I was intrigued by Richard Sproat's challenge to the Minimalist 
community. As I understand it, the challenge calls for a renewed 
connection between computational and theoretical linguistics. The 
idea strikes me as intriguing -- it could create intellectual excitement 
and lead to new discoveries (not to mention more of those bright 
computer-science students in our syntax classes, always a good 
thing). 
 
Currently, the goals of the two fields are quite distinct: generative 
syntactic theory seeks to identify the universal structural principles 
constraining human language, which often manifest themselves most 
clearly in rather obscure and rarely-used aspects of language, like 
multiple wh-questions or passives, as well as in child language errors, 
selective impairments in aphasia, and so forth. By contrast, 
computational linguistics (at least the kind Richard Sproat is 
describing) seeks to develop an engineering tool that can reliably and 
efficiently parse a corpus into syntactic structures. Not surprisingly, 
practitioners in the two fields spend their time solving very different 
kinds of problems. Generally, I imagine, computational linguists have 
enough to do without also internalizing the current theoretical 
literature, and syntacticians have enough to do without also 
internalizing the current computational literature. 
 
Still, it seems as though there's a potential for convergence, if both 
theoretical and computational linguistics could be brought to bear on a 
computational model of the human language faculty, with all 
its 'inefficiencies' and limitations, as well as its capacities. A promising 
area for establishing an initial common ground would be sentence 
processing. For example, Ted Gibson's work on sentence processing 
is framed within a computational approach; and Colin Phillips' 1996 
MIT dissertation proposes a very intriguing union of syntactic theory 
with sentence processing. This syntax-processing-computation nexus 
seems like an ideal starting point for anyone seeking to implement a 
computational model of language processing within a current 
principles-and-parameters approach. 
 
I do hope that Richard Sproat's challenge will result in a surge of 
interest and activity in this area. However, it seems to me that if such 
an effort is to succeed, it will not be made by scholars in one field 
submitting to the goals and assumptions of another. It will require 
genuine collaboration, with syntacticians, psycholinguists, and 
computational linguists working together to develop collective goals 
and assumptions, and committing their knowledge, time, and 
resources to cross-disciplinary interaction and the training of graduate 
students. This kind of collaborative research is challenging and very 
time-consuming, but it can pay big intellectual dividends. I imagine 
there are many linguists who would be happy to act as resources on 
Minimalist theory, if someone is looking for collaborators. 
 
Cheers, 
Martha 
 
_____________________________________________ 
Martha McGinnis, Assistant Professor 
Linguistics Department, University of Calgary 
2500 University Drive NW, Calgary AB T2N 1N4 
phone: (403) 220-6119 fax: (403) 282-3880 
http://www.ling.ucalgary.ca/~mcginnis/ 
_____________________________________________

In issue 16.1156, Richard Sproat and Shalom Lappin challenge the 
Minimalist community to ''to produce, by May of 2008, a working P&P 
[Principals and Parameters Framework] parser that can be trained in a 
supervised fashion on a standard treebank, such as the Penn 
Treebank, and perform in a range comparable to state-of-the-art 
statistical parsers,'' between 80% and 90% accuracy under certain 
conditions. 
 
The goals of the P&P approach to language acquisition are 
dramatically different from those of statistical approaches, which 
makes a comparison in terms of accuracy alone uninformative. The 
P&P framework seeks to connect typological universals to the 
mechanism of language learning, in effect explaining those universals 
as properties of the initial state of the trainable parser. A statistical 
parser can, within physical limitations, recognize and learn any 
statistically significant pattern, not merely those patterns that occur in 
human languages. The P&P approach finds this disadvantageous, 
because the P&P framework seeks to answer the question ''What is a 
possible human language (type)?'' The P&P parser that Sproat and 
Lappin envision would answer this question; comparable statistical 
parsers do not. 
 
A successful P&P parser would not only acquire the target language 
accurately, it would behave like a language learner in its acquisition 
timeline and would fail to acquire languages that violate language 
universals. It would have to display these properties in order to 
successfully learn the target language, because these properties 
ought to be inherent in the parameters underlying the system. 
 
So the P&P parser that Sproat and Lappin envision would accomplish 
much more than comparable statistical parsers, which makes the 
proposed accuracy metric a poor yardstick for comparison, and 
furthermore, I suspect, it makes the three-year timeline unrealistic, 
especially since there is no reason to believe that the discovery of 
parameters and implicational relations among them is finished at the 
present time and ready to form the basis of a trainable parser. 
 
Nonetheless, I hope someone takes up the challenge (it's not my 
field), since the attempt can only benefit the P&P framework. Perhaps 
there should be a prize. 
 
Peter Hallman 
Department of Linguistics 
McGill University 
Montreal, Quebec, Canada