Prague School influence: summary

[George Lakoff]

Dear Fritz,

I'm delighted to see you looking at the influence of the Prague School. Let
me add a bit to what you have.

First a correction:

To my knowledge, Halliday had no influence at all on the introduction of
discussions of topicality into generative semantics. In my recollection, we
starting discussing topicality in 1965-66, under the influence of Dwight
Boliger
(who was my Harvard colleague and audited the course Haj and I gave there
and at MIT in fall 1966) and Susumu Kuno, who ran the project Haj and I
worked on and was talking with us about topicality from the perspective of
the Japanese syntax,  where wa-constructions indicate topicality and are
one of the most prominent syntactic features of the language. McCawley, of
course, was also thinking about wa-constructions in Japanese. I never met
Halliday till 1968, when he came to Cambridge and, as I recall, talked
about other issues.

Incidentally, the person who puched Prague School ideas on information flow
the most here at Berkeley during the 70's was Wally Chafe, who of course
was a major force behind the formation of the functionalist school first
here and later at Santa Barbara. His influence here has been persistent.

About features: I believe they came into generative grammar through the
influence of Roman Jakobson. I studied with Jakoboson in 1961, while I was
taking Halle's introductory course. Morris was using the Jakobson, Fant and
Halle text to provide phonetic motivation for features in phonology. During
1962 while I was starting grad school at Indiana, I read Lees' MIT thesis
(he was Noam's first PhD student) and the early Stockwell-Schacter papers
from UCLA, both of which used really ugly subcategorization rules with no
features. In the spring of 1963, I wrote a paper for Fred Householder at
Indiana suggesting the introduction of features into transformational
grammar, as a way of improving on the work of Lees and Stockwell-Schacter.
When I went to Cambridge on spring break I showed it to Noam, who showed no
particular interest, and said that Hugh Mathews had written a paper in 1958
suggesting the use of features in syntax while he was working on Yngve's
mechanical translation project.I never found that paper. I continued during
my graduate work to think and write using features. My first paper on
generative semantics in the summer of 1963 used them. When Postal came out
to Indiana in the summer of 1964 to teach at the Linguistic Institute, he
was using features as well and said that Chomsky had adopted them. And of
course, the 1964 paper on semantics in genrative grammar by Katz and Fodor
used them.

About dependency grammar:  My introduction to dependency grammar came from
the writings of David Hayes and Jane Robinson in computational linguistics,
which
I read as a grad student at Indiana around 63 or 64. As I recall, the
Kuno-Oettinger computational linguistics project at Harvard around 1965
used dependency representations (but my memory is hazy). I assumed they
came from the Hayes tradition. Bill Woods worked on that project. When I
moved to Michigan I began playing around with dependency representations
(in unpublished notes) trying to see if they could replace phrase
structures.  Woods came out with his ATN grammars in the early 70's. When I
read his work, I went back to my earlier notes and realized that phrase
structures and transformations could indeed be eliminated using gramamtical
relations formalized as dependecy structures. Henry Thompson and I made
such a proposal in our 1975 BLS paper on Cognitive Grammar. These ideas
were carried over in my 1977 CLS paper "Linguistic Gestalts" which was a
precursor to my early work with Fillmore moving toward construction
grammars. Those dependency grammar ideas have now come to prominence again
in the current development of the Neural Theory of Grammar here at
Berkeley. I don't know where Hayes got his dependency ideas from, but I
would not be surprised if Tesniere were the source.

Hope the fills out the picture a bit.

George







At 7:46 PM -0800 2/10/99, Frederick Newmeyer wrote:
>A few weeks ago, on this list and on one other, I posted a query about the
>influence of the Prague School on current North American and Western
>European work in syntax. I would like to thank the following, who gave me
>very helpful replies: Machtelt Bolkestein, John Connolly, Geert Craps,
>Deborah DuBartell, Tom Givon, Frank Gladney, Eva Hajicova, Paul Hopper,
>Dick Hudson, John Mackin, Salvador Pons Borderia, Petr Sgall, Sanna-Kaisa
>Tanskanen, Jess Tauber, Yishai Tobin, Marina Yaguello, Fumiko Yoshikawa,
>and several individuals who asked me not to cite them by name.
>
>All respondents agreed that several current schools of syntactic analysis
>that originated in Western Europe owe a great debt to pre-war work in
>Prague. These include Dik's 'Functional Grammar', Halliday's 'Systemic
>Functional Grammar', and Alarcos Llorach's 'Funcionalismo'. I am told that
>these approaches are quite explicit about their debt to Prague, from which
>they derive an integrated structural-functional approach to syntax.
>
>As far the debt of generative grammar is concerned, all agree that any use
>of feature notation is ultimately a Praguean influence. Dependency-based
>generative approaches appear to derive from Prague and, in fact, Charles
>Fillmore in 'The Case for Case' cites Lucien Tesniere, a member of the
>Prague School, for the idea of 'sequence-free representations'. It was
>suggested that the approach within generative semantics to topic and focus
>derived from Halliday, and hence ultimately from the Prague School. And
>there were suggestions that the work on these issues by, say, Michael
>Rochemont (within formal syntax) and Barbara Partee (within formal
>semantics) are generative reinterpretations of Prague School-originated
>generalizations. Partee is also reported to be writing a joint book with
>two Prague linguists.
>
>However, there was wild disagreement among the respondents on the degree
>to which mainstream North American functionalism (and the similar German
>functionalism represented by linguists such as Haspelmath, Heine, and
>Lehmann) is indebted to Praguean work. The opinions I received ranged from
>'deeply indebted' to 'no debt whatsoever'. Those who took the former
>position pointed to the centrality of Prague-originated notions like
>'functional sentence perspective' and 'communicative dynamism' in American
>functionalism (even if these terms are not generally used) and suggested
>linguists like Bolinger, Chafe and Greenberg as being instrumental in
>passing Prague School conceptions on to them. Those who took the latter
>position say that the 'discovery' of Prague work was post hoc and that
>'foundational differences' exist between Prague School functionalism and
>US functionalism. Prague School work was described as being, at one and
>the same time, 'too structural' and 'not structural enough'. Too
>structural in the sense that American functionalists have tended to reject
>the Saussurean idea (adopted by Prague) that a grammar is a system 'ou
>tout se tient'. Not structural enough in the sense that the Prague School
>has tended to advocate a dependency-based analysis, and therefore, unlike
>much of US functionalism, does not formulate generalizations involving
>syntactic constituent structure and the structural relationships based on
>that. (IMPORTANT NOTE: I am summarizing here, not editorializing!)
>
>I realize that neither 'the Prague School' nor 'American functionalism'
>are homogeneous entities, so conflicting responses might well be drawing
>on the work of different scholars within these schools carried out at
>different times. Still, since the respondents did not generally qualify
>their answers by citing persons and times, I have not done so in this
>summary.
>
>Fritz Newmeyer
>University of Washington
>fjn at u.washington.edu