21.04.2013 Views

Eckhard Bick - VISL

Eckhard Bick - VISL

Eckhard Bick - VISL

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Therefore, the @ADV-@ADVL distinction 178 can be made almost entirely at the "early"<br />

syntactic level - i.e., the mappings section of the CG-rules file, where most @ADV<br />

cases can be identified, depending heavily on lexical context (especially, lexical valency<br />

information), and less on syntactic reasoning.<br />

Material-wise, CG will map morphological adverbs only as @ADVL, @ADV on<br />

the clause level, and mostly @>A or @A< on the group level. Prepositional phrases<br />

(PPs), however, will in addition be mapped as @PIV (prepositional object) @SC<br />

(subject complement) or @PRED (free predicative) on clause level, and @N< (postnominal)<br />

on group level. In addition, the internal structure of PPs can be much more<br />

complex than that of most adverb-phrases, which makes relevant clause level context<br />

more distant and less visible to the CG disambiguation rules. I will therefore focus on<br />

the syntactic function of PPs throughout the rest of this chapter.<br />

Though valency bound, and therefore marked lexically on the verb, @PIV is<br />

much more difficult to disambiguate than a PP @ADV, since it covers a wide range of<br />

prepositions and semantic roles. PP @N< is somewhat easier to handle - at least where<br />

there is relevant valency information on the head -, since group level attachment allows<br />

my CG to rule out interfering clause level complements or adjuncts. At the same time,<br />

in the case of PP @N< hierarchies, leftward attachment can be left underspecified. In<br />

analogy to the clause-level distinction between @PIV/@ADV and @ADVL, also PP<br />

@N< can be differentiated into valency-bound "objects" and circumstantial "adjuncts":<br />

discussão sobre a Dinda (discussion about Parliament) vs. discussão na Dinda<br />

(discussion in Parliament). Again, the first invites lexical solutions, while the second is<br />

a real competitor to the @ADVL reading.<br />

Just how big the syntactic ambiguity potential of PPs is in my CG-description,<br />

can be seen in (1). A simple adverbial tag (@ADVL) would be the obvious default<br />

reading for most of the categories below, - if the tag set was to be reduced for<br />

pedagogical reasons, to ensure a very low error rate, or for transformation into or<br />

comparison with other - less detailed - tagging systems. In my discussion of the other<br />

function tag alternatives for PPs, I will therefore focus on those traits that distinguish<br />

them from the @ADVL "prototype", as well as on the disambiguation tools employed.<br />

(1a) @ADVL VEJA, 30. Dez. 92<br />

(1b) @ Em 1992, o desemprego foi recorde.<br />

Ao retornar, em 1986, encontrou o país transformado ...<br />

(1d) @ No Rio, onde morava antigamente.<br />

178 After this, one of the two classe, @ADVL, will still be ambiguous, - with regard to other adverbial classes like @>A<br />

(adverbial pre-modifier).<br />

- 308 -

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!