Monday, October 31, 2016

Cultural Logic or Transcendental Interpretation? – Golumbia on Chomsky’s Computationalism

New working paper. Title above, download sites, abstract, TOC, and Introduction below.

Abstract: In The Cultural Logic of Computation David Golumbia offers a critique of Chomsky and of computational linguistics that is rendered moot by his poor understanding of those ideas. He fails to understand and appreciate the distinction between the abstract theory of computation and real computation, such as that involved in machine translation; he confuses the Chomsky hierarchy of language types with hierarchical social organization; he misperceives the conceptual value of computation as a way of thinking about the mind; he ignores the standard account of the defunding of machine translation in the 1960s (it wasn’t working) in favor of obscure political speculations; he offers casual remarks about the demographics of linguistics without any evidence, thus betraying his ideological preconceptions; and he seems to hold a view of analog phenomena that is at odds with the analog/digital distinction as it is used in linguistics, computation, and the cognitive sciences.


Introduction: How can you critique what you don’t understand? 3
Computing, Abstract and Real 3
Computationalism 5
Why Chomsky’s Ideas Were so Popular 8
Hierarchy in Theory and Society 10
The Demise of MT: Blow-Back on Chomsky? 11
White Males, Past Presidents of the ACL, and a Specious Binary 13
Is David Ferrucci a Neoliberal Computationalist? 15
Conclusion: Cultural Logic as Transcendental Interpretation? 17
Appendix 1: Pauline Jacobson on Linguistics and Math 18
Appendix 2: Analog vs. Digital: How general is the contrast? 19

Introduction: How can you critique what you don’t understand?

As many of you know, David Golumbia is one of three authors of an article in the Los Angeles Review of Books that that offered a critique of the digital humanities (DH): Neoliberal Tools (and Archives): A Political History of Digital Humanities. The article sparked such vigorous debate within the DH community that I decided to investigate Golumbia’s thinking. I’ve known about him for some time but had not read his best known book:
David Golumbia. The Cultural Logic of Computation. Harvard University Press, 2009.
I’ve still not read it in full. But I’ve read enough to arrive at a conclusion: Golumbia’s grasp of computation is shaky.

That conclusion was based on a careful reading of Golumbia’s second chapter, “Chomsky’s Computationalism,” with forays into the first, “The Cultural Functions of Computation,” and a look at the fourth, “Computationalist Linguistics”. I published my remarks to the web on July 15, 2010 under the title "Golumbia Fails to Understand Chomsky, Computation, and Computational Linguistics". Since then I’ve made other posts about Golumbia and participated in discussions about his conception of computing at Language Log, a group blog devoted to linguistics. I’ve taken those discussions and added them to my original post to produce this working paper.

* * * * *

Let’s begin where I started in July, with a question: Why Chomsky?

Chomsky is one of the seminal thinkers in the human sciences in the last half of the twentieth century. The abstract theory of computation is at the center of his work. His early work played a major role in bringing computational thinking to the attention of linguists, psychologists, and philosophers and thus helped catalyze the so-called cognitive revolution. At the same time Chomsky has been one of our most visible political essayists. This combination makes him central to Golumbia’s thinking, which is concerned with the relationship between the personal and the political as mediated by ideology. Unfortunately his understanding of Chomsky’s thinking is so tenuous that his critique of Chomsky tells us more about him than about Chomsky or computation.

First I consider the difference between abstract computing theory and real computing, a distinction to which Golumbia gives scant attention. Then I introduce his concept of computationalist ideology and criticize his curious assertion that computational linguistics “is almost indistinguishable from Chomskyan generativism” (p. 47). Next comes an account of Chomsky’s popularity that takes issue with Golumbia’s rather tortured account, which emphasizes their appeal to a neoliberal ideology that is at variance with Chomsky’s professed politics. From there I move to his treatment of the Chomsky Hierarchy, pointing out that it is a different kind of beast from hierarchical power relations in society. The next two sections examine remarks that are offered almost as casual asides. The first remark is a speculation about the demise of funding for machine translation in the late 1960s. Golumbia gets it wrong, though he lists a book in his bibliography that gets it right. Then I offer some corrective observations in response to his off-hand speculation about the ideological demographics of linguistics.

No comments:

Post a Comment