Gosse Bouma (gosse@let.rug.nl)
Mon, 12 May 1997 09:58:24 +0200 (METDST)
Frederik Fouvry writes: > The problem: I ran into a spurious > ambiguity: in the case of certain verbs that allow two constructions > in subordinate clauses which collapse in the main clause (due to > linearisation effects): > > (1) ... dat hij een huis probeert te bouwen. > <... that he a house tries to build.> > (2) ... dat hij probeert een huis te bouwen. > <... that he tries a house to build.> > (3) Hij probeert een huis te bouwen. > <He tries a house to build.> > > Those verbs (here "proberen" - to try) are in the ANS (the standard > Dutch grammar) classified as NVG (non-obligatory group forming), which > means that they are not obligatorily followed by the infinitive. In > (1) it is followed by the infinitive, but not in (2). Both word > orders collapse in a V2 clause: first the verb `goes' in front, and > then the subject or the complements. Or in a diagram: > > (4) Hij[NP] probeert[V] een_huis[NP] te[MRK] bouwen[V] > --------------VP-------------- > ("proberen" has 1 argument, like in (2) - lcomps of "bouwen" satisfied) > > (5) Hij[NP] probeert[V] een_huis[NP] te[MRK] bouwen[V] > -----NP----- -------V--------- > (argument inheritance: "proberen" has 2 arguments, like in (1)) > > This effect seems unavoidable to me, unless I start tinkering with > the rule sequences, which would make the grammar messy, to say the > least. > Anything I missed? Well, one might argue that the ambiguity is actually worse, as `proberen' also behaves as an `obligatory group forming' (verb raising) verb in some cases. This distinction is only relevant if `proberen' is governed by an auxiliary (i.e. non-obl group forming subsumes group forming as far as word order is concerned. Auxiliaries, however, govern an infinitive, in case this infinitive is an obl group former, and a participle in all other cases. Since we can have either `een huis heeft geprobeerd te bouwen' and `een huis heeft proberen te bouwen', proberen must be both non-obl and obl group forming). Therefore, there is a third analysis here, with proberen as an obl group forming verb. > Any comments? This spurious ambiguity problem is not restricted to the non-obligatory group forming (third construction, partial extraposition) verbs, but arises with all verbs selecting a complement that may either precede or follow the head in subordinate clauses. Examples are verbs selecting a PP complement (i), and argument inheritance verbs which allow their (lexical) verbal complement to go either left or right of the head, such as modals (ii) and auxiliaries (iii): (i) a. dat hij aan deze oplossing twijfelt that he on this solution doubts `that he has doubts about this solution' b. dat hij twijfelt aan deze oplossing c. hij twijfelt aan deze oplossing (ii) a. dat hij het boek lezen wil that he the boek read wants `that he wants to read the book' b. dat hij het boek wil lezen c. Hij wil het boek lezen (iii) a. dat hij het boek gelezen heeft that he the book read has `that he has read the book' b. dat hij het boek heeft gelezen c. Hij heeft het boek gelezen > Any solutions? I don't think there is an easy solution. Any approach which accounts for the difference between main and subordinate word order by means of syntactic rules (as oppposed to using different lexical entries) will run into this problem. For instance, accounting for v1 and v2 clauses by stipulating a verbal trace in the clause final position (as proposed by Netter and others), does not help, as the trace could be in two different positions. Alternatively, using an approach to word order inspired by the topological fields theory (as in Kathol's dissertation) doesn't help either, as the `right sentence bracket' is empty, and thus it is unclear where the `mittelfeld' ends and the `nachfeld' starts. If one really wants to get rid of this ambiguity, it seems one must assume that the lexical entry for verbal heads in main and subordinate clauses are actually different. This option is common in CG. In some of Steedman's papers on Dutch you find a `carrousel rule' which will rotate the arguments of a finite verb, in order to account for v1 and v2 word order. (In other papers by Steedman, and in Hepple's diss, there is a more spectacular account, involving lots of (disharmonic) type raising of the arguments. I don't think that is a feasible solution in HPSG.) Such an account would become more plausible if you can find arguments that the two must be distinguished for other reasons as well. One situation that comes to mind is the inflection of finite, second person singular, verbs: (iv) a. dat je een boek leest that you a book read b. Je leest een boek c. Lees je een boek I.e. in main clauses, the finite verb looses its -t ending, if it is followed by a second person singular subject (je/jij). This suggests that the rules for inflection must be able to impose constraints on the clause type that the verb will appear in (or else, if you assume a Borsley style of account of `inversion', must ensure that the subject goes to COMPS instead of SUBJ). There might be other arguments for making a distinction. (Steedman has argued for a radically different constituent structure in main and subordinate clauses on the basis of serial coordination (noch .. noch, neither .. nor. That might be relevant.) Finally, you might try to argue that the ambiguity is not really spurious, i.e. that even in main clauses you need to assume that in some cases a complement is in the `mittelfeld', whereas in others it is in the `nachfeld'. For instance, adverbs may either precede or follow a complement, but must precede the verb in subordinate clauses: (v) a. dat hij nooit aan deze oplossing twijfelde. that he never on this solution doubted b. dat hij aan deze oplossing nooit twijfelde. (* dat hij aan deze oplossing twijfelde nooit.) c. Hij twijfelde nooit aan deze oplossing. d. Hij twijfelde aan deze oplossing nooit. Thus, (c) and (d) suggest that the PP must at least be able to appear in the `mittelfeld'. Now, if we also had elements which must follow the head in subordinate clauses, and we encounter the same word order variation, the PP must also be able to appear in the `nachfeld'. Here is a try: (vi) a. dat ik een boek te leen gaf aan marie over grammatica that I a book for borrowing gave to mary on grammar `that I borrowed mary a book on grammar b. dat ik een boek te leen gaf over grammatica aan marie (* dat ik een boek gaf te leen over grammatica aan marie) c. Ik gaf een boek te leen aan marie over grammatica d. Ik gaf een boek te leen over grammatica aan marie Here, the PP modifier of the NP `a book' can be extraposed into the nachfeld. The pp complement of give can also go there. Both orders are allowed. In main clauses, we find that `over grammatica' may follow the extraposed pp (note that the idiomatic `te leen' ensures that the PP-modifier is actually extraposed). Thus, `over grammatica' must be in the nachfeld as well. (Other examples might perhaps be constructed using verbs which select a seperable prefix.) (v) and (vi) together suggest that PP complements must sometimes appear in the mittelfeld and sometimes in the nachfeld, and that this distinction is relevant even in main clauses. Gosse. -- Gosse Bouma, Alfa-informatica, RUG, Postbus 716, 9700 AS Groningen gosse@let.rug.nl tel. +31-50-3635937 fax +31-50-3636855
This archive was generated by hypermail 2.0b3 on Fri Dec 18 1998 - 20:36:06 PST