question formation in various constructions, and a full expla­
  nation of restrictions on question formation must address the effects lexical
  semantics can have on the information structure of sentences.
  As is well known, languages vary with respect to the restrictions placed
  on question formation, and the approach sketched here makes it possible to
  identify the parameters along which languages will vary. There are two
  major ones: (1) how "deep" into the sentence does (111) apply?, and (2)
  how much can lexical semantic factors override (111)? (Northern) German
  and Russian, for example, permit question formation only out of main
  clauses, and therefore (111) does not apply to them at all. It is an empirical
  question as to how deep (111) may be valid, and it may turn out that it is an
  all or nothing opposition: either a language allows question formation out
  of at least some subordinate clauses, as in English and Lakhota, or it does
  not allow it out of any subordinate clauses, e.g. Russian and German.^65
  The second parameter is dependent on the first; if a language restricts
  question formation to main clauses, then there is nothing for lexical
  semantic factors to override. This parameter is meaningful only in lan­
  guages in which question formation is possible out of at least some subordi­
  nate clauses. English and Danish provide a contrast here. Erteschik-Shir
  (1973) and Erteschik-Shir & Lappin (1979) show that in Danish it is possi­
  ble to have question formation out of some relative clauses, if the main
  clause is relatively semantically empty and the head noun is non-specific or
  generic; for example, the Danish equivalent of What are there many who
  like? is grammatical. Most English speakers seem to reject questions like
  this one, and therefore it appears that lexical semantic factors may play a
  stronger role in Danish in licensing question formation in violation of (111)
  than in English. The influence of lexical semantic factors in these construc­
  tions in different languages can only be ascertained through empirical inves­
  tigation.