Читать книгу A Companion to Chomsky - Группа авторов - Страница 65

SPE is explicit in its commitment to nativism:

The significant linguistic universals are those that must be assumed to be available to the child learning a language as an a priori, innate endowment. That there must be a rich system of a priori properties—of essential linguistic universals—is fairly obvious from the following empirical observations. Every normal child acquires an extremely intricate and abstract grammar, the properties of which are much underdetermined by the available data. This takes place with great speed, under conditions that are far from ideal, and there is little significant variation among children who may differ greatly in intelligence and experience. The search for essential linguistic universals is, in effect, the study of the a priori faculté de langage that makes language acquisition possible under the given conditions of time and access to data.

It is useful to divide linguistic universals roughly into two categories. There are, first of all, certain “formal universals” that determine the structure of grammars and the form and organization of rules. In addition, there are “substantive universals”² that define the sets of elements that may figure in particular grammars. [p. 4]

Despite these clear statements, nativism has fared badly in the phonological literature in recent years, and we'll discuss two influential trends. Optimality Theory (OT) (e.g. Prince and Smolensky 1993; McCarthy and Prince 1993) began as a strongly nativist model of phonology, consisting of an innate constraint set, built on an inventory of substantive universals, such as the subsegmental features used in SPE. In OT, a universal algorithm determines the output form for a given phonological input based on a language‐specific ranking. In OT, then, the only locus of language variation is in the ranking of the universal constraints in a given language and the set of morphemes that happen to be present in the lexicon of (although there is some confusion with respect to the latter issue, represented in discussion of the so‐called Richness of the Base). For example, an underlying /i/ is able to surface in a closed syllable in English [bit] “beet,” but not in Québec French [pt^sIt], despite the two languages having the same constraints, because the two differ in their constraint ranking.

The initially elegant OT model appeared to offer solutions to some longstanding issues, and introduced a set of different problems—a normal situation in discussion of incommensurable theories in any scientific field. However, it quickly became apparent that the innate component of phonology in OT terms was implausibly rich, requiring an extensional characterisation of a large set of constraints as part of innate endowment. More recently, the issue of nativism is just ignored in most OT literature, but has to be understood as tacitly rejected given the specificity of the constraints posited, such as “Assign one violation for each contrast between N and NC in which NC does not have an oral release that belongs to category of 4 or larger along the RELscale,” which is an actually proposed constraint in a paper in the journal Phonology (Stanton 2019). In addition, various scholars posit constraints that are not just specific to one language, but to the realization of one morpheme in a particular language, such as the ALIGN‐um‐L markedness constraint that refers to the Tagalog morpheme /um/ (Kager 1999, p. 122).

Outside of the OT literature, we find anti‐nativist titles such as The emergence of distinctive features (Mielke 2008) and “Phonology without universal grammar” (Archangeli and Pulleyblank 2015). How is it possible that Chomsky's stature in phonology is universally recognized by the phonological community, as is his development of the idea of Universal Grammar, and yet, so much current work, even by his former students, is anti‐nativist, anti‐UG? We think we have the answer, but first let's document (with added boldface) the not uncommon denial of nativism via the denial of the argument of the poverty of the stimulus (APoS) in the phonological literature.

Perhaps it is not surprising that a psychologist like Peter MacNeilage rejects the APoS:

Example (6.1) Peter MacNeilage, The origin of speech (2008: 41):

however much poverty of the stimulus exists for language in general, there is none of it in the domain of the structure of words, the unit of communication I am most concerned with. Infants hear all the words they expect to produce. Thus, the main proving ground for UG does not include phonology.

This anti‐nativist perspective is also implicitly anti‐internalist since it suggests that words are out there for children to hear, a view at odds with the generative program as understood by, say, a syntactician like Howard Lasnik (2000, p. 3):

The list of behaviors of which knowledge of language purportedly consists has to rely on notions like “utterance” and “word.” But what is a word? What is an utterance? These notions are already quite abstract. Even more abstract is the notion “sentence.” Chomsky has been and continues to be criticized for positing such abstract notions as transformations and structures, but the big leap is what everyone takes for granted. It's widely assumed that the big step is going from sentence to transformation, but this in fact isn't a significant leap. The big step is going from “noise” to “word.”

MacNeilage's view also has an unfortunate English‐centred perspective on words. In an agglutinative language like Turkish, a given root can appear in hundreds or thousands of words, such as the root ev “house” in the noun evlerimizdekilerinki meaning “the one belonging to the ones in our houses” (Hankamer 1989, p. 397). Children clearly do not hear all the words that they can produce.

Phonologist Jeff Mielke's work arguing against an innate set of phonological features also rejects APoS:

Example (6.2) Mielke (2008), The Emergence of Distinctive Features

 “Many of the arguments for UG in other domains do not hold for phonology. For example, there is little evidence of a learnability problem in phonology.” [p. 33]

 [Most of the evidence for] “UG is not related to phonology, and phonology has more of a guilt‐by‐association status with respect to innateness.” [p. 34]

Although Mielke dedicates a whole monograph to arguing that features can be learned, he never addresses logical arguments against such a view presented by Fodor (1980) and others, or the clear assertion by Chomsky and Halle (1965) that “one does not ‘construct features from scratch for each language’” in a response to the American Structuralist linguist Fred Householder (1965).

In a discussion of Chomsky's legacy, it is most relevant to note the divergent perspectives on such matters of phonologists who have worked in the generative tradition:

Example (6.3) Archangeli and Pulleyblank (2015) “Phonology without universal grammar”

 “See Mielke [2004/8] on why features cannot be innately defined, but must be learned”

 “[Children face] the challenge of isolating specific sounds from the sound stream”

 “the predictions of [Emergent Grammar] fit the data better than do the predictions of UG.”

Example 6.4 Blevins (2003, p. 235), Evolutionary Phonology

“Within the domain of sounds, there is no poverty of the stimulus. [I offer] general arguments against the ‘poverty of stimulus’ in phonology, …[there is no evidence that] regular phonological alternations cannot be acquired on the basis of generalizations gleaned directly from auditory input.”

These passages clearly locate segments in the auditory input, the “sound stream,” and thus reject the radical rationalist‐internalist perspective of Chomskyan phonology alluded to above, simply expressed by Hammarberg (1976, 354): “[I]t should be perfectly obvious by now that segments do not exist outside the human mind.” Scholars denying a poverty of the stimulus problem for phonology fail to recognize the well‐established difficulty of defining invariant acoustic correlates of segments, a challenge known as the Problem of the Lack of Invariance, which is discussed insightfully by philosopher Irene Appelbaum (1996).

Finally, here is yet another phonologist, author of a popular text and co‐editor of a philosophically oriented volume on phonological knowledge (Burton‐Roberts et al. 2000) rejecting phonological internalism and nativism in one breath:

Example (6.5) Carr (2006), “Universal grammar and syntax/phonology parallelisms”

“Phonological objects and relations are internalisable: there is no poverty of the stimulus argument in phonology. No phonological knowledge is given by UG.”

Carr's quotation makes clear the logical relationship between internalism and nativism, and helps us to explain the rejection of nativism as a logical consequence of a failure to appreciate internalism: if phonology is internalisable, it need not be innate. A non‐Chomskyan linguist or psychologist studies the mind, and so is an internalist, but he or she may very well deny any interesting domain‐specific innate knowledge—think of your average connectionist. So, internalism definitely does not imply nativism. However, there is a valid implication in the other direction: Universal Grammar is a claim of innate knowledge, so a nativist in cognitive science has to be an internalist. By contraposition we know that if “nativism implies internalism” is true, then “Not‐internalism implies Not‐nativism” is also true. And that's the problem: For many phonologists, it is logically impossible that they be nativists, because they are not internalists.

Example (6.6) The relationship between internalism and nativism:

So, given the rejection of the internalist legacy of SPE, the nativist legacy stands no chance. This anti‐internalism is clear from claims in phonology, phonological acquisition, and speech perception literature that features, segments, alternations, patterns and so on are in the signal, to be found by the listener / learner. We know, however, that just as a Necker cube is not discovered by our visual faculty, but instead is invented by the mind each time it is perceived or imagined (Marr 1982), in the same way, the mind constructs phonological representations out of a limited set of innately available resources such as features. We even see this happening when we interpret the sounds of a parrot or Hoover the “talking” seal as speech—the words and segments we “hear” are definitely not out in the world.