HARRIS THE REVOLUTIONARY: PHONEMIC THEORY BRUCE E. NEVIN Bolt Beranek & Newman/University of Pennsylvania Abstract Harris the Revolutionary: Phonemic Theory Chomsky and Halle's well-known attack on ªtaxonomic phonemics| fails if it is applied to Harris's phonological theory, because for Harris phonemic contrast is an observational primitive. Bloomfield had observed that the study of language required a ªspecial assumption,| which he termed ªthe fundamental assumption of linguistics: we must assume that in every speech-community some utterances are alike in form and meaning| (1933:78). He could see no way to determine repetition (or its obverse face, contrast) except by knowing the meanings of utterances, something which science could not yet specify (1933:75-78). But unsupported assumptions should if possible be eliminated from science. Many of those who followed Bloomfield's line of research sought to eliminate this special assumption in a negative way: they tried to define phonemic contrast by the relative distribution of features in phonetic transcriptions of utterances without reference to their meanings (e.g. Bloch 1948, 1953). Harris (1951 [1947]) took a positive approach. In place of Bloomfield's special assumption he devised the pair test, which captures native speakers' intuitions of repetition/contrast in a controlled way. This makes contrast an observational primitive of the science, rather than something to be defined distributionally. Thus, even the initial segmentation of utterances is a phonemic representation of the contrasts, rather than a merely phonetic transcription. This has numerous ramifications. 1. Introduction Leonard Bloomfield posed the question, when we approach an unknown language, how do we know which utterances are the same and which are different? He said that we can only tell by knowing their meanings. + The features of sound in any utterance, as they might be recorded in the laboratory, are the gross acoustic features of this utterance. Part of the gross acoustic features are indifferent (non-distinctive), and only a part are connected with meanings and essential to communication (distinctive). [+] Since we can recognize the distinctive features of an utterance only when we know the meaning, we cannot identify them on the plane of pure phonetics. (1933:77) Draft July 29, 1993 Not For Attribution -1- -1- Draft July 29, 1993 Not For Attribution Phonemic Theory To recognize the distinctive features of a language, we must leave the ground of pure phonetics and act as though science had progressed far enough to identify all the situations and responses that make up the meaning of speech-forms. In the case of our own language, we trust to our everyday knowledge to tell us whether speech-forms are ªthe same| or ªdifferent.| + In the case of a strange language we have to learn such things by trial and error, or to obtain the meanings from someone that knows the language. (1933:77-78) The study of significant speech-sounds is phonology or practical phonetics. Phonology involves the consideration of meanings. (1933:78) Because there is no antecedent science of meaning, contrast and repetition could be brought into linguistics only by a ªspecial assumption|: The study of language can be conducted without special assumptions only so long as we pay no attention to the meaning of what is spoken. (1933:75) The meanings of speech-forms could be scientifically defined only if all branches of science, including, especially, psychology and physiology, were close to perfection. Until that time, phonology and, with it, all the semantic phase of language study, rests upon an assumption, the fundamental assumption of linguistics: we must assume that in every speech-community some utterances are alike in form and meaning. (1933:78, emphasis in original) This presented a problem: as far as possible, science should avoid special assumptions.1 One way to resolve the problem was to try to define contrast without reference to meaning, by analyzing the distributional relations among observed phonetic features in utterances. To most American linguists following Bloomfield, this seemed to be the only scientific approach. After all, if you are concerned with a relation of contrast, you must first identify the things contrasted, phonetically. On this view, the ªgross acoustic features| are taken as observational primitives, and phonemic contrast is to be defined with respect to the distributional relations found among these primitives in the corpus. Bernard Bloch exemplified this approach. ªContrast between sounds can be defined, I think, on the basis of distribution alone, without the customary appeal to meaning| (1953:59[224]). Since ªthe facts of pronunciation [are] the only data relevant to phonemic analysis| (Bloch 1941:95), phonemes had to be closely identified with their phonetic detail. In his carefully worked-out postulates, Bloch based the segmentation of utterances on articulatory movements (Bloch 1948, postulates 11-16), and made explicit the requirement that all the members of a phoneme have some characteristic ____________________ 1 Thanks are due to to Victor Yngve for clarifying this issue. -2- Draft July 29, 1993 Not For Attribution Harris the Revolutionary phonetic feature in common: ªThe class of all segments + containing a given feature is a phoneme. [+] The feature common to all members of a given phoneme is the characteristic of the phoneme|2 (Bloch 1948, definitions 53.2, 53.12). This constraint appeared necessary in order to determine contrast. Without it, one could not know what was in contrast with what. It turned out not to be so easy to define contrast in this way. Problems arose, for example when the membership of a given phonetic segment in one phoneme or another was indeterminate (neutralization) or ambivalent (overlapping). Bloch (1941) brought these difficulties into focus. These difficulties were targeted in the well-known attack by Generative phonologists on ªtaxonomic phonemics| (Halle (1959), Chomsky (1964), Chomsky and Halle (1965)).3 But the facts of pronunciation are not the only data relevant to phonemic analysis, nor even the uniquely fundamental data. During the same period,4 Zellig Harris proposed a second way to determine contrast without unsupported assumptions. If you want to know which utterances are the same and which are different, ask+a native speaker can reliably tell you. On this basis, Harris devised the pair test and various substitution tests. The pair test distinguishes contrast from repetition. The substitution tests determine a ªlinguistically relevant| segmentation of utterances+linguistically relevant because the segments represent the contrasts and locate them relative to one another within an utterance. Because the segmentation is a segmentation of utterances, phonetic detail is associated with each segment. However, the facts of pronunciation do not determine the segmentation. The linguistic ELEMENTS are defined for each language by associating them with particular features of speech+or rather, differences between portions or features of speech+to which the linguist can but refer. They are marked by symbols, whether letters of the alphabet or others, and may represent simultaneous or successive features of speech, although they may in ____________________ 2 Bloch's definition 53.2 refers to both ªsegments and spans| to cover the case of suprasegmental phonemes, for example where pitch extends over a syllable. The elision is immaterial here. 3 Chomsky introduced the new coinage ªtaxonomic phonemics| (1964:406) as a label for all structuralist phonological theories. Sometimes in later Generativist writings the term ªautonomous| phonemics has been used: ªautonomous because some phonemicists refused to admit grammatical information into their phonological analysis, and taxonomic because sounds were merely classified, ignoring important phonological generalizations expressible by rule| (Hyman 1975:82). 4 Harris developed this solution to the problem before 1947, the signature date in the preface of Harris (1951). Draft July 29, 1993 Not For Attribution -3- Phonemic Theory either case be written successively. The elements will be said to represent, indicate, or identify, rather than describe, the features in question. For each language, an explicit list of elements is defined. The statement that a particular element OCCURS, say in some position, will be taken to mean that there has occurred an utterance, some feature of some part of which is represented linguistically by this element. (1951:14) This makes a fundamental change in phonemics. It means that even the earliest linguistically relevant segmentation, which may be quite redundant, is a representation of the phonemic contrasts of the language. The further work of phonological analysis is no longer for the sake of defining contrast (which is already given), but only for the sake of transforming a less efficient representation into a more efficient one. ªMore efficient| or ªconvenient| means that the grammar may be more simply stated. Harris begins with an arbitrary segmentation + ªarbitrary| because it is not necessarily relevant to the distinctions between utterances which we wish to represent. Purely phonetic considerations are arbitrary in this sense, that is, they cannot ensure the relevance of the segmentation, because the contrasts are socially determined, albeit within physical bounds, and because the articulations and sounds of speech are continuous, and not discrete. Fortunately, it is possible to represent each continuous speech event in such a way that we can then compare various speech events and say that the first is different from the second to such and such an extent. Our ability to do this rests on the observation that in each language we can substitute a close imitation of certain parts of one utterance for certain parts of another utterance without getting any consistent difference of response from native hearers of the changed second utterance. + We therefore set out to represent every utterance by segmental elements which are substitutable for segments of other utterances. (1951:25-26). We represent an utterance by a succession of segments which end at arbitrary points along its duration. + Linguists usually select the segments [ + by articulatory or acoustic criteria, or perceived similarity to] what they have elsewhere + learned to regard as `one sound.' However, neither these nor any other criteria can always show us what points of division will turn out later to be most useful (i.e. which will come out at the boundaries between the eventual phonemes). + This uncertainty leads to no loss in exactness, because later procedures will determine the boundaries of these segments. If the segment divisions arbitrarily selected here do not pass the test of the later procedures, they can be adjusted, and if necessary the utterance can be recorded, anew, -4- Draft July 29, 1993 Not For Attribution Harris the Revolutionary with the symbols that will be chosen for the adjusted segments. (1951:26&fn4) Here, we see a first instance of Harris's characteristic way of ªbootstrapping| his analysis, in which a first approximation is later refined in the light of results that could not have been reached without it. In the substitution tests of Harris (1951, Chapter 4), we imitate an utterance, replacing one of the segments we heard in it with a phonetically different segment heard in other utterances, and we ask (directly or indirectly) whether the two pronunciations are the same; or we may observe a native speaker pronouncing differently what he judges to be repetitions of the same utterance.5 The resulting segmentation is linguistically relevant because each segment corresponds one-one (biuniquely) to a contrast between the pairs of utterances. [T]he representation of speech as a sequence or arrangement of unit elements is intimately connected with the setting up of phonemic distinctions between each pair of non-equivalent utterances. If each utterance were considered by itself, it might be represented as a continuum or as a simultaneity of features which change with time. [+] However, if we match utterances [as described in chapter 4, Phonemic Distinctions], we obtain some individual difference between the members of each particular pair of utterances; that is, we obtain discrete elements each of which represents some particular inter-utterance difference [+ and] which can be combined together. These elements are phonemic distinctions, rather than phonemes; i.e. they are the difference between /k/ and /p/ (more exactly, between tack and tap, between sack and sap, etc.) rather than being /k/ and /p/ themselves. However, for convenience, we will set up as our elements not the distinctions, but classes of segments so defined that the classes differ from each other by all the phonemic distinctions and by these only. [+] In this way we define /k/ to represent all the paired distinctions in which [k] was a member, /l/ to represent all the distinctions in which [l] was a member, and so on. The classes, or phonemes, are thus a derived (but one-one) representation for the phonemic distinctions. The segmentation of Chapter 3 was carried out in order to permit the representation of continuously varying speech to express the discrete elemental phonemic differences. A phonemically written form therefore is not a direct record of some spoken ____________________ 5 Of course, this may work the other way: if the native hearer judges as non-repetitions two utterances we had recorded with the same arbitrary phonetic segments, we may be led to notice some phonetic difference that had previously escaped our notice. Draft July 29, 1993 Not For Attribution -5- Phonemic Theory form, but rather a record of its difference from all other spoken forms of the language. (1951:34-35) So even from the outset there is a one-one or biunique correspondence of segments to contrasts. However, any correspondence between a given segment-contrast pairing and the phonetic features found in one segment or another is coincidental. A phone-contrast or phone-segment correspondence is neigher sought nor required. A given segment at this first stage + a given bit of speech identified as distinctive by this technique + may turn out to "contain" more than one contrast, or to contain the same contrast as some other segment that initially is taken to be different; all of this is sorted out by later methods. The segmentation is carried out only so far as is necessary to distinguish utterances that are found to contrast, and the locations, extent, and temporal relationships of individual points of contrast between utterances may turn out to be quite different from those suggested by this initial segmentation. The phonetic features associated with an element representing one contrast may turn out to extend for the length of a morpheme or word, for example in vowel or consonant harmony. This is equivalent to underspecification and spread of features in Generative phonology today. Halle (1954:335) and Chomsky (1957:234[343]) refer to the pair test as a fundamental starting place for identification of repetitions. However, the ramifications of contrast being given in advance are not developed in Generative phonology. Instead, the notion of contrast is presented just as in the work of Bloch and other ªNeo-Bloomfieldians|, as a function of phonetic differences between physically defined segments, as given by an antecedent study of phonetics. Chomsky (1964:410fn17) says that Bloomfield (1926) took contrast as a primitive notion, but as we have noted it was for Bloomfield a primitive by assumption rather than by observation (test), that is, it was a socalled ªprimitive| with no empirical standing whatsoever. Chomsky (1957, 1964, 1968) argued in favor of phonological representations organized around the distinctive feature theory of Jakobson, Fant, and Halle, carrying forward one line of development of the Prague school phonology of Trubetzkoy and others. The shift from a segmental representation to a distinctive feature representation changed the character of descriptive statements of phonology and in particular the character of rules deriving phonetic descriptions from morphophonemic representations. Bloomfield (1933) had conceived of the phoneme as a bundle of distinctive features, but it appears that he thought of the bundle (the phoneme) as the fundamental thing, an ªindivisible unit| (1933:79), and the features as a part of their descriptive analysis. In the proposals of Jakobson, Fant, Halle, and now Chomsky, the converse is true: the distinctive features were seen as fundamental, with alphabetic, segmental representations -6- Draft July 29, 1993 Not For Attribution Harris the Revolutionary clearly secondary. Indeed, Chomsky & Halle (1965:472) argue that a segmental representation may be dispensed with entirely: We conclude, therefore, that only feature notation has linguistic significance, and that segments are simply to be regarded as conventional abbreviations, utilized to cope with the exigencies of printing but having no linguistic significance in themselves. We will now review the attack on segmental phonemes by Chomsky and Halle. As we will see, it does not apply to Harrisian phonemics, and misrepresents Harris's views. 2. The Four ªTaxonomic| Conditions A number of Chomsky's (1964) arguments hinge on the claim that socalled ªtaxonomic phonemics| adheres to four principles or conditions: linearity, invariance, biuniqueness and local determinacy.6 2.1 Linearity The linearity condition says two things. It says that each phoneme corresponds to a phone or to a sequence of consecutive phones, and that the linear order of phonemes in any given utterance is the same as the linear order of corresponding phones. If phones a, b, and c correspond to phonemes x, y, and z respectively, then linear order abc can only correspond to linear order xyz,never yxz and so on. Chomsky (1964:407) defines the linearity condition as follows: The linearity condition + requires that each occurrence of a phoneme in the phonemic representation of an utterance be associated with a particular succession of (one or more) consecutive phones in its representing matrix, as its ªmember| or ªrealization|; and, furthermore, that if A precedes B in the phonemic representation, then the phone sequence associated with A precedes (is to the left of) that associated with B in the phonetic matrix. ____________________ 6 See also John Fought's discussion of these conditions in Hymes & Fought (197*:**). Where it may be helpful in locating a cited passage, I give references to Chomsky (1964) and Harris (1951) in the form page.paragraph; for example, 1964:417.3 is in the third full paragraph on page 413, and 1951:41.0 refers to a partial paragraph carried over from the preceding page. Page references to Chomsky (1964) are for the reprint in Makkai (19**). Draft July 29, 1993 Not For Attribution -7- Phonemic Theory There is a simpler formulation in an earlier review (1957a:346-347):7 + speech is taken to be literally constituted of a sequence of phonemes, each with its distinctive and redundant features; accordingly, the phonetic value of a sequence of phonemes is the sequence of phonetic values of these phonemes. Harris violates the linearity condition whenever doing so is ªconvenient| for obtaining a simpler or more perspicacious grammar. Such violations are found, for example, in his treatments of simultaneous components (starting with suprasegmentals), and in his analysis of partial overlapping. In cases of vowel or consonant harmony, for example, a phonetic feature is associated with a segment written at the end of a stretch of the utterance and is spread over the preceding segments of that stretch by rule. 2.1 Invariance Chomsky says that the invariance condition + asserts that each phoneme P has associated with it a certain set f(P) of defining features (that is, P=Q if and only if f(P)=f(Q)) and that wherever P occurs in a phonemic representation, there is an associated occurrence of f(P) in the corresponding phonetic representation. The invariance condition has no clear meaning unless the linearity condition is also met; I will assume, then, that it is inapplicable when linearity is violated. [+] Where linearity and invariance are both met by a taxonomic phonemic representation, the string of phones is segmented into successive segments, each of which contains, along with redundant (determined) features, the defining features f(P) of some phoneme (P), and the phonemic representation is just the sequence of these phonemes. ____________________ 7 Chomsky (1957:347) argues, with reference to the writer/rider case, that the distinctive feature theory of Jakobson and Halle would benefit from abandoning the linearity condition or, rather, from abandoning a presumption of phonetic realization on which the linearity condition rests, in favor of an idealized ªhypothetical constructed standard| that need not ever be explicitly realized as such in speech. He cites Harris's (1951:70-71) discussion of ªthis and similar examples| + see also e.g. t- r 1 (39.3, 60.1), fainting (32.1, 41.0), painting-paining (92.1). However, in Harris's account, as in Joos' original report (Joos 1942), the phonetic difference cited for the writer/rider contrast is in the quality of the vowel preceding the flap, not its length (see John Fought's discussion in Hymes & Fought 1981:209-211). An element of length is much more convenient, however, for Chomsky's argument against linearity. -8- Draft July 29, 1993 Not For Attribution Harris the Revolutionary Obviously, the invariance condition is simply a restatement of the requirement, made explicit by Bloch (in his Postulates, 1948, loc.cit.), that all members of a phoneme have some characteristic phonetic feature or features in common. We have already seen that this requirement of phonetic invariance across allophones of a phoneme was not binding on Harris, and why. For Harris, invariance is ªconvenient| for making the description simpler and more efficient (1951:64): It is convenient to have the definitions of the various segments within a phoneme simply related to each other. We may try to group segments into phonemes in such a way that all the segments of each phoneme represent sounds having some feature in common which is not represented by any segment of any other phoneme. If a phonetic feature is shared by all the segments that are grouped as members of a phoneme, then descriptive statements (rules) may refer to ªthe phoneme as repesenting this common feature, rather than as being a class of segments. Relations between phonemes would then represent relations between sound features| (1951:65). Harris makes clear that this formulation is intended to be equivalent to the distinctive feature theory of Trubetzkoy, Jakobson, and their followers (Harris 1951: 125fn4, 146-149), with the crucial proviso (noted e.g. in Harris 1941:346) that, as always, the elements must be defined relative to one another, as representations of contrast, rather than according to any absolute scale. Chomsky distinguishes a strong and weak form of invariance. The difference is in the form of the defining features f(P). The strong form requires that f(P) be specified in terms of absolute phonetic descriptors, and forbids partial overlap. The weak form specifies f(P) as scalar values of phonetic parameters, and admits partial phonemic overlap (1964:408.2): One can distinguish two versions of the invariance condition, depending on whether the features are taken to be relative (i.e., more or less along a certain phonetic dimension) or absolute. [+] Under the absolute invariance condition, partial overlapping is excluded. Chomsky (409.3) identifies Bloch as requiring the strong form, and Jakobson and Harris as accepting the weak form. This is not the same use of the terms ªrelative| and ªabsolute| by Harris. Harris's elements representing the contrasts are associated with phonetic data just because his analysis starts with a segmentation of phonetically specified utterances. It does not matter whether the phonetic specification associated with the segments is given in terms of absolute phonetic data (as nearly as one could approach that) or relative terms as differentiated values on scales of phonetic measurement. The crucial thing is that the phonological elements+the representations of the contrasts+are defined relative to one another. Draft July 29, 1993 Not For Attribution -9- Phonemic Theory It appears that Chomsky intends that the notion of contrast should entail invariance. In discussions of the theory of distinctive features, the features are sometimes spoken of as though they were the contrasts, e.g. +/-voice is ªthe voicing contrast,| etc. This step of reification is seductive, but unwarranted. The feature [+/-voice] is a representation of a contrast (in many languages) which might be represented instead by, perhaps, [+/- delayed VOT]; neither representation is the contrast itself. Generative phonology proposes that a set of phonetic descriptors suffices to identify the contrasts of any language, universally. However, the features of this ªuniversal alphabet| are still labels or representations of the contrasts, not the contrasts themselves. Indeed, they can only be thought of as the contrasts themselves on a presumption of phonetic invariance. If this supposition is correct, then Chomsky's use of the term ªcontrast| presumes weak invariance with respect to universal phonetic parameters such as [+/-voice], and those parameters are thought of as being the contrasts themselves. Phonetic invariance may come into conflict with the next condition, biuniqueness. This is the crux of phonemic overlapping, if you require invariance and linearity as in order to define contrast. But however useful or convenient it may be, for Harris biuniqueness is foremost among the ªmore powerful reasons| (1951:65) that may override invariance. 2.3 Biuniqueness Chomsky's definition of biuniqueness (1964:408.4) is that it asserts that each sequence of phones is represented by a unique sequence of phonemes, and that each sequence of phonemes represents a unique sequence of phones. Harris had introduced the term ªbi-unique relation| in his paper on phonemic long components (1944a:187-188, footnote suppressed):8 Finally, if we are ready to admit partial overlapping among phonemes, we may agree to have different components in different environments represent the same phonetic value. So long as we do not have a component in one environment represent two phonetic values which are not freely interchangeable, or two components or component-combinations in the same ____________________ 8 Hymes & Fought refer erroneously to Harris (1944b), and Anderson (1985) evidently assimilates the error from them, although the page range for (1944b) is 196-211. I am not aware of any other place that the word occurs in Harris's publications. One can only presume that he used it in classes that Chomsky attended. In Harris (1951 [1947]), the earlier term was abandoned in favor of the phrase ªone-one relation|. -10- Draft July 29, 1993 Not For Attribution Harris the Revolutionary environment represent the same phonetic value, we are preserving the bi-unique one-to-one correspondence of phonemic writing. (The term bi-unique implies that the one-to-one correspondence is valid whether we start from the sounds or from the symbols: for each sound one symbol, for each symbol one sound.) In subsequent writings, he used the phrase ªone-one correspondence| instead, and, as we have seen, in Harris (1951) the correspondence is not between phones and phonemes, as in Chomsky's definition,9 but between the phonemic contrasts in an utterance and the elements in a representation of the utterance.10 Harris identifies his initial elements not by consulting phonetic theory, but by consulting the linguistic intuitions of native speakers. Phonetic properties are associated with each segment because, after all, this is a segmentation of utterances. Biuniqueness is not achieved by subsequent procedures, it is entailed by this method of identifying the contrasts. There is intrinsically a one-one correspondence between these segments and the contrasts that they represent. Each distinct utterance is represented by a different sequence of segments, and each different sequence of segments represents a distinct utterance. The ªbiuniqueness condition| is therfore not a formal condition holding between representations at two different levels of description, a phonetic level and a phonemic level. Rather, it is a methodological requirement that one not lose or obscure the fundamental observational data for a science of language, namely, which utterances are repetitions and which are not. ____________________ 9 Hyman (1983:68) appears to equate biuniqueness with invariance. Chomsky's comments (1964:414) on Bloch's two treatments of Japanese phonemics also confuse biuniqueness with invariance, since it was surely not biuniqueness per se but rather phonetic invariance that made for what Bloch saw as the ªgreater accuracy| of his later treatment, and which forced him to accept an analysis that actually obscured the patterning in the language and complicated statements of phonology, morphophonemics, and morphology + something that Harris (or Sapir, or for that matter Bloomfield) would never have done. Some of Bloch's personal motivation here may perhaps be seen in his lifelong interest in dialectology, which naturally focusses on differences of phonetic detail. 10 When Harris (1951) speaks of a one-one correlation of phonemes to segments, the latter (the segments being grouped into a given phoneme) were set up at an earlier stage of analysis so as to have a one-one correspondence to the distinctions (either directly or by way of the segments of still earlier stages of representation). In other words, the biunique relation of representations to the primitive distinctions is transitive through successive reformulations of the representations. Draft July 29, 1993 Not For Attribution -11- Phonemic Theory Harris occasionally speaks of a one-one relation between phonemes and segments.11 This is a telegraphic usage, to avoid awkward repetition of the definition of relationship given at (1951:34-35, quoted earlier), where after all he had said ªfor convenience, we will set up as our elements not the distinctions, but classes of segments so defined that the classes differ from each other by all the phonemic distinctions and by these only.| When Harris describes the grouping of segments into a phoneme, it might appear that one of the requirements involves creating a biunique relation between the phoneme and its member segments, as though he were talking of a biunique relation of phonemes to phones. In fact, the requirement is for the phoneme to preserve the biunique relation to contrasts that its member segments had. All the operations of linguistic analysis, aimed at transforming the initial representation into a more efficient or useful one, must preserve the biunique relation between representations and contrasts. We will return to this under the topic of complementary distribution. We will take a moment to clear up a possible source of confusion about what is intended by the expression ªone-one correspondence.| The contrasts are observed as contrasts between utterances. When one arrives at a maximally efficient representation for the contrasts, the contrasts appear to be relations between the elements of that representation. The (weak) phonetic invariance that we have noted as an implicit tenet of Generative phonology results in a presumption of the form and content of a maximally efficient representation for the contrasts, and a reification of the distinctive features as being the contrasts. This might be a source of confusion. If the contrasts are identified in this way with a non-redundant representation, then it appears that Harris is saying that more than one segment of the initial representation corresponds to the same contrast, and it does not make sense to refer to this as a one-one correspondence. But the contrasts are observed in paired utterances, and the distinctive features are defined across the ªcomplete phoneme stock| (section 7.423). 2.4 Local Determinacy Local determinacy is a qualification of the biuniqueness condition. Chomsky (1964:409.1) defines it as [a one-one] correspondence such that the unique phonemic representation corresponding to a given phonetic form can be determined by ªpurely phonetic| considerations, or perhaps, considerations involving only ªneighboring sounds.| ____________________ 11 See for example 65fn14, 80.0, 85fn16. See also the passage from Harris (1944a:187-188) quoted earlier, although it is possible that when Harris wrote this he had not yet realized the priority of contrast over segmentation. -12- Draft July 29, 1993 Not For Attribution Harris the Revolutionary He says that the condition of local determinacy is difficult to state precisely, because the linguists whose views he is attacking have been frustratingly vague as to what they intended by it. And yet he claims that it is the basic and most widely held sense of the biuniqueness condition.12 So this is Chomsky's attempt to restate, in clear and explicit form, a widely held but inchoate view of what was necessary to maintain linguistics on a scientific footing An obvious candidate is separation of levels. This was the concern, which was indeed widely prevalent in post-war American linguistics, that it would be unscientific to make use of morphology, or other information from ªhigher levels| of the grammar, in the work of determining the phonemic system. This appears to have arisen from an acute awareness of the essential abstractness of linguistic patterning, and a determination that conclusions about it should be reached from a solid, physical foundation. Thus, this concern is quite consistent with phonetic invariance and linearity, and might almost be thought of as following logically from them, or from the same premisses as they. Each step of linguistic analysis must be firmly anchored, through an explicit and rigorously defined succession of preceding steps, to the phonetic ªfacts of pronunciation.| Here is a definition of local determinacy in negative terms as biuniqueness with no ªmixing of levels|: LD. Local determinacy prohibits recourse to the results of morphological or syntactic analysis, since they can be carried out only after the phonological analysis is complete. Chomsky, however, is at pains to distinguish separation of levels, as a methodological condition, from local determinacy, as a substantive (or formal) condition. This distinction arises out of the dichotomy, made at the outset of Chomsky (1964), between an acquisition model and a perceptual model. A methodological condition pertains to an acquisition model, and concerns information relevant to determining the correct choice of a phonemic system. A substantive condition is a formal condition on a phonemic system and the rules that relate it to sound, pertaining to a perceptual model. Separation of levels is a methodological condition for learning the phonemic system, or for a linguistic discovery procedure. Local determinacy is a formal or substantive condition for recognizing ____________________ 12 He says (409) ªapparently it is this, rather than literal biuniqueness in the technical sense, that is required in taxonomic phonemics,| and farther on (413) ªbiuniqueness [is] usually construed + in terms of local determinacy|, and (418fn24) ª[o]ne or another form of this is implicit in all substantive discussions of linguistic procedures that I have been able to locate|. He makes no further substantive reference to this condition, apparently intending ªbiuniqueness| thereafter to signify ªbiuniqueness with local determinacy|. Draft July 29, 1993 Not For Attribution -13- Phonemic Theory phonemes in the phonetic continua of speech once the phonemic system has been determined (or learned). Chomsky argues that the former can shed no light on the latter. All of this has little relevance to Harris. The only indication of local determinacy in Harris's work is as a practical matter: one should extend the environments that one tests no farther than is sufficient.13 A condition of local determinacy would contravene the characteristic ªbootstrapping| approach to linguistic analysis that Harris followed in all his work, in which a first approximation is later refined by criteria that are not initially available, in which tentative guesses as to the results of later stages of analysis (e.g. morphology) are used as guides at an earlier stage, subject to correction when the later work is carried out more fully, and in which earlier results are always subject to re-evaluation in the light of results at a later stage, always with an eye to the overall simplicity of the grammar. This is what Harris means when he says that operations must be ªcarried out for all the elements simultaneously| without any ªarbitrary point of departure| (1951:7). Thus, subsequent rephonemicization could take into account the results of later stages of analysis, including phonemic juncture and the boundaries of morphemes and words. Harris did not have qualms about the scientific status of invoking such entities as junctures and boundaries, and even morphophonemic alternations, in preliminary guesses and later for the sake of rephonemicization, because all his results refer back to the data of contrast (biuniqueness, properly understood). Given this touchstone of validity and scientific rigor, Harris was free to ªbootstrap| his description by later refinements of early approximations and guesses without loss of methodological rigor. This freedom is unavailable to linguists for whom contrast is not an observational primitive, but rather something to be defined by distributional analysis of phonetic primitives. We will turn next, then, to the procedures of distributional analysis. ____________________ 13 There is occasional brief consideration of a ªshort-range requirement|, e.g. /ay__/ rather than /nay__/ as the environment for a morphophoneme /F/ (1951:234fn34), but as is frequently the case Harris is here laying out one of a range of options, subject to various tradeoffs, and not a rigid constraint. One extends environments as far as suffices. For Harris, the reconsideration of the phonemic representation for the sake of simplifying morphophonemic statements (1951:233f) is in a somewhat different sense even more `locally determined' than the earlier stages of phonemic analysis are: although the relevant phonemic environments are no smaller in extent, only the occurrences of those phonemic environments in alternants of morphemes are considered. -14- Draft July 29, 1993 Not For Attribution Harris the Revolutionary 2.5 Summary These four conditions are obligatory if one has only ªthe data of pronunciation| to work with. Without linearity, invariance, biuniqueness (in the phone-phoneme sense) and local determinacy, one would be at a loss to say what might be in contrast with what. They are optional for Harris because his fundamental data are native speakers' intuitions of contrast, identified by the pair test and located in elements of a representation by the substitution tests. The question is not what contrasts with what, but rather what means are available for representing the contrasts that are present. In Harris's (1991) view, the same question faces the child learning its first language. 3. Complementary Distribution Chomsky attacks complementary distribution as (1964:414.2): the central concept of taxonomic phonemics [+], basically, the principle of biuniqueness converted into a procedure. Regarded as an analytic procedure, its goal is to provide the minimally redundant representation meeting the conditions of biuniqueness and local determinacy. Talk of a procedure as ªproviding| or ªleading to| the correct representation suggests that it is a practical discovery procedure, such as might be implemented in a computer algorithm. Harris explicitly states that this is not what he is after (on the discovery procedure canard, see Ryckman 1986, Nevin 1993). More importantly, distributional procedures do not and cannot produce biuniqueness, they preserve it, exactly as operations in mathematical logic preserve truth value. It is essential to preserve the correspondence of representations to contrasts as one wrestles the representations into a form that supports the simplest grammar. 3.1 Tentative Phonemes Claiming to summarize Harris (1951: Chap. 7), Chomsky describes complementary distribution as follows (414.3, italics in original): Given a set of representations in terms of phones, let us define the distribution D(x) of the phone x as the set of (short-range) phonetic contexts in which x occurs. The relation of complementary distribution holds between phones x and y if D(x) and D(y) have no element in common. A tentative phoneme is a class of phones related pair-wise by the relation of complementary distribution. A tentative phonemic system is a family of tentative phonemes meeting a condition of exhaustiveness. We find the phonemic Draft July 29, 1993 Not For Attribution -15- Phonemic Theory system (or systems) by applying additional criteria of symmetry. The most important of the distortions here are these: 1. The two items being compared, x and y, are not of equal status. Chomsky describes the comparison of distributions of two phones x and y. Harris describes the comparison of distributions of a segment x and a ªtentative phoneme| in the making, y. 2. The comparison of D(x) and D(y) is not successive, but recursive. Chomsky describes a linear sequence. A sort of distributional analysis component outputs a set of tentative phonemic systems as alternative candidates, which are subjected first to a test for exhaustiveness, then to ªadditional criteria of symmetry|. The winner output from the test component is ªthe phonemic system (or systems)|. Chomsky objects (1964:414-415) that in some cases ªthe class of `tentative phonemic systems' + will not include the optimal biunique system as a member, so that no supplementary criteria will suffice to select it from this class.| Harris describes a recursive process. The criteria are not applied to a set of alternative phonemic systems produced by distributional analysis, they are applied at every step of merging a segment with restricted distribution into a partially defined phoneme. The comparison of x and y is one step of this recursive process, where x is a segment and y is a phoneme-in-the- making. If the environments of segment x are complementary to those of phoneme y, and if the criteria indicate that an x-y grouping is superior to other concurrent possible combinations, then x is merged into y. That is, the environments of y now include those in which segment x occurred. It is to the less restricted tentative phoneme y that subsequent comparisons are made. The result in the end is of course that all the member segments of any given phoneme are complementary to each other, since complementarity is transitive over this succession of comparisons. However, the critical issue concerns not the end result but the recursive process of attaining it. At each step of the process, the environments of a tentative phoneme y are augmented by those of a new segment x. Then, before proceeding further, all the environments that formerly contained the now-included segment x must be restated in terms of the redefined phoneme. Chomsky reframes this rather obvious housekeeping step as an ad hoc procedure brought in just to save taxonomic phonemics from its flaws. 3.2 Criteria for Grouping Segments The ªadditional criteria of symmetry| are not criteria for evaluating alternative ªtentative phonemic systems| produced by analysis of the complementary distribution of -16- Draft July 29, 1993 Not For Attribution Harris the Revolutionary phones relative to one another. They are criteria for making each choice grouping segments, one by one, into partially defined phonemes (Harris 1951:62-63). Another quotation affords a convenient way to review these criteria. Chomsky argues (1964:401.2) that the ªphonetic substance| of phonemes cannot be supplanted by distributional or other criteria for grouping segments into phonemes: No procedure has been offered to show why, for example, initial [ph] should be identified with final [p] rather than final [t], in English, that does not rely essentially on the assumption that the familiar phonetic properites (Stop, Labial, etc.) are the ªnatural| ones. Harris might be interpreted as suggesting that a non-phonetic principle can replace reliance on absolute phonetic properties when he concludes (1951:66) that ªsimplicity of statement, as well as phonetic similarity, decide in favor of the p- ph grouping|; but this implication, if intended, is surely false. The correct analysis is simpler only if we utilize the familiar phonetic properties for phonetic specification. With freedom of choice of features, any arbitrary grouping may be made simpler. Turning to the indicated place in Harris (1951), we find ourselves in the midst of his discussion (section 7.4) of criteria for grouping segments into phonemes. By now, of course, we are aware that Harris's segments (identified in Chapter 5, with extraction of suprasegmental contours in Chapter 6) are not phonetic primitives, they are (inefficient) representations of phonemic distinctions. As representations, they are logical symbols (1951:8, 16&fn17, 18) serving to represent the contrasts; it is the contrasts that are the primitive elements (1951:34-35). Because they are derived from a segmentation of the speech stream, phonetic properties are associated with the representations. Some of these phonetic properties may themselves be made to serve as symbols for the contrasts in a given language or for classes of contrasts. Equivalently, various marks or symbols may be used to indicate the contrasts, and among the alternatives are marks naming phonetic properties. When we undertake the work leading to a more efficient representation for the contrasts, ªin most cases there will be more than one way of grouping segments into phonemes [+] It is therefore necessary to agree on certain criteria which will determine which of the eligible segments go together into a phoneme| (1951:63). Harris's concern at this stage of analysis is to guide the process of grouping segments so that a simpler description is possible.14 The criteria that ____________________ 14 Ironically, Chomsky makes the same point when he inveighs against what he calls an ªatomistic| criterion proposed by Trubetzkoy for deciding a problem of distribution: ª[T]he acceptability of an analysis hinges on its effect on the grammar as a whole|, etc. (1964:411.2). He then argues (1964:412) that such Draft July 29, 1993 Not For Attribution -17- Phonemic Theory he applies to this end may be thought of as evaluation metrics, with the caveat that we have just noted in the preceding section: one should not think of some sort of `distributional analysis component' churning out a set of candidate tentative phonemicizations, followed by an `evaluation component' that applies the criteria and selects a winner as the phonemic system. ____________________________________________________________ [g]eneral systematic considerations are, however, foreign to the point of view of taxonomic phonemics, and, in fact, they have often been criticized as circular (cf. e.g., Twaddell, 1935, 66). This criticism is correct, given the general ªprocedural| bias of modern phonology; but it shows only that the attempt to develop a taxonomic phonemics on the basis of analytic procedures of segmentation and classification, supplemented by such ad hoc rules as [Trubetzkoy's, as cited], is ill conceived from the start. Chomsky surely was aware that Harris's overriding concern, second only to preserving the data of contrast, is the overall simplicity of the description. It is unlikely that Harris failed ever to discuss this concern with Chomsky; indeed, it is plausible to suppose that Chomsky learned something of this emphasis from Harris. -18- Draft July 29, 1993 Not For Attribution Harris the Revolutionary In the main, the range of evaluation criteria15 by which Harris proposes to guide distributional analysis (section 7.3) to a more efficient representation for the phonemic distinctions (contrasts) are stated in terms of symmetry in the representation of sounds (section 7.42) and symmetry of environment (section 7.43), but in each case Harris clearly and unequivocally states that the motivation for the criterion is not merely symmetry for its own sake but rather the simplifications that symmetry makes possible in the descriptive statements (rules) of the grammar. One metacriterion (condition for applying the other criteria) concerning the simplicity of the description is as follows: M1. Number and freedom: as few phonemes as possible with maximum freedom of occurrence among the phonemes (section 7.41). A second metacriterion is as follows: M2. The effect on the whole phonemic system [should] be always kept in mind: ªIn all cases of associating segments + the final decision rests with the way the grouping in question affects the whole stock of phonemes| (1951:71). As we shall see, many of Chomsky's examples turn out to be straw men because they do not meet this metacriterion.16 ____________________ 15 Harris did not view the criteria that he proposed as absolute or universal, and he accepts non-uniqueness at every level of linguistic description, that is, that there is no unique ªcorrect| solution even for a given purpose and a given set of evaluation measures that might appropriately be associated with that purpose. The terms of evaluation depend in part on what is important to the investigator. Harris's interest is in finding a most efficient representation, for reasons related to information theory. He readily allows that other motivations would lead one to devise other kinds of tests by which to evaluate alternative representations for the contrasts. An example is given in the appendix to 10.1-5, distinctive features. For another example, Bloch was interested in work with dialect variation, for which more phonetic detail is required than for Harris's purposes. 16 This continual reference to the pattern of the whole system is characteristic of Harris's work at every level of analysis. Isolated examples are merely anecdotal evidence, and have no import unless contextualized (explicitly or by the reader) within a description of the whole pattern of which they exemplify but a fragment. It is important not to overlook the context of the whole description against which Harris always sets his examples. Also important are the temporal and logical relations among the forms of the data at various stages of analysis (a central methodological concern of Harris 1951). The case at hand, to which we will return presently, provides a trivial example: Harris's discussion of a [p]-[ph] grouping as against a [p]-[t] grouping, following a criterion of parallel allophonic Draft July 29, 1993 Not For Attribution -19- Phonemic Theory The three criteria that hinge on phonetic considerations (ªsymmetry in the representation of sounds|) are as follows: 1. ªIdentity of representation among segments,| that is, their having phonetic features in common (section 7.421). This is ªconvenient| because it makes for a simpler phonological description to be able to ªspeak of the phoneme as representing this common feature, rather than as being a class of segments. Relations between phonemes would then represent relations between sound features.|17 2. ªIdentity of inter-segmental relation among phonemes,| that is, having the same phonetic difference between parallel pairs of allophones (section 7.422). ªIt is also convenient to have the relation among segment definitions within one phoneme identical with the relation in other phonemes.| This makes for great simplification in the phonological description, because a single statement may account for the parallel allophony of multiple phonemes. 3. Choosing phonetic features that apply across the ªcomplete phoneme stock| (section 7.423). ªIf the objective is a minimal stock of phonemes, the definition of each of which is to be as simple as our other criteria permit, it follows that the selection of the common features should be governed by the generality of these features and differences among all the segmental elements of the language.| This is ªconvenient| because it broadens the set of phonemes to which a statement (rule) affecting a given feature can potentially be made to apply, simplifying the grammar. ªWe can discover which groupings [of segments] will yield the most simply defined phonemes by testing the differentiation, upon which we propose to assign particular segments, throughout all the segments.|18 ____________________________________________________________ differences, cannot legitimately be quoted out of context of his immediately prior discussion of grouping by phonetic similarity. 17 This is not to say that the sound features would be the primitive elements. The contrasts are the primitive elements, as always, and the sound features would in this treatment be their representations. 18 Harris notes the correspondence of such features to those proposed by Trubetzkoy, Jakobson, and linguists of the Prague Circle, and refers us to chapter 10, fn. 48. This is in an appendix in which he discusses distinctive features at greater length, as supplanting the segmental phonemes. -20- Draft July 29, 1993 Not For Attribution Harris the Revolutionary Harris presents two other kinds of criteria which are ignored in Chomsky's (1964) argument. These criteria are also to be applied so as to increase the generality of statements (rules) and simplify the grammar. The first is the criterion of environmental symmetry (section 7.43). This is not mere formalism for its own sake: In all cases of associating segments on the basis of environmental symmetry, as in associating them by phonetic symmetry, the final decision rests with the way the grouping in question affects the whole stock of phonemes. Assigning a segment, in some environment, to a particular phoneme not only affects the membership and environmental range of that phoneme, and its similarity in these respects to other phonemes, but also prevents any other phonemes from having that segment in that environment. The other criterion that Chomsky neglects to mention is the criterion of morphemic identity (appendix to 7.4) with the obvious benefit of simplifying morphophonemics. In a nod to those who are concerned about ªseparation of levels|, Harris observes that this criterion is not necessary for phonemics, though helpful, and that assignments based on this criterion are necessarily tentative at this stage, based on ªguesses| as to the shapes and boundaries of morphemes, subject to revision when the morphological analysis is carried out in full. He notes the obvious corollary that if on the other hand phonemic analysis is done first (ªin order of rigorous analysis, not of time|, p. 196) the phonemes are subject to adjustment later anyway, during the course of morphemic analysis. Because he had the data of contrast as his scientific bedrock, mixing levels was for Harris merely a matter of preference for the ªorder of rigorous analysis|, and no bugbear. Let us now examine the context from which Chomsky extracted the quotation, ªsimplicity of statement, as well as phonetic similarity, decide in favor of the p-ph grouping| as opposed to a possible grouping of [t] and [ph] into a single phoneme. This occurs in Harris (1951:66), section 7.422, criterion 2 (parallel allophonic differences). Harris says: This [criterion] requires that the segments be grouped into phonemes in such a way that several phonemes have correspondingly differing allophones (i.e. segment members) in corresponding environments. E.g. English [p, t, k] all occur in /s___V/, as in stone; [ph, th, kh] all occur in #___V/ as in tone. We could have grouped [p] and [th] together, since they are complementary. But the above criterion directs us (barring other relevant relations) to group [p] with [ph] into /p/, and similarly for /t/, /k/. For if we do so, we can say that the /#___V/ member of all these phonemes is virtually identical with the /s___/ member except that [h] is added; such a simple general Draft July 29, 1993 Not For Attribution -21- Phonemic Theory statement would not have been possible if we had grouped the segments differently. At the close of this passage occurs the footnote from which Chomsky has quoted: Symmetrical statements can often be made for several alternative arrangments of segments. For instance we can group [p] with [th], [t] with [kh], [k] with [ph] and say that the [#___V] member of each phoneme involves aspiration plus a shift of the point of closure one place back or two places forward (ªplace| being defined in terms of the tongue-palate contact positions recognized for the other phonemes). However, simplicity of statement, as well as phonetic similarity, decide in favor of the [p]-[ph] grouping. In context, then, we can see that Chomsky's proposed interpretation distorts Harris's obvious intent. ªSimplicity of statement| in the quoted sentence means that it is much simpler to talk of the group of sounds having an added [h] in the #___V environment than it is to talk of them having this same [h] plus a shift of place of articulation. In other words, Chomsky's major point of contention, that simplification of the grammar should be the overriding criterion for evaluating alternative analyses, is precisely what Harris had been advocating, and is indeed what Harris means by the disputed phrase ªsimplicity of statement|. The phrase ªas well as phonetic similarity| does not mean ªthat a non-phonetic principle can replace reliance on absolute phonetic properties|, as Chomsky contends, but on the contrary refers to the fact that the case in hand would already have been decided by the immediately preceding criterion of 7.421, namely, that segments should preferably have a phonetic feature in common. At the place in the disputed text where this quotation occurs Harris has himself already made the point that a phonetic-feature criterion is not only relevant but that it is (of course) logically and methodologically prior to the criterion of phonetic parallelism (7.422) then under discussion. 3.3 Complementarity and Biuniqueness Chomsky characterizes complementary distribution (1964:414) as a procedure whose ªgoal is to provide the minimally redundant representation meeting the conditions of biuniqueness and local determinacy.| Minimal redundancy is indeed a primary aim of Harris's procedures, but a set of elements that correspond one-one to contrasts between utterances is not necessarily minimally redundant. Many potential systems of representation might bear a biunique relation to the contrasts between utterances, including an initial segmentation determined (non-uniquely) by tests for repetition under substitution. All such representations are `phonemic' in the essential sense that the contrasts are phonemic distinctions. Most of -22- Draft July 29, 1993 Not For Attribution Harris the Revolutionary these possible systems of representing the contrasts have more elements than are needed, and elements that are more restricted than necessary in their combinations with one another. Harris himself said (1951:62-63) that complementary distribution alone is not a sufficient criterion to guide the recurseive process of grouping segments into phonemes to a preferred or optimal representation, and as we have seen he deployed a range of criteria for grouping segments so as to represent the phonemic contrasts more efficiently, that is, so as to simplify the grammar (ªin most cases there will be more than one way of grouping segments into phonemes [+] It is therefore necessary to agree on certain criteria which will determine which of the eligible segments go together into a phoneme| (1951:63)). It is difficult, therefore, to motivate Chomsky's demand that complementary distribution, reduced to a mechanical discovery procedure, should alone ªprovide the minimally redundant representation|. This is nonetheless the only possible construal to put on Chomsky's ensuing discussion of complementary distribution, as he turns (1964:414) to an example of phonemic overlapping due to Bloch. In the dialect that Bloch describes, alveolar flap [D] occurs intervocalically after stress in e.g. ªBetty|, and after q in e.g. ªberry|. Chomsky says: The requirement of biuniqueness is preserved if we set up the phonemes /t/, with the allophone [D] in intervocalic, post-stress position, and /r/, with the allophone [D] after dental spirants. Given a phone in a phonetic context, we can now uniquely assign it to a phoneme; and given a phoneme in a phonemic context, we can uniquely determine its phonetic realization (up to free variation). However, this solution, which is the only reasonable one [+] is inconsistent with the principle of complementary distribution. In fact, the allophones [D] and [r] of /r/ are not in complementary distribution since they both occur in the context [be-y] (`Betty', `berry'). Hence complementary distribution is not a necessary condition for biuniqueness. The sleight-of-hand here is in the phrase ªthe allophones [D] and [r] of /r/ + both occur in the context be-y]|. The [D] that occurs medially after stress in `Betty' is of course not an allophone of /r/. Chomsky's argument here goes through only if one insists that all occurrences of the phone [D] be assigned to one and only one phoneme, either /r/ or /t/. Without notice, Chomsky is assuming the strong form of the invariance condition prohibiting even partial phonemic overlapping. Underlying the invariance condition, especially clearly in its strong form, is the familiar presumption that biuniqueness is a relation between phones (e.g. [D]) and phonemes (e.g. either /r/ or /t/). But Harris showed that the relation of biuniqueness instead properly holds between distinctions (e.g. `Betty' vs. `berry' vs. `throw') and various means of representing them. Draft July 29, 1993 Not For Attribution -23- Phonemic Theory See e.g. Harris (1951:34-35) on segments representing distinctions, ªthe representation of speech as a sequence or arrangement of unit elements is intimately connected with the setting up of phonemic distinctions between each pair of non-equivalent utterances,| etc., and the discussion of contrast in section 1, above. The relation between contrasts and representations in this example may be considered as follows: Word Contrasts: "Betty" "berry" "throw" Representation #1: [beDiy] [beriy] [qDow] Representation #2 [betiy] [beriy] [qrow] The segments in [beDiy], [beriy], [qDow] constitute a less efficient representation of the contrasts between words in English (including these three words) than do the segments in [betiy], [beriy], [qrow], but both representations preserve a biunique relation to the contrasts. Chomsky is either misrepresenting Harris, or he has not understood him. 3.4 Neither Necessary Nor Sufficient Chomsky continues (414-415): Furthermore, the class of ªtentative phonemic systems| as defined in the preceding paragraph will not include the optimal biunique system as a member, so that no supplementary criteria will suffice to select it from this class. This passage betrays the linear, non-recursive conception that Chomsky has of distributional analysis, in which he sees evaluation criteria as selecting one out of a set of phonemic systems output by a kind of complementary distribution module. In this case, to be sure, the evaluation criteria are not those that apply at each step of distributional analysis. These ªsupplemental procedures| are described in Harris (1951) chapter 8, Junctures, and chapter 9, Rephonemicization. Even so, the ªsupplemental procedures| by which problems of overlapping and neutralization such as these are resolved do so not by ªselecting| one of the tentative solutions given with distributional analysis and Harris's criteria of Chapter 7, but by generating new solutions not otherwise available (although in some cases they may confirm ªguesses| as to the results of morphological analysis). Crucially for this example, Chomsky ignores Harris's description of dividing the distributional range of a segment. We divide the [D] segment into two distributionally defined (but phonetically identical) elements, and assign the one that occurs after [q] to /r/ and the one that occurs intervocalically after stress to /t/. This procedure is a further extension of distributional analysis, yielding a new system that had -24- Draft July 29, 1993 Not For Attribution Harris the Revolutionary previously not been considered. Harris describes this procedure and its justification as follows (1951:91-92): The phonemic representation of a language may be simplified by means of this operation [ªdividing the segment|] when the segment A cannot be put into any phoneme without disturbing the over-all symmetry, and when it is possible to partition A into such segments A1 and A2 as would fit well into the phonemes of the language. Assignment of both A1 and A2 to some other phonemes should yield a more symmetrical or otherwise convenient phonemic stock than assigning the original A to some phoneme. (The last sentence of this passage is a criterion, not a prediction.) As we have seen, contrast is a primitive observation of the science, and the goal of determining or defining contrast was met at the outset. At this stage, it is an inappropriate goal. This is why complementary distribution is not necessary for determining contrast, and indeed no criteria are necessary other than those of the pair test. Complementary distribution of segments is, however, necessary for the segments to be grouped into a phoneme. Only under a condition of phonetic invariance, however, could this be construed as complementary distribution of phones. Under such a condition, the phone [D] could not have been split into two distributionally defined segments, one in one phoneme and one in another. The thrust of Chomsky's argument against complementary distribution is that complementary distribution is not sufficient to determine contrast. This is explicit in his next example (415.1): But now observe further that the class of tentative phonemic systems, as defined, will contain systems that fail the principle of biuniqueness. Thus, for example, [k] and [a] are in complementary distribution in English (and, furthermore, share features shared by nothing else, e.g., in Jakobson's terms, the features Compact, Grave, Lax, Non-Flat). Hence they qualify as a tentative phoneme, and there is a tentative phonemic system in which they are identified as members of the same phoneme /K/.19 ____________________ 19 I have omitted the circumflex over the vowel [a], as it has no bearing whatsoever on the argument, and as Chomsky in fact omits it in other examples involving the identical sound in Bloch's dialect. The intent may have been to encourage us to think that just one phonetic variant of vowel a is to be grouped with k, but the only phonetic specification given (ªCompact, Grave, Lax, Non-Flat|) makes no such differentiation. It is amusing that Chomsky had to use Jakobson-Fant-Halle distinctive features to set up this straw man, since such counterintuitive groupings had actually been criticised as a weakness of the theory. I am not aware of any other phonetic specifications actually in Draft July 29, 1993 Not For Attribution -25- Phonemic Theory He goes on with examples of word pairs, e.g. ªsocked| and ªScot|, that would have an identical representation under this proposal, in violation of biuniqueness. From this he concludes that ªthe principle of complementary distribution does not even provide a sufficient condition for biuniqueness.| But this is scarcely a discovery, since Harris himself motivates his criteria for grouping segments with the observation that complementary distribution by itself is not sufficient (1951:62-63): The operation of 7.2-3 [i.e. complementary distribution] determines whether segment X can be associated with segment Y in a single phoneme. But it is not sufficiently selective to determine which of two complementaries, X and Z, shall be included with Y (if X and Z are not mutually complementary, so that only one of them, but not both, can be associated with Y). [+] It is therefore necessary to agree on certain criteria which will determine which of the eligible segments go together into a phoneme. Further, Chomsky's way of stating the issue reverses the priorities. Biuniqueness is a one-one relation between contrasts (utterance-utterance distinctions) and a representation of those contrasts. It is a requirement or condition only in the tautological sense that anything that does not preserve this relation is ipso facto not a representation of the contrasts. The logical relation of complementarity guarantees that if you group elements whose distribution is complementary into a new element, the new element necessarily retains the biunique relation that held between the original elements and the utterance-utterance distinctions. It is a logical operation that preserves biuniqueness in a way analogous to the manner in which operations in mathematical logic preserve truth value. There is a perhaps more obvious problem with Chomsky's example of a complementary grouping of [k] and [a] into /K/, however, and that is that it would never be pursued very far, even by a mechanical discovery procedure, because it does not generalize. A tentative phonemicization that groups [k] with [a] exploits the complementarity of consonants with vowels.20 But having exploited that ____________________________________________________________ use by linguists as criteria for phonemicization that would suggest this sort of grouping. For example, in more recent versions of the universal alphabet of features, [k] is [+high, -low] and [a] is [-high, +low]. 20 These are distributional classes that would be set up in all tentative distributional solutions for a given language if they were set up for any of them. See (1951:61fn7), where Harris introduces the convention of using V and C in environmental statements: ªV indicates any segment of a group which we call vowel. In most languages it is convenient to set up this group, on distributional grounds, in contrast to consonants (C).| There are of course well known acoustic and articulatory grounds supporting these -26- Draft July 29, 1993 Not For Attribution Harris the Revolutionary complementarity for the [k]-[a] pair, it is no longer available for all C-V pairs as a class. Major distributional regularities would be lost in favor of a smaller and more restricted grouping that adds to the complexity of the description. Descriptive statements (rules) could no longer apply to the class of vowels, or of consonants, or of stops, and so on. Thus, it is scarcely surprising that this socalled ªproblem has received little attention| (Chomsky 1964:415.1). Every step of this purported argument having fallen apart, Chomsky's conclusion falls too (415.1): Since it provides neither a necessary nor a sufficient condition for biuniqueness, and, apparently, has no motivation except for its connection with biuniqueness, the principle of complementary distribution appears to be devoid of any theoretical significance. 3.5 Adjusting Environments in the Course of Phonemicization Chomsky next takes up Harris's discussion (7.31, p. 62) of the need to redefine environments during the course of distributional analysis, so that when segments being tested against environments are merged into a single segment they are also merged where they occur in the statements of the environments before proceding with further analysis. We briefly discussed this recursive property of distributional analysis earlier. In a footnote (1951:62fn10, substituting small caps for subscript [|]), Harris points out the consequences of overlooking what is after all an obvious requisite for carrying out the work of linguistic analysis in a systematic and logical way: If we did not do this, but had included [R] and [r] in one phoneme /r/ ([R] after [T], [r] after [k]) and [T] and [k] in one phoneme /T/ ([T] before [R], [k] before [r]), we would have try and cry both phonemically written /Tray/. This would conflict with a basic consideration of phonemics, namely, to write differently any two utterances which are different in segments [so that the writing preserves the biunique relation to the primitive distinctions that the segments had + BN]. This inadmissible situation does not arise if we group [R] and [r] into /r/ while keeping [t] and [k] phonemically distinct from each other, since they contrast before the new /r/. As we have seen (3.1, Tentative Phonemes), this is essentially a housekeeping step that must be taken each time ____________________________________________________________ classes, but, as always, behind Harris's use of the word ªconvenient| is the overriding criterion of simplifying the grammar, and Chomsky's straw-man proposal fails that test immediately, as soon as we examine distributional alternatives that might fit together with the [k]-[a] grouping. Draft July 29, 1993 Not For Attribution -27- Phonemic Theory a segment is combined with an existing phoneme-in-process (under the criteria identified in Chapter 7). However, Chomsky reframes it as an ad hoc procedure brought in to save taxonomic phonemics from its flaws. He proposes a ªtentative phonemic system| (sic, see caveat earlier) in which (415.3): + we could have a phoneme /T/ with allophones [T] before [R] and [k] before [r], and a phoneme /R/ with allophones [R], [r]. But now both `try' and `cry' would be represented /TRay/. To avoid this, Harris suggests that we first group [R] and [r] into /r/, and then redefine distributions in terms of the newly specified contexts, in which [T] and [k] now contrast before /r/. This procedure will avoid the difficulty in the particular case of `try', `cry', but not in the cases described above. The ªcases described above| to which he refers are the pseudo-problems with pairs like ªsocked| vs. ªScot| that we have just seen evaporate upon examination. 3.6 Ou Tout Se Tient There is a broader methodological issue lurking behind this discussion, concerning the handling of data in relation to emerging results. Chomsky's attack on distributional analysis continues: Furthermore, the same procedure could just as well be used to group [t] and [k] into /T/, thus keeping [R] and [r] phonemically distinct (in further justification, we could point out that this regularizes distributions, since now /t/ occurs neither before /r/ or /l/, instead of, assymetrically, only before /r/). Hence, as in the case of the procedures discussed above, it fails to distinguish permissible from impermissible applications. The same difficulty faces this pseudo-proposal as did Chomsky's earlier suggestion that [k] and [a] should be grouped in one phoneme. The analysis of distributional patterning cannot be done atomistically by treating one isolated example at a time, but rather, as Sapir taught and exemplified, only by holding the emerging structure of the whole always before one. Furthermore, what Chomsky touts as a gain in symmetry by increasing the restrictions on /t/, in fact goes against Harris's stated aim of eliminating as many restrictions as possible. Again, it is not clear whether Chomsky has misunderstood Harris or is misrepresenting him. -28- Draft July 29, 1993 Not For Attribution Harris the Revolutionary Having diminished a broadly applied housekeeping principle to the status of an ad hoc ªprocedure| aimed at rescuing distributional analysis from awkward counterexamples like the grouping of [t] and [k] into /T/, Chomsky argues that such a procedure violates a global requirement that Harris must retain in order to avoid use of rule ordering. Continuing at (415.3): Finally, the procedure [sic] as stated is inconsistent with Harris's general requirement on the set of linguistic procedures (1951:7), namely, that operations must be ªcarried out for all the elements simultaneously| without any ªarbitrary point of departure.| Turning to Harris (1951:7), we find the following: In both the phonologic and the morphologic analyses the linguist first faces the problem of setting up relevant elements. To be relevant, these elements must be set up on a distributional basis: x and y are included in the same element A if the distribution of x relative to the other elements B, C, etc., is in some sense the same as the distribution of y. Since this assumes that the other elements B, C, etc., are recognized at the time when the definition of A is being determined, this operation can be carried out without some arbitrary point of departure only if it is carried out for all the elements simultaneously. The elements are thus determined relatively to each other, and on the basis of the distributional relations among them. One could wish that Chomsky had read this passage more carefully. It does not express a requirement imposed by Harris, but rather a requirement imposed by the character of the material being worked with, for there is no a priori basis for identifying the elements of language other than relative to one another, no metalanguage external to and prior to language itself.21 Harris continues (1951:7): It is a matter of prime importance that these elements be defined relatively to the other elements and to the interrelations among all of them.22 The ____________________ 21 For discussion, see Ryckman 1986, Harris 1991, Nevin 1993. In a footnote to this passage, Harris (1951) discusses how there is no need for recourse to meanings expressed in some purportedly antecedent metalanguage (really, only glosses in the linguist's native language). Of course, it is transparent that the `meanings' contributed by glosses are distributional in character when the glosses happen to be in the same language as that being studied, as they are for dictionary definitions. The same considerations apply to `semantic features' in a supposedly separate and antecedent metalanguage. 22 A footnote here refers us to Sapir for ªthe most explicit statement of the relative and patterned character of the Draft July 29, 1993 Not For Attribution -29- Phonemic Theory linguist does not impose any absolute scale upon a language, so as to set up as elements, for example, the shortest sounds, or the most frequent sounds, or those having particular articulatory or acoustic properties [emphasis added]. Rather, + he sets up a group of elements (each by comparison with the others) in such a way as will enable him most simply to associate each bit of talking with some construction composed of his elements. When Harris talks of the elements of language being relatively defined, and not absolutely, he is referring as Sapir did to the patterning and configuration that characterizes and indeed constitutes language. Chomsky has shrunk this global perspective down to an ad hoc demand placed upon mechanical discovery procedures. It is precisely the need to consider all the elements together in a systematic whole that precludes absurdities such as Chomsky's proposed merger of [k] with [t] or with [a], as we have seen. Recall that the elements being defined are logical symbols with which phonetic properties are associated (1951:8, 16&fn17, 18, discussed above in 3.3, Criteria for Grouping Segments). The issue cannot be predetermined by the phonetic properties associated with these elements (although phonetic likeness is a desirable optional criterion), precisely because of cases, such as the one cited above, in which we wish to divide a phonetically defined element like [D] into two distributionally defined elements, [D] after [q] vs. intervocalic [D], so as to yield by their subsequent inclusion into more general elements a systemically simpler and more useful result. Chomsky's rather curious suggestion that Harris wishes to avoid rule ordering is in his next sentence at the cited place (1964:415): In fact, this requirement [that linguistic patterning be dealt with as a whole, that elements be defined relative to one another without any arbitrary starting point defined in absolute terms] was what made it possible for Harris to avoid Bloomfield's use of descriptive order [that is, rule ordering]. But it is violated by the procedure just discussed. We will take up the question of rule ordering in section 4, noting here only that it would be absurd to suggest that Harris abjured Bloomfield's use of ordered rules. The relevant passages in Harris (1951) are in the Appendix to 14.32 (237-238), where an ªexact statement of the representation of the morphophonemes| is an alternative way of stating an example from Bloomfield's account of Menomini, and in the Appendix to 16.21 (283), where the use of descriptive order is given as the alternative. They are presented as alternatives, with prejudice to neither. Harris (307-308fn14) also suggests that descriptive order is ____________________________________________________________ phonologic elements|. Harris also notes Saussure and Trubetzkoy. -30- Draft July 29, 1993 Not For Attribution Harris the Revolutionary essentially connected with his overarching goal of simplicity, in discussing the selection of a morpheme alternant as the base form for morphophonemic descriptions: ªThe criteria for selecting a basic alternant are not meaning or tradition, but descriptive order, i.e. resultant simplicity of description in deriving the other forms from the base.|23 3.7 Chomsky's ªCondition C| Chomsky closes his survey of ªtaxonomic phonemics| in general and of complementary distribution in particular with the following generalization, labelled ªcondition C| at the end of his 4.3, p. 416): C. if phone sequences X and Y contrast, then their phonemic representations must differ. Condition C is followed by the claim that ªthere are no known distributional procedures for defining phonemes that guarantee that this condition will be met, and, in particular, the principle of complementary distribution fails in actual cases.| These statements nicely crystallize the misconceptions (or misrepresentations) of Harris's work that pervade this discussion. Chomsky's Condition C states the case in terms of phone sequences that contrast. This amounts to substituting Chomsky's ªuniversal phonetic alphabet| in place of Harris's segmentation based upon substitution preserving contrast/repetition. Harris's initial segments are defined relative to one another, but Chomsky assumes the initial segments are ªphones,| elements defined in phonetic terms. Harris of course uses terms of phonetic theory as descriptors of phonetic properties associated with the segments, but the phonetic properties do not determine the segmentation or the linguistic relevance of the segments. Harris's segmental representations of the contrasting utterances always necessarily differ, no matter whether they count in Chomsky's reckoning as ªphone sequences,| as ªtentative phonemes,| or as phonemic representations. They always differ because Harris's representations of utterances are representations of contrasts between utterances, and therefore necessarily have a biunique correspondence to the ____________________ 23 This shows that the issue of Bloomfield's ªdescriptive order| is broader than simply the ordering of rules. In section 15.31 (246-248), descriptive order means setting up first those classes of morphemes that are most easily treated, then setting up other classes in relation to them. Chomsky might as easily have argued that this violates Harris's ªrequirement| to have no arbitrary starting point. But this is of course another instance of ªbootstrapping| the description. The initial, most easily defined classes are subject to later redefinition in the light of results with the other classes, all of them being adjusted reciprocally as parts of a single system. Draft July 29, 1993 Not For Attribution -31- Phonemic Theory contrasts between utterances. Chomsky's Condition C is met from the outset. The satisfaction of Condition C is preserved under distributional redefinition of the phonemic elements, not created by it. The procedures of distributional analysis ensure that each subsequent redefinition, refinement, and rephonemicization of the representation preserves the biunique correspondence that the prior one held with respect to the contrasts. Biuniqueness is a transitive relation, back to the initial representation, such that each new representation of utterances X and Y a fortiori has a one- one correspondence to the primitive contrasts between utterances X and Y. As a consequence, ªif phone sequences X and Y contrast|, then the representations of X necessarily differ from the representations of Y under all these redefinitions. The criteria of simplicity and symmetry (in the several senses of 7.421, 7.422, 7.423, and 7.43) help to determine one of the solutions that best meet the overarching criterion of simplicity or efficiency.24 4. The Argument from Rule Ordering Surrounding the discussion of ªtaxonomic phonemics| is a presentation of Halle's argument (Chomsky, 1964:412-413) ªthat it is generally impossible to provide a level of representation meeting the biuniqueness condition without destroying the generality of rules, when the sound system has an assymetry|. To demonstrate this, derivations of different forms are laid out in parallel, showing a segmental representation after each application of rules. Rule Stop Affricate 1. 2. For each segmental representation in these derivational series that we might propose as the phonemic ªlevel of representation|, it is shown that some rule applies before that level as a morphophonemic rule to some forms (i.e. in one or more of the parallel columns), and that the same rule after that level as an allophonic rule for other forms. In Halle's now familiar example, voicing is contrastive for stops in Russian, but not for the affricate. (This is the ªassymetry| in Chomsky's statement.) A rule that voices obstruents before voiced obstruents is morphophonemic for stops, but allophonic for affricate. The key is to realize that it is the distinctive features that are the representations of contrasts. Harris had observed that his simultaneous components in general ____________________ 24 Chomsky's extended discussion of ªseparation of levels| in the ensuing pages avowedly does not apply to Harris, and I will not concern myself with it here, though there are many problematic issues in it, particularly in Chomsky's treatment of Bloomfield and Sapir. -32- Draft July 29, 1993 Not For Attribution Harris the Revolutionary (1944a:205, 1951:133.3) and distinctive features in particular (unit-length components defined for the whole stock of phonemic contrasts of a language, 1951:147.2) may supplant the segmental phonemes. Chomsky proposes to eliminate the phonemic ªlevel| of representation: i. physical phonetics ii. systematic phonetics (distinctive features) iii. taxonomic phonemics (segments, eliminated) iv. systematic phonemics (morphophonemic) In fact, the representation of phonemic contrasts is simply shifted from segmental phonemes to distinctive features. What Chomsky calls ªsystematic phonetics|, using distinctive features, is no less phonemic than the segmental representation that it supplants. The effect of the actual changes in representation of sound systems is as follows: i. phonetics ii. contrasts: segmental representation + feature representation iii. base representation (morphophonemic) If you state your rules in terms of distinctive features, then your phonemic representation is distributed throughout the successive stages of a derivation + in fact, everywhere that the distinctive features are used. The question ªwhere are the phonemes| is sensible only if one sees the task of phonology as defining contrast; when contrast is a primitive element given by the pair test, this is a non-issue. Neither the features nor the segments have any privileged ontological status. It is the contrasts that are ªreal|. This broaches the peculiar ontological status of language. In the familiar Berkeleyan example, if a tree falls in a forest and no one is there to perceive it, we nonetheless assume the reality of the tree and of the event, and, insofar as sound is defined in physical terms as pressure waves in the atmosphere, there is a sound. However, if a tape recording of the Gettysburgh Address is played in the forest and no one is there, there is only sound. There can be no language present+no words, no contrasts or distinctive features+in the absence of a hearer who controls perceptions of the words, contrasts, etc. of English. The contrasts may be represented by phonemic segments, phonemic components, distinctive features (unit-length phonemic components), or in some other way, but they cannot be defined absolutely, in purely physical terms. They can only be defined relatively, in terms of the contrasts between utterances that they represent. And the contrasts are not physically given, they are socially given. This is why the pair test is necessary for determining the phonemic distinctions or contrasts in a language. The purpose here has been to to set the record straight and to explicate Harris's insights into the nature of linguistic contrast. The consequences of reinstating these insights appropriately in the theory and practice of Draft July 29, 1993 Not For Attribution -33- Phonemic Theory linguistics today go beyond what is possible to consider in this context. -34- Draft July 29, 1993 Not For Attribution Harris the Revolutionary 7. References Anderson, Stephen R. 1974. The Organization of Phonology. San Diego: Academic Press. __________. 1985. Phonology in the Twentieth Century: Theories of rules and theories of representations. Chicago: Univ. of Chicago Press. Bloomfield, Leonard. 1926. A set of postulates for the science of language. Language 2.153-164. Reprinted in IJAL 15.195-202 (1949) and in Joos (1957:26-31). __________. 1933. Language. New York: Holt, Rinehart & Winston. Bloch, Bernard. 1941. Phonemic overlapping. American Speech 16.278-284. Reprinted in Joos (1957:93-96). __________. 1948. A set of postulates for phonemic analysis. Language 22.200-248. __________. 1953. Contrast. Language 29.59-61. Bruner, Jerome Seymour. 1983. Child's Talk: Learning to use language. With the assistance of Rita Watson. New York: W. W. Norton. Chomsky, Noam A. 1957. Review of Fundamentals of Language, by R. Jakobson and M. Halle. IJAL 23.234-242. Reprinted in Makkai (1972:) __________. 1964. Current Issues in Linguistic Theory. The Hague: Mouton. Chapter 4 reprinted in Makkai (1972:401-423). __________. 1971. ªOnward and Upward With the Arts: John is easy to please| The New Yorker, May 8, 1971: 44-87. Interview with Noam Chomsky by Ved Mehta. Chomsky, Noam A., & Morris Halle. 1965. Some controversial questions in phonological theory. Journal of Linguistics 1.97-138. Reprinted in Makkai (1972:457- 485). __________. 1968. The Sound Pattern of English. New York: Harper & Row. Gibson, James J. 1982. Reasons for Realism: Selected essays of J. J. Gibson. Hillsdale, N.J.: Lawrence Erlbaum Associates. Halle, Morris. 1954. The strategy of phonemics. Word 10.197-209. Reprinted in Makkai (1972:). __________. 1959. The Sound Pattern of Russian: A linguistic and acoustical investigation. With an excursus on the contextual variants of the Russian vowels by Lawrence G. Jones. The Hague: Mouton. Harris, Zellig S. 1941. Review of Trubetzkoy's Grundzge der Phonologie. Lg. 17.4:345-9. __________1944a. Simultaneous components in phonology. Language 20.181-205. Reprinted in Joos (1957:124-138). __________. 1944b. Yokuts structure and Newman's grammar. IJAL 10.4:196-211. __________1951 [1947]. Methods in Structural Linguistics. Chicago: Univ. of Chicago Press. Reissued as Structural Linguistics. Draft July 29, 1993 Not For Attribution -35- Phonemic Theory __________. 1988. Language and Information. Number 28, Bampton Lectures in America Delivered at Columbia University. New York: Columbia Univ. Press. __________. 1991. A Theory of Language and Information: A mathematical approach. Oxford: Clarendon Press. Harris, Zellig S., Michael Gottfried, Thomas Ryckman, Paul Mattick, Jr., Anne Daladier, Tzvee N. Harris, and Suzanna Harris. 1989. The Form of Information in Science: Analysis of an immunology sublanguage. Preface by Hilary Putnam. Boston Studies in the Philosophy of Science 104. Boston: Kluwer Academic Publishers. Hyman, Larry M. 1975. Phonology: Theory and analysis. New York: Holt, Rinehart & Winston. Hymes, Dell, & John Fought. 197*. American Structuralism. The Hague: Mouton. Joos, Martin. 1942. A phonological dilemma in Canadian English. Language 18.141-144. __________. 1957. Readings in Linguistics. Washington: ACLS. Reissued as Readings in Linguistics I. Chicago: Univ. of Chicago Press. Liberman, Alvin M., and Mattingly, Ignatius G. 1983. The motor theory of speech revised. Cognition 21.1-36. Lisker, Leigh. 1978. Rapid vs. rabid: A catalogue of acoustic features that may cue the distinction. Haskins Laboratories Status Report on Speech Research, SR-54, 127-132. Makkai, Valerie Becker. 1972. Phonological Theory: Evolution and current practice. Lake Bluff, IL: Jupiter Press. McGurk, H. & J. MacDonald. 1976. Hearing lips and seeing voices. Nature 264.746-748. Nevin, Bruce E. 1991. Obsolescence in Achumawi: Why Uldall Too? Papers from the American Indian Languages Conferences, held at the University of California,Santa Cruz, July and August 1991, Occasional Papers on Linguistics 16:97-126. Carbondale, IL: Southern Illinois University Department of Linguistics. __________. 1993. A minimalist program for linguistics: a perspective on the work of Zellig Harris. Historiographia Linguistica XX:2/3 (1993). Powers, William T. 1973. Behavior: The control of perception. Chicago: Aldine. __________. 1989. Living Control Systems: Selected papers of William T. Powers. Gravel Switch, KY: CSG Book Publishing. __________. 1992. Living Control Systems II: Selected papers of William T. Powers. Gravel Switch, KY: CSG Book Publishing. Robertson, Richard J. & William T. Powers. 1990. Introduction to Modern Psychology: The control theory view. Gravel Switch, KY: CSG Book Publishing. -36- Draft July 29, 1993 Not For Attribution Harris the Revolutionary Ryckman, Thomas Alan. 1986. Grammar and Information: An investigation in linguistic metatheory. PhD. dissertation, Columbia Univ. Sapir, Edward. 1929. ªThe Status of Linguistics as a Science.| Language 5.207-214. (Repr. in Sapir 1949:160166. [Page references are to the reprint.]) Schane, Sanford. 1974. Generative Phonology. Englewood Cliffs, N.J.: Prentice-Hall. Stevens, Kenneth N. 1972. Quantal nature of speech. In E. E. David, Jr., and P. B. Denes (eds.), Human Communication: A unified view. New York: McGraw-Hill. Stevens, Kenneth N., and Sheila E. Blumstein. 1988. Speech Physiology, Speech Perception, and Acoustic Phonetics. Cambridge: Cambridge Univ. Press. Twaddell, William Freeman. 1935. On Defining the Phoneme. Language monograph 16. Baltimore: LSA. Reprinted in Joos (1957:55-80). Werker, J. F. & R. C. Tees. 1984. Cross-language speech perception: Evidence for perceptual organization during the first year of life. Infant Behavior & Development 7.49-63. Williams, Gregory. 1992. A Bibliography of Hierarchical Perceptual Control Theory. Gravel Switch, Kentucky: CSG Publications. Draft July 29, 1993 Not For Attribution -37-