Thursday 13 January 2011

Recommended Reading 1

Over the last few months, quite a few of you have been kind enough to read some of the other posts on this site. Today, I have no particular axe to grind, though I'm mindful of the other pieces that are still pending, including the final section of Sapir-Whorf Redux, which should appear early next week.) Instead, nach dem Motto* "Anyone can do any amount of work, provided it isn't the work they are supposed to be doing that the time (Robert Benchley)," I'd like to use this piece to draw your attention to a writer and critic who—in the world of linguistics, at any rate—has been sorely neglected in recent years. This is a real pity, not just because it's a waste of everyone's time to start thinking about linguistic issues from scratch when someone else has done the groundwork, but also because he writes so well: whether scathing or complimentary, his prose is unfailingly precise and elegantly constructed.

The writer in question is George Steiner. You can find out more about him  by following this link. Rather than rehearsing that description, here are three extended quotes from Language and Silence, which I happen to have in front of me. The collection After Babel is more relevant still, but I don't have a copy to hand.

  • From Humane Literacy
When he looks back, the critic sees a eunuch's shadow. Who would be a critic if he could be a writer? Who would hammer out the subtlest insight into Dostoyevsky if he could weld an inch of the Karamazovs, or argue the poise of Lawrence if he could shape the free gust of life in The Rainbow. All great writing stems from le dur désir de durer, the harsh contrivance of spirit against death, the hope to overreach time by force of creation. 'Brightness falls from the air'; five words and a trick of darkening sound. But they have outworn three centuries. Who would choose to be a literary critic if he could set verse to sing, or compose, out of his own mortal being, a vital fiction, a character that will endure. Most men have their dusty survivance in old telephone directories (it is a mercy that these are kept at the British Museum): there is in the literal fact of their existence, less of life's truth and harvest than in Falstaff or Mme de Guermantes. To have imagined these. (Language & Silence: 21)
  •  From The Retreat from the Word
The Apostle tells us that in the beginning was the Word. He gives us no assurance as to the the end.
It is appropriate that he should have used the Greek language to express the Hellenistic conception of the Logos, for it is to the fact of its Greco-Judaic inheritance that Western civilization owes its essentially verbal character. We take this character for granted. It is the root and bark of our experience and we cannot readily transpose our imaginings outside of it. We live inside the act of discourse. But we should not imagine that a verbal matrix is the only one in which the articulations and conduct of the mind are conceivable. There are modes of intellectual and sensuous reality founded not on language, but on other communicative energies such as the icon or the musical note. And there are actions of the spirit rooted in silence. It is difficult to speak of these, for how should speech justly convey the shape and vitality of silence? But I can cite examples of what I mean...
...The tools of mathematical analysis transformed chemistry and physics from alchemy to the predictive sciences they now are. By virtue of mathematics, the stars move out of mythology into the astronomer's table. And as mathematics settles into the marrow of a science, the concepts of that science, its habits of invention and understanding, become steadily less reducible to those of common language.
It is arrogant, if not responsible, to invoke such basic notions in our present model of the universe as quanta, the indeterminacy principle, the relativity constant of the lack of parity in so-called weak interactions of atomic particles, if one cannot do so in the language appropriate to them — that is to say, in mathematical terms. Without it, such words are phantasms to deck out the pretence of philosophers or journalists. Because physics has had to borrow from the vulgate, some of these words seem to retain a generalized meaning; they give a semblance of metaphor. But this is an illusion. When a critic seeks to apply the indeterminacy principle to his discussion of action painting, or of the use of improvization in certain contemporary music [or of Minimalist syntax? NGD], he is not relating two spheres of experience; he is merely talking nonsense**...
  • From To Civilize our Gentlemen
...[T]he student of literature now has access to and responsibility towards a very rich terrain, intermediate between the arts and science, a terrain bordering equally on poetry, on sociology, on psychology, on logic, and even on mathematics. I mean the domain of linguistics and of the theory of communication.
Its expansion in the post-war period is one of the most exciting chapters of modern intellectual history. The entire nature of language is being re-thought and re-examined as it has not been since Plato and Leibniz. The questions being asked about the relations between verbal means and sensory perception, about the ways in which syntax mirrors or controls the reality-concept of a given culture, about the history of linguistic forms as a record of ethnic consciousness — these questions go to the very heart of our poetic and critical concern. The precise analysis of verbal resources and grammatical changes, which may soon be feasible by means of computers — these may have a bearing on literary history and interpretation. We are within reach of knowing the rate at which new words new words enter a language. We can discern graphic contours and statistical patterns relating linguistic phenomena to economic, sociological changes. Our whole sense of the medium is being re-evaluated.
Let me give only two examples which are familiar to any student of modern linguistics. There is a Latin American Indian language, indeed there are a number, in which the future — the notion of that which is yet to happen — is set at the back of the speaker. The past, which he can see, because it has already happened, lies all before him. He backs into the future unknown; memory moves forward, hope backwards. This is the exact reversal of the primary co-ordinates by which we ourselves organize our feelings in root metaphors. How does such a reversal affect literature or, in a larger sense, to what extent is syntax the ever renewed cause of our modes of sensibility and verbal concept? Or take the well-known instance of the astounding range of terms — I believe it is in the region of one hundred — by which the gauchos of the Argentine discriminate between the shadings of a horse's hide. Do these terms in some manner precede the perception of the actual nuance of colour, or does that perception, sharpened by professional need, cause the invention of new words? Either hypothesis throw rich light on the processes of poetic invention and on the essential fact that translation means the meshing of two different world images, of two different patterns of human life (Language & Silence: 86-87)

Of course, there is much to disagree with in Steiner's writing. Take, for example, his easy acceptance of what might be called 'the Whorfian fallacy': his discussion of the colour terms of Argentinian gauchos immediately calls to mind the old saw about the Eskimos having a hundred words for snow (see Geoff Pullum's corrective The Great Eskimo Vocabulary Hoax). It is also true that much of Steiner's writing reflects a near obsessive concern with the Holocaust and with Germany and the German language. As understandable as this may be, given his personal history and the immediate post-war period that informed his criticism, this concern overshadows many of his insights, at times to the point of obscurity. In spite of all this, though, one cannot help but be respectful of the scholarship, engaged by the writing, and impressed by the humanism. To read more, click on the links below, or borrow the books from your university library.





*Ill-translated by dict.cc as 'along the lines of'. One of the great things about Steiner's writing, but perhaps also a reason why he is less read than he deserves to be, is that he freely mixes French, German, Italian, Latin and Greek terms and quotations into his English writing, without pandering glosses or translations: in other words, he assumes that his reader is an educated European. Unfortunately, as he himself is painfully aware, this is simply not true, the complacent—often willful—ignorance of foreign languages among even highly educated people being one of the most unattractive features of middle-class British and American chauvinism.

**In a footnote, Steiner comes close to retracting this last comment. On balance, my own judgement is that the original assertion was correct: most analogies to physics and mathematics—in formal linguistics, at least— are more pretentious than insightful.

Sunday 9 January 2011

Defining Goals (continued)

[This is a draft excerpt from Chapter 2 of a proposed monograph on Vietnamese, in which I try to tackle some general theoretical problems. As ever, I really would appreciate comments, and will incorporate feedback in future drafts. Thank you. PS. If you wish to cite this, please reference it as Duffield, Nigel Particles and Projections in Vietnamese Syntax, draft ms., University of Sheffield.]

...Part of the difficulty here stems from an inconsistency within Chomsky’s own writings about the goals of generative theory. On the one hand, there is the notion of Explanatory Adequacy, which as just discussed asserts a direct inferential relationship between a particular analysis of some core properties of grammar and the ability of children to acquire their native language. I shall return to this notion presently. On the other hand, in separate passages (1981, 1985, 1988, etc.) Chomsky has explicitly and consistently distinguished questions about language knowledge from those of language acquisition, in terms that again should be very familiar. The following excerpt from Chomsky (1988) is representative:
(i) What is Knowledge of Language? What is in the mind/brain of the [adult native] speaker of English or Spanish or Japanese?; (ii)  How does this system of knowledge arise in the mind/brain? How is this knowledge put to use in speech (and secondary systems such as writing)?; (iii) What are the physical mechanisms that serve as the material basis for this system of knowledge and for the use of this language? (Chomsky 1988: 3).

Chomsky’s “three questions”—which crucially are five—notionally distinguish epistemological from psychological or physiological concerns. As such, they have been used to demarcate separate disciplines: Linguistics, Psycholinguistics, and Neurolinguistics, respectively. If one takes the first of these questions (What is Knowledge of Language?) as a starting point, rather than the notion of Explanatory Adequacy, then not only does the generativist approach to theory construction become much more perspicuous, but the practice of neglecting acquisition data becomes markedly more understandable. Primarily, this is because the goal of linguistic theory can then be stated as the development of an abstract Theory of Linguistic Knowledge: a Level 1 (Computational) Theory, in the sense of Marr (1982).[i]  By definition, such a theory is at least one step removed from questions of psychological representation or process, and several more steps removed from issues of neurophysiological implementation.[ii]

Notice, however, that Chomsky implicitly identifies a Level 1 question—What is Knowledge of Language?—with what is properly a Level 2 question—namely, What is in the mind/brain of the [adult native] speaker of English or Spanish or Japanese?—implicitly treating the second as a paraphrase of the first.[iii] For Chomsky and many mainstream generativists,[iv] the identification of these two questions is natural and unproblematic, and this is the crux of the difficulty: for reasons that I suppose are once again ultimately grounded in apriori commitment to innateness, Mainstream Minimalism assumes that essentially the same grammatical knowledge is in the mind/brain of the adult speaker of any language, be it English, Spanish, Japanese…or Vietnamese, and thus that there is no theoretically interesting variation in L from one language to the next. This is explicit in the following quote:
‘In the absence of compelling evidence to the contrary, assume languages to be uniform, with variety restricted to easily detectable properties of utterances. (Chomsky 2001:2)’
This assumption is licensed by the identification of L (a theory of the attained knowledge of a particular language) with FL (a theory of the initial state). The following quotation is instructive:
I understand L (for me, some variety of English) to be an attained state of a genetically-determined faculty of language FL... (Chomsky, cited in Stemmer 2009).
In fact, it is reasonable to conclude from much of Chomsky’s writing that the only theoretically relevant (or interesting) properties of L are those that also properties of FL. This strict identification of L with FL constitutes a theoretical watershed of sorts for many linguistic researchers: certainly, it is one of the clearest points of divergence between the perspective adopted in this monograph and that of Mainstream Minimalism, inasmuch as I am open to the idea that many of the interesting features (including universal properties) of L are emergent properties of particular grammatical systems).[v]
To ground this discussion—which has already become unduly abstract—the question that this monograph attempts to answer is the following more explicitly stated version of Chomsky’s second question (that is, “question (1b)”):
(Leaving aside Saussurean arbitrariness), what syntactic knowledge is in the mind/brain of the adult native speaker of Vietnamese, and how does this differ from that of the grammatical knowledge in the mind/brain of the adult native-speaker of Chinese or Thai or English?
Implicit here—contrary to Mainstream Minimalist assumptions—is the suggestion that there are significant, theoretically interesting differences between the attained state of LViet and (as it might be) the attained state of LEnglish, and also the assumption that a formal theory of L can and should describe and explain those differences: that is to say, that local variation arising from linguistic experience and interaction during language development is an intrinsic property of particular theories of L, not simply a contingent property of the interfaces between an invariant syntax and an arbitrary lexicon. Ideally, the same theory should also account for the structural commonalities between LViet and LEnglish, as well as for the absence of certain types of non-occurring L’s; see, for example, Pinker & Jackendoff (2009), for discussion of this point. Absent from this question, it should be noted, is any direct reference to FL (the theory of the initial state) either as an explanatory device or as a justification for empirical inquiry.[vi]

Of course, it may turn out that there are no interesting grammatical differences between LViet and LEnglish (‘beyond Saussurean arbitrariness’), and that the formal commonalities across languages are then best explained in terms of the initial state (FL): in other words, that Chomsky’s various identifications are justified. But given this construal of the goals of linguistic theory, these become entirely distinct empirical questions of fact, not apriori assumptions. And while there may be reasons to hope—at least to those favourably predisposed to the overall generativist programme—that there are universally shared properties of attained states, it is worth noting that the results of fifty years of research into (surface) language universals have led many to the conclusion that this hope is a vain one (Evans & Levinson 2009, and supportive commentaries). Nevertheless, as discussed in Duffield (2010) and further below, this may tell us very little about FL, where I believe there are more grounds for optimism.

An Aside: Taxi Drivers and the Initial State

In order to appreciate the difference in perspective that results from the dissociation of theories of the initial state (FL) from those of the attained grammatical knowledge of a particular language (LEnglish, LViet, etc), it is instructive to consider the development to steady state of other localized and specialized cognitive skills that also have a significant innate component. 

One pertinent example is human spatial memory, as it applies to navigational skills. It is uncontroversial to claim that the capacity to record spatial information about one’s environment, and to use this stored knowledge to navigate through space in daily life, is in certain crucial respects innate: all arguments that might be adduced from Poverty of Stimulus, lack of negative evidence, absence of instruction, relatively uniform success in threshold acquisition, selective neuropsychological breakdown, and so forth, apply equally to the development of spatial memory as to language, such that it is appropriate to speak of a spatial memory faculty (‘FSM’). Moreover, there is general consensus about neural localization of spatial memory: in all mammals, the hippocampus is crucially implicated in the storage and processing of memories about spatial positioning and orientation; while in humans, there is also believed to be an asymmetric involvement of neocortical structures in spatial memory, with significantly greater right hemisphere involvement in processing spatial memory (see, for example, Nunn et al, 1999).

For reasons that seem to have little to do with the intrinsic properties of the phenomena themselves, and almost everything to do with folk psychology, people strongly, if erroneously, believe that they teach their children language, but don’t believe that they (need to) teach their children to remember where their bed is, or how to find the back yard. In other words, the initial state of the FSM is of little theoretical interest to Cognitive Psychology, since its innateness is uncontentious. By contrast, what is of central interest to psychological theories of spatial memory is the characterization of the attained steady states, developmental changes in spatial memory representations, and the extent to which interaction with the environment can produce adaptive changes in cognitive representation and processing.

A dramatic example of such adapation beyond the initial state is provided by the findings of Maguire et al. (2000), who studied the neurophysiological correlates of acquired spatial navigation skills in London taxi-drivers: the researchers found that the posterior regions of the hippocampi of taxi-drivers were significantly larger than those of a control group, that the anterior portions were significantly smaller; furthermore, that this physical asymmetry increased with years of taxi-driving experience. Results such as these provide striking evidence of adult adaptations in brain regions associated with spatial memory and navigation. The point of this example is that adequate theories of spatial memory, however concerned they may be with general, universal properties of this mental faculty, treat their object of inquiry as an adaptive, dynamic system of knowledge, whose character is partially determined by local environmental factors, and which is subject to theoretically interesting variation (even if this variation is ultimately constrained by biology). An adequate theory will accommodate and explain this emergent variation and identify as precisely as possible which properties of the input are able to induce adaptation in the internalized system of knowledge. A theory of spatial memory that restricted attention only to apriori or initial state knowledge, and which dismissed all subsequent development or relegated such variation to ‘legibility conditions at the interface’ would hardly be considered ‘explanatorily adequate’.[vii]

Returning to language, my judgment is that the Innateness question has led many Minimalists to a near obsessive concern with the initial state of the language faculty (FL). This obsession has antagonized, and continues to provoke, a significant segment of the Cognitive Science community. More importantly however, it has distracted attention from arguably more interesting questions about attained states and the constraints on variation in adult grammars (L). Whether one accepts the innateness and domain-specificty of the capacity for language creation linguistic—as I do—or rejects it (as do many others), should be largely irrelevant to understanding and developing theories of steady state knowledge (just as, for the most part, theories of human genetics and embryology have been irrelevant to cognitive theories of spatial memory). Dissociating FL from L, as is proposed here, and focussing attention on the latter allows one to ask much more dispassionate questions about observed commonalities and differences across I-languages, and about the design properties of theories of L, without the distraction of a set of particular ideological commitments (and the concomitant impulse to trivialize variation while accentuating apparent similarities).

[to be continued]

[i] It should be noted that Chomsky himself directly rejects the relevance of Marr’s levels for linguistic inquiry:
“As for Marr's famous three levels of analysis, he was concerned with input-output systems (e.g., the mapping of retinal images to internal representations). Language is not an input-output system. Accordingly, Marr's levels do not apply to the study of language, though one could adapt them to the very different problem of characterizing cognitive systems accessed in processing and production (Chomsky, cited in Stemmer 2009).”
There is much that one might take issue with here, but this is not an appropriate place for such a debate.
[ii] It is at least debatable whether Linguistic Theory—or Cognitive Psychology more generally—really has, or should aspire to, this “Level 1 status”: see Peacocke (1986a, 1986b) and commentaries, cf. Laurence (2003), Soames (2008), for discussion, or indeed, whether such a classification is relevant to linguistics at all (Chomsky denies this: see previous note). But if one assumes this to be the case, then the generative perspective on questions concerning “lower level” theories becomes much more understandable: it would seem absurd to deny that if Knowledge of Language (KOL) is mentally represented that it should not be put to use in language comprehension and production, or that it should be physiologically represented in any other part of the body than in areas of the brain associated with higher cognition: the job of the psycholinguist and neurolinguist, respectively, then comes to be to determine precisely how KOL is put to use in processing language, how it is acquired, and how it is neurophysiologically realized (i.e, Questions “two” and “three” (really, three and five)).
[iii] This move immediately recalls a different, though equally unhelpful case of “systematic ambiguity”, namely, that concerning grammar:
Using the term ‘grammar’ with a systematic ambiguity to refer, first, to the native speaker’s internally represented ‘theory of his language’ and, second, to the linguist’s account of this, we can say that the child has developed and internally represented a generative grammar in the sense described. […] we are again using the term ‘theory’ — in this case ‘theory of language’ rather than ‘theory of a particular language’ — with a systematic ambiguity to refer both to the child’s innate pre-disposition to learn a language of a certain type and to the linguist’s account of this (Chomsky 1965: 25).
[iv] Notwithstanding the prior quote from Chomsky (1975) above.
[v] An imperfect, but passable analogy here might be to theories of automobile construction in a world where car engines were sealed at the factory door: the Mainstream Minimalist assumption would be that all cars (Ferraris or Fords) are powered by precisely the same engine—whatever is built up around this core represents only superficial difference. The alternative Minimalist view assumes certain fixed and abstract initial conditions—for example, all engines might be assumed derive power through internal combustion, or to transmit power to the driveline via a transmission system: beyond these abstract design properties, the alternative Minimalist view would admit a large range of ‘locally optimal’ solutions to the problem/fact of ground propulsion.  (The principal imperfections of this analogy are of course that the car engine, unlike the language faculty, is an inorganic, manufactured construct, whose theory is largely defined by its function: as a consequence, it is physically and functionally separable from the rest of the system in which it is embedded, and shows no positive development or growth from the point of initial manufacture to ultimate scrappage. Only the most extreme nativist would claim that the latter characteristics are properties of the language faculty.)
[vi] It should be stressed that I do not consider questions about the initial state FL to be irrelevant to present concerns. On the contrary: the following stages of this discussion are intimately and directly concerned with such issues. However, the point is that I assume that there can be very different answers to the questions—What is L?, What is FL?—respectively (especially where L = LEnglish, arguably less so where L = LViet); also, that cross-linguistic grammatical differences are theoretically at least as interesting as cross-linguistic universals.
[vii] Similar remarks apply to theories of musical competence; see Duffield (2011c).