Discussion Details
| Title: | Re: A Challenge to the Minimalist Community |
| Submitter: | Ash Asudeh |
| Description: | Some confusion has arisen in the subsequent discussion of the
Sproat-Lappin challenge. Most of the subsequent posts discuss statistical parsing versus P&P parsing. However, the challenge has nothing to do with statistical parsers per se, although it's true that much of the original challenge discussed these parsers. There are, however, non-statistical or hybrid broad-coverage parsers in theories other than P&P, such as the PARC LFG parser mentioned in the original Sproat-Lappin challenge and the HPSG Lingo/LKB parser at Stanford. For theorists who for whatever reason think that statistical parsers are a different sort of thing from the P&P parser that the challenge focuses on (they are not, for the reasons sketched by Sproat and Lappin themselves and also by Ken Shan (LL16.1288)), these can perhaps serve as a reference point instead. There is also, in my opinion, a confusing aspect to the Sproat-Lappin challenge. They are not only asking for a P&P parser, but also asking for a large P&P grammar. Grammars and parsers are different things, at least for non-statistical "deep grammar" approaches (purely statistical methods instead induce the grammar), such as the HPSG and LFG ones noted above and, presumably, like the P&P one in question. For example, the PARC LFG parser has been applied to analyses of a large number of languages, although the English grammar is what Sproat and Lappin seem to have in mind in their message, given the citation of Riezler et al. Another example is the LKB system, which implements an HPSG parser. The English Resource Grammar (ERG) is used with this parser, but they are developed separately (although with strong interactions -- same goes for the PARC XLE parser and the ParGram grammars). Carson Schütze (16.1288) wrote: In addition to capturing the distinction between learnable and unlearnable languages, P&P has as an important goal capturing the distinction between well-formed (grammatical) and ill-formed (ungrammatical) sentences within a language. As I understand it, the challenge demands only correct parsing of grammatical sentences, not correct rejection of ungrammatical ones. This represents another case where the P&P system, by virtue of the goals of the theory, is being subjected to greater demands than the statistical parsers. I don't understand the substance of this objection. All grammars, those used in statistical parsing or otherwise, attempt to reject ungrammatical sentences: Nobody wants their grammar/parser to overgenerate. Even if the claim is true of statistical parsers (I don't think it is), it certainly isn't true of the LFG and HPSG parsers and grammars noted above. Peter Hallman (LL16.1251) and Martha McGinnis (also LL16.1251), although positively inclined to the challenge, also raise objections. The substance of the objections are that P&P is attempting to do much more than just parse sentences (Hallman) and that the goals of P&P are different to those of computational linguistics (McGinnis). I think there is merit to both these statements, but they are ultimately non sequiturs to the challenge. P&P is not just concerned with how the adult grammar results from the inititial state (although this is a principal goal): it is also concerned with the state of the final grammar (Chomsky, 1986: "What constitutes knowledge of language?"). The requirement of capturing the adult grammar also means that it's insubstantial whether the goals of P&P are those of computational linguistics: P&P is still expected to capture adult grammatical competence in the end, even if this isn't a *motivation* for a lot of its practitioners. Lastly, I agree with Ken Shan (LL16.1288): "All other things being equal, poor (or unknown) parsing performance indicates failure at (resp. disinterest in) answering Q [What is a possible human language?]." It is no surprise that the objections to the challenge so far have sought to argue that all other things are not equal. Like Ken, however, I think that this challenge is a *good* thing for P&P and I too am optimistic for P&P. In any case, the attempt to meet the challenge will without doubt reap huge rewards, not just for computational linguistics, but especially for P&P theory. Ash Asudeh |
| Date Posted: | 29-Apr-2005 |
| Linguistic Field(s): |
Computational Linguistics
Syntax Discipline of Linguistics |
| LL Issue: | 16.1364 |
| Posted: | 29-Apr-2005 |
