[developers] Fwd: HPSG implementations

[Emily M. Bender]

Hi Stefan,

A few more ideas:

--- Are you also pointing to implemented work in LFG (and *TAG and
other frameworks)?

--- I have a couple of papers in which I argue that grammar engineering
is essential for hypothesis testing (and in fact, that grammars are complex
enough that without implementation, we can't really tell if they work
the way we as linguists think they do).  One is in TLSX (bibtex below).
The other is to appear very shortly in LiLT (preprint attached separately).
These papers amplify Oepen & Flickinger's point that "the interaction
of lexicon
and phrase structure apparatus can be subtle and make it hard to predict how
even modest changes to the grammar affect system behaviour."

Emily

@inproceedings{Bender:08,
  author = {Bender, Emily M.},
  title = {Grammar Engineering for Linguistic Hypothesis Testing},
  booktitle = {Proceedings of the {T}exas {L}inguistics {S}ociety {X}
Conference: Computational Linguistics for Less-Studied Languages},
  editor = {Nicholas Gaylord and Alexis Palmer and Elias Ponvert},
  address = {Stanford CA},
  publisher = {CSLI Publications ONLINE},
  year = {2008},
  pages = {16--36},
}

@article{Oep:Fli:98,
  author = {Stephan Oepen and Daniel P. Flickinger},
  title = {Towards Systematic Grammar Profiling.
           {T}est Suite Technology Ten Years After},
  editor = {Robert Gaizauskas},
  journal = {Journal of Computer Speech and Language},
  volume = {12 (4) (Special Issue on Evaluation)},
  year = 1998,
  pages = {411$\,$--$\,$436}
}


On Mon, Feb 8, 2010 at 3:05 AM, Stefan Müller
<Stefan.Mueller at fu-berlin.de> wrote:
> Hi Emily (and DELPH-IN developers by cc),
>
> Thanks for all the pointers! I am really impressed how much is done in
> HPSG in terms of implementations.
>
> The point that I am trying to make in the book is that 50 years of
> Chomskian linguistics (in the narrow sense) did not result in a single
> implementation that reflect the ideas one by one. If you know of such a
> thing I would be happy to learn more about it. Currently I mention the
> following GB-inspired systems:
>
> @inproceedings{AC86a,
> author = {Steven Abney and Jennifer Cole},
> title  = {A Government-Binding Parser},
> booktitle = {Proceedings of North Eastern Linguistic Society 16. GLSA},
> address = {University of Massachusetts, Amherst},
> pages = {1--17},
> year = 1986,
> }
>
> @Book{Marcus80a,
>  author      = {Mitchell P. Marcus},
>  title       = {A Theory of Syntactic Recognition for Natural Language},
>  address     = {London, England -- Cambridge, Massachusetts},
>  publisher   = {The MIT Press},
>  year        = 1980
> }
>
> @PhDThesis{Fong91a-,
>  author      = {Sandiway Fong},
>  title       = {Computational Properties of Principle-Based Grammatical
> Theories},
>  school      = {MIT Artificial Intelligence Lab},
>  url = {http://www.neci.nec.com/homepages/sandiway/pappi/index.html},
> url_checked =   {\urlchecked{10}{10}{2002}},
>  year        = 1991
> }
>
> @inproceedings{Correra87a,
> crossref = {acl87},
> author = {Nelson Correa},
> title = {An Attribute-Grammar Implementation of {Government-Binding
> Theory}},
> url = {http://acl.ldc.upenn.edu/P/P87/P87-1007.pdf},
> url_checked = {31.03.2008},
> pages = {45--51},
> }
>
> @article{Nordgard94a,
> author = {Torbj{\o}rn Nordg\r{a}rd},
> title  = {{E-parser}: An Implementation of a Deterministic {GB}-Related
> Parsing System},
> journal = {Computers and the Humanities},
> volume  = 28,
> Number = {4--5},
> pages = {259--272},
> year = 1994
> }
>
> However, these systems are only loosely based on Chomsky's ideas. That
> is, they do not use transformations.
>
> My point is that Chomsky explicitly refused further formalization of
> linguistic theory in publications like this:
>
> @Article{Chomsky90a,
>  Author         = {Chomsky, Noam},
>  Title          = {On formalization and formal linguistics},
>  Journal        = {Natural Language and Linguistic Theory},
>  Volume         = {8},
>  Pages          = {143--147},
>  year           = 1990
> }
>
> I think that this is one reason for the lack of implementations in this
> area of linguistics.
>
> So what I am looking for are examples that show that there is a fruitful
> interaction between theory building and theory verification by
> implementation (for whatever purposes the implementation was done,
> sometimes the implementation may be even useful in practical systems =;-).
>
> Since the whole thing may cause some discussion, it is important to
> point to fragments of a relevant size and to fragments that are
> well-documented. The best thing is of course journal papers or high
> profile conference papers.
>
> In the discussion in the book, I mention the Matrix, but this was not
> posted to the HPSG-L since the question was about individual languages.
>
> You will find the updated list here at the first page of Chapter 8:
>
> http://hpsg.fu-berlin.de/~stefan/Pub/grammatiktheorie.html
>
> If you read German and happen to have time, I would be happy to get
> general comments.
>
> I am still working on it, especially the Nativism, Language Acquisition
> and Psycholinguistics stuff takes a lot of time ...
>
> I hope to finish everything in March.
>
> Thanks again and best wishes
>
>        Stefan
>
> --
> Stefan Müller       Tel: (+49) (+30) 838 52973
>                    Fax: (+49) (030) 838 4 52973
> Institut für Deutsche und Niederländische Philologie
> Deutsche Grammatik
> Habelschwerdter Allee 45
> 14 195 Berlin
>
> http://hpsg.fu-berlin.de/~stefan/
>
> http://hpsg.fu-berlin.de/~stefan/Babel/Interaktiv/
>
>