I am currently working on my dissertation on the Machine Translation of
prepositional phrases (not all prepositional phrases, actually only a
subset, and focusing on French and English for the time being, but
that is not important for the current message).
In preparing my lit. review, I have uncovered a large amount of material
concerning prepositions, prepositions and case, verbal argument structure,
etc..., but I have had some problems locating previous work in the
following areas. Any pointers or other assistance would be most welcome.
If there is interest, I will, of course, post a summary of responses to
the list.
I am currently in search of:
1. articles directly addressing the problem of _Machine Translation_ of
prepositional phrases - in particular, previous proposed solutions and/or
implementations/evaluations of those solutions. I have come across
several inspiring papers, which are listed at the end of this e-mail, but
I feel that there certainly must be more out there.
2. articles relating to evaluation of the current state of the art in PP
translation. This is the most troublesome, because, as most of you
probably know, there are many varying viewpoints on the subject of
evaluation of MT systems, ranging from the more academically-focused
to the more commercially-focused.
I have found many articles on MT evaluation methodologies (including an
entire, very interesting, issue of CL), but haven't run across anything
that specifically addresses current MT systems' performance with respect
to PPs. It's not necessarily enough to say that this is recognized as a
'hard problem' (i.e. it's listed as such in conference calls, and
anecdotally, people seem to agree that the issue needs to be addressed.)
I need to be able to make the case that, yes, it is agreed that PP
translation is a difficult and important problem, AND CURRENT SYSTEMS
AREN'T VERY GOOD AT IT.
I don't think that any large-scale studies have been done to address this
particular issue, and the development (let alone the implementation) of an
MT evaluation methodology to fill this gap could be another whole
dissertation in itself.
Both a) pointers to articles and
b) suggestions for handling the fact that there may not be numeric
(hard, factual) metrics that demonstrate that PP translation is an area in
MT that needs improvement would be appreciated.
As a sidenote, I will be out of the office next week (July 25-Aug 1),
so I won't be able to summarize responses until after that.
Thank you in advance for any assistance you are able to provide.
----- Keith J. Miller
Geogetown University
Linguistics Department
Computational Linguistics
millerk@gusun.georgetown.edu
References relating directly to the MT of PPs:
Brée, D. S., R. A. Smit, and J. P. Van Werkhoven. "Translating Temporal
Prepositions Between Dutch and English." Journal of Semantics 7 (1990): 1-
51.
Japkowitz, Nathalie, and Janyce M. Wiebe. "A System for Translating Locative
Prepositions from English into French." 29th Annual Meeting of the
Association for Computational Linguistics 29 (18-21 June 1991): 153-60.
Sumita, Eiichiro, and Hitoshi Iida. "Example-Based NLP Techniques: A Case
Study of Machine Translation." In Statistically-Based Natural Language
Programming Techniques: Papers from the 1992 AAAI Workshop; March 25 - 27,
Stanford
University, 81-87. Technical Report SS-92-01. Menlo Park, California: AAAI
Press, 1992.
Trujillo, A. "Locations in the Machine Translation of Prepositional
Phrases."
Quatrième Colloque international sur les aspets théoretiques et
méthodologiques de la traduction automatique Fourth International
Conference on Theoretical and Methodological Issues in Machine Translation
Méthodes
empiricistes versus méthodes rationalistes en TA Empiricist vs.
Rationalist Methods in MT 4 (1992): 13-20. Montréal.