Sum: Word-and-Paradigm morphology in HPSG

[Raúl Aranovich]

Thanks to all who answered my query about W&P morphology and HPSG. I
received a lot of useful comments and suggestions. I have summarized them
for the list, and I have added some more thoughts and questions.

A group including Farrell Ackerman, Greg Stump, Jim Blevins, Andrew
Spencer, Louisa Sadler, Rachel Nordlinger, Ana Luis, among some others,
have been exploring incorporating word and paradigm models into
unification-based lexicalist theories. Their formalism is essentially that
of LFG, but their results could be also applied to HPSG. I have included
some references to their work below. Jim Blevins is working on a monograph
on that topic. One of the central thrusts of their work is to show that
analytic and synthetic forms can all be part of the same paradigm. CSLI
will publish a book edited by Spencer and Sadler on this topic.

There are several works in HPSG morphology that adopt a model more similar
to the Item-and-Arrangement approach. In this approach morphemes are
treated as signs, and they are related to form words more or less with the
same mechanisms that relate signs (daughters) to form phrases (i.e.
word-syntax). Frank Van Eynde wrote a Habilitationsschrift about
auxiliaries and verbal affixes in 1994, in this framework. John Nerbonne
and Uli Krieger also defended this approach in their work.

There has been a growing interest for some time now in HPSG (some would
think from the very beginning) on implementations of Word-and-Paradigm
morphology, or, more generally, realizational theories of morphology. Some
of this work includes J. Calder (1989, 1990), A. Kathol (1999), J-P. Koenig
(1999), D. Meurer (2001), S. Mueller (to appear, 2002) S. Riehemann (1998).
John Nerbonne writes that "most of the published work on lexical relations
in HPSG does NOT assume the existence of inflectional morphemes as signs
beginning with the sketchy remarks in Pollard and Sag, 1987, 1994, which
suggest a separate function responsible for form, f_{3sg}.  An excellent
overview of the different positions may be found in Bouma, van Eynde and
Flickinger's paper".

I have found Tomaz Erjavec's PhD thesis "Unification, Inheritance and
Paradigms in the Morphology of Natural Languages" very useful and
informative. This work explores theoretical and computational aspects of
paradigm-based morphology for the analysis of Slovene verbal inflection.
Part of the work consists of an extension to ALE that supports the
implementation of such an approach. Erjavec's work is similar in spirit to
the approaches in Calder (1989, 1990), Riehemann (1998), and Bird and Klein
(1994). I believe it goes one step further into a radical implementation of
realizational morphology in HPSG by separating morpholexical types from
morphophonological types. This work may answer Kathol's remark that
"relying exclusively on realizational functions for inflectional morphology
[has the downside that it] pretty much completely separates (inflectional)
morphology from the constraint-based mechanisms used in HPSG for describing
syntactic (and semantic) well-formedness". I am told that Berthold Crysmann
gave a talk on this very topic at the Lille conference on Sept 19 to 21.
The title of his communication was: "Paradigm functions in HPSG". His
abstract (and his email adress) can be found on the web site of the
conference at: <gdr-morphologie.linguist.jussieu.fr>. His dissertation
proposes to implement Stump's division between rules of exponence and their
combination into paradigms by means of type hierarchies of forms.

I started from the assumption that the HPSG formalism (i.e. a formalism
based on typed feature structures) is more naturally fitted to the
Item-and-Arrangement model of morphology. That assumption is wrong. As Jim
Blevins put it to me, "constraint-based approaches have tended to adopt an
IP or IA model of morphology, but that mostly reflects familiarity rather
than anything to do with intrinsic compatibility, as far as I can tell."
This seems to be another case in which the formalism should not be confused
with the theory. Typed feature structures with unification let us formalize
I&A, I&P, or W&P models of morphology. The advantage of using such
formalism is that the claims or predictions of one's analysis are made
explicit in a rigurous way. But deciding among the I&A, I&P, or W&P models
is another matter, and that is when one is doing 'linguistic theory'. At
this point is when I can see someone shooting the criticism "HPSG lets you
do anything, so it is not a theory of language". Well, agreed: but
expecting that the formalism will do the work of the theory is not right
(paraphrasing Thomas Frank's socio-political analysis of globalization I
would call this 'formal populism').

For instance, one ofthe issues that is in my mind now is whether the W&P
model should be extended to cover derivational morphology as well. Anderson
(1982) leaned against ding that, but Beard (1995) has some arguments for
going "all the way", formulating a realizational theory of derivational
morphology. This question is independent of whether one uses typed feature
structures to model morphological entities. However, the formalism puts one
particularly irritant issue in the spotlight: Inflectional morpho-lexical
rules are monotonic (i.e. they add information), whereas derivational
morpho-lexical rules are not. Beard (1995) glosses over this issue, and its
consequences for his arguments in favor of radical realizational
morphology. The formalism of HPSG will not give an answer to this problem
on its own, but it does not let us sweep it under the rug. I would
appreciate any comments on these thoughts.

REFERENCES

Ackerman, Farrell, and Gregory Stump 'Paradigmsand Periphrastic Expression:
A Study in Realization-based Lexicalism'. [Availble on-line at:
http://ling.ucsd.edu/~ackerman/]

Ackerman, Farrell, and Gert Webelhuth (1998) A Theory of Predicates,
Stanford: CSLI Publications.

Bird, S. and Klein, E (1994). "Phonological Analysis in Typed Feature
Systems". Computational Linguistics, 20(3), 455-492.

Blevins, James. 'Stems and paradigms' Morphomic stems in West Germanic

  ----- 'Periphrasis as syntactic exponence'  Verbal periphrasis in German
and Estonian.

  ----- 2001  'Realisation-based Lexicalism'. Journal of Linguistics
37(2):317-27. [Review of Ackerman and Webelhuth]

Bouma, Gosse, Frank van Eynde, and Dan Flickinger. 1998. Constraint-based
lexicons. [online version is available at http://www.let.rug.nl/alfa/
"Papers" > "1998"]

Calder, Jo (1989). 'Paradigmatic Morphology'. Proceedings of the 4th
conference of the European chapter of the Association for Computational
Linguistics.

  ----- (1990). An Interpretation of Paradigmatic Morphology. Ph.D. Thesis
Univ. of Edinburgh.

Erjavec, Tomaz (1996) "Unification, Inheritance and Paradigms in the
Morphology of Natural Languages". Ph.D. dissertation, Ljubljana University.
[Available on-line from http://nl.ijs.si/et/Thesis/.]

Kathol, Andreas (1999). Agreement and the Syntax-Morphology Interface in
HPSG. In Robert Levine and Georgia Green (eds.) Pp. 223-274. Cambridge:
Cambridge University Press.

Koenig, Jean-Pierre (1999). "Lexical Relations". Stanford: CSLI publications.

Krieger, H.-U. and Nerbonne, J. (1991). 'Feature-based Inheritance Networks
for Computational Lexicons'. Proceedings of the ACQUILEX Workshop on
Default Inheritance in the Lexicon. Cambridge: University of Cambridge
Computer Laboratory.

Krieger, H.-U., Pirker, H., and Nerbonne, J. (1993). 'Feature-based
Allomorphy'. Proceedings of the 31st Anual Meeting o the Association for
Computational Linguistics, 140-147.

Luis, Ana and Louisa Sadler. 2002. `Object clitics and marked morphology.'
Paper presented at the 2001 CSSP, Paris, October 2001. To appear in an
edited volume, 2002. [available on-line at:
http://privatewww.essex.ac.uk/~louisa/newpapers/]

Meurer, Detmar (2001) On Expressing Lexical Generalizations in HPSG. Nordic
Journal of Linguistics 24 (2). pp. 161-217. [An electronic copy of the last
version Detmar had the copyright for is at:
http://ling.osu.edu/~dm/papers/lexical-generalizations.html]

Mueller, Stefan (to appear, 2002). Complex Predicates: Verbal Complexes,
Resultative Constructions, and Particle Verbs in German. Stanford: CSLI
publications.

Nordlinger, Rachel and Louisa Sadler. 2002. `Morphological composition
revisited.' Paper presented at the LFG02 Conference, Athens, July 2002. (To
appear in _Proceedings of the LFG02 Conference_).

Riehemann, Susanne (1998) "Type-Based Derivational Morphology". Journal of
Comparative Germanic Linguistics 2, 49-77.

Sadler, Louisa and Andrew Spencer. 2001. `Syntax as an exponent of
morphological features.' In Geert Booij and Jaap van Marle (eds.) _Yearbook
of Morphology 2000_, 71-96. Dordrecht: Kluwer.

Spencer, Andrew. (2001a) 'Periphrastic paradigms in Bulgarian'. Submitted
for publication in Junghanns & Szucsich (eds.) Syntactic Structures and
Morphological Information. [Revised version, 11 December 2001.]

  ----- (2001b) 'The Word-and-Paradigm approach to morphosyntax'.
Transactions of the Philological Society, 99:279-313,

Stump, Gregory (2001). Inflectional Morphology. Cambridge University Press.

Van Eynde, Frank  (1994). Auxilaries and verbal affixes. A monostratal
cross-linguistic analysis. Habilitationsschrift, University of Leuven,
1994. (especially chapters 4 and 5).