For the purposes of the present discussion, the term structure will be used in the following non-rigorous sense: A set of phonemes or a set of data is structured in respect to some feature, to the extent that we can form in terms of that feature some organized system of statements which describes the members of the set and their interrelations (at least up to some limit of complexity). In this sense, language can be structured in respect to various independent features. And whether it is structured (to more than a trivial extent) in respect to, say, regular historical change, social intercourse, meaning, or distribution — or to what extent it is structured in any of these respects — is a matter decidable by investigation. Here we will discuss how each language can be described in terms of a distributional structure, i. e. in terms of the occurrence of parts (ultimately sounds) relative to other parts, and how this description is complete without intrusion of other features such as history or meaning. It goes without saying that other studies of language — historical, psychological, etc. — are also possible, both in relation to distri-butional structure and independently of it.


Distributional Relation Distributional Fact Distributional Structure Relative Occurrence Test Utterance 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    The investigation of historical regularity without direct regard to descriptive (synchronic) structure was the major achievement of the linguists of the late eighteen hundreds. There are incipient studies of historical-descriptive interrelations, as in H. M. Hoenigswald, ‘Sound Change and Linguistic Structure’, Lg. 22 (1946), 138–43; cf.Google Scholar
  2. A. G. Juilland, ‘A Bibliography of Diachronic Phonemics’, Word 9 (1953), 198–208. The independent study of descriptive structure was clarified largely by Ferdinand de Saussure’s Cours de linguistique générale, the Prague Circle in its Travaux du Cercle linguistique de Prague, Edward Sapir in various writings, and Leonard Bloomfield’s Language.Google Scholar
  3. 2.
    These approaches are discussed by Martin Joos, ‘Description of Language Design’, Journal of the Acoustical Society of America 22 (1950), 702–8, and W. F. Twaddell, ibid. 24 (1952), 607-11.CrossRefGoogle Scholar
  4. 3.
    For a discussion of simplicity in this connection, see a forthcoming article by Noam Chomsky,’ some Comments on Simplicity and the Form of Grammars’.Google Scholar
  5. 4.
    Y. R. Chao, ‘The Non-Uniqueness of Phonemic Solutions of Phonetic Systems’, Bulletin of the Institute of History and Philology, Academia Sinica 4 (1934), 363–98. Cf. the two solutions of Annamese phonemes in M. B. Emeneau, Studies in Vietnamese (Annamese) Grammar; 9-22.Google Scholar
  6. 5.
    This kind of formulation is best expressed in the work of Sapir and Newman; cf. reviews of Selected Writings of Edward Sapir (ed. by D. Mandelbaum) in Language 27 (1951), 289-92 (Paper XXXIII of this volume); and of Stanley Newman, Yokuts Language of California in International Journal of American Linguistics 10 (1944), 196–211 (Paper XII of this volume).CrossRefGoogle Scholar
  7. 6.
    An opposition has sometimes been claimed between real facts and mathematical manipulation of structure. This claim ignores the fact that science is (among other things) a process of indicating much data by few general statements, and that mathematical methods are often useful in achieving this. Mathematical and other methods of arranging data are not a game but essential parts of the activity of science.Google Scholar
  8. 7.
    As pointed out by Kurt Goldstein, Language and Language Disturbances, 71, 103.Google Scholar
  9. 8.
    E. g. in Edward Sapir, ‘La réalité psychologique des phonèmes’, Journal de Psychologie Normale et Pathologique 30 (1933), 247–65 (translated in David Mandelbaum, ed., Selected Writings of Edward Sapir, 46-60). (See Paper XXXIII of this volume.)Google Scholar
  10. 9.
    C. F. Hockett, review of Recherches structurales in International Journal of American Linguistics 18 (1952), 98.CrossRefGoogle Scholar
  11. 10.
    As pointed out to the writer by A. W. Holt.Google Scholar
  12. 11.
    This applies to the grammatical innovation involved in new formations; the selection of morphemes within a class is determined, not only by these “grammatical” associations but also semantically. Cf. the first paragraph of §1.1 above.Google Scholar
  13. 12.
    Here we have discussed whether the distributional structure exists in the speakers as a parallel system of habits of speaking and of productivity. This is quite different from the dubious suggestion made at various times that the categories of language determine the speakers’ categories of perception, a suggestion which may be a bit of occupational imperialism for linguistics, and which is not seriously testable as long as we have so little knowledge about people’s categories of perception. Cf. for the suggestion, Benjamin L. Whorf, ‘The Relation of Habitual Thought and Behavior to Language’, Language, Culture and Personality (Sapir Memorial Volume) (ed. by A. I. Hallowell, L. Spier, and S. Newman), 75-93; ‘Languages and Logic’, The Technology Review, 1941, 43-6; and against it, Eric H. Lenneberg, ‘Cognition in Ethnolinguistics’, Lg. 29 (1953), 463–71Google Scholar
  14. Lewis S. Feuer, ‘Sociological Aspects of the Relation Between Language and Philosophy’, Philosophy of Science 20 (1953), 85–100.CrossRefGoogle Scholar
  15. 13.
    In E. G. Schachtel’s ‘On Memory and Childhood Amnesia’, Psychiatry 10 (1947), 1–26 it is suggested that the experiences of infancy are not recallable in later life because the selection of aspects of experience and the classification of experience embodied in language, which fixes experience for recall, differs from the way events and observations are experienced (and categorized) by the infant.Google Scholar
  16. 14.
    Human Nature in the Light of Psychopathology: The William James Lectures for 1938-39, ch. 3.Google Scholar
  17. 15.
    The following analysis can be fully understood only if one checks through the actual lists of Cherokee forms. The few forms cited here are taken from William D. Reyburn, ‘Cherokee Verb Morphology II’, International Journal of American Linguistics 19 (1953), 259–73. For the analysis, see the charts and comments in Reyburn’s work and in Z. S. Harris, ‘Cherokee Skeletal Grammar’, and ‘Cherokee Grammatical Word Lists and Utterances’, in the Franz Boas Collection of the American Philosophical Society Library.CrossRefGoogle Scholar
  18. 16.
    This assumption is based on the fact that each morpheme has a different distribution (§2.36), so that same feature of environment points to the same morpheme.Google Scholar
  19. 17.
    Since new formations of these combinations do not appear, we cannot apply the productivity tests of §2.1 to discover the speakers’ morphemic recognition.Google Scholar
  20. 18.
    This particular pair was suggested to me by Y. Bar-Hillel, who however considers that distributional correlates of meaning differences cannot be established.Google Scholar
  21. 19.
    It should be clear that only after we discover what kinds of distributional regularities there are among successive elements or sections in discourses can we attempt any organized semantic interpretation of the successions discovered. Various types of discourses have various types of succession (of sentences, clauses, or other intervals). In mathematics and the constructed ‘languages’ of logic, certain conditions are imposed on what sentences can appear in succession in their connected discourses (proofs): each sentence (line in a proof) has to be a theorem or else derived from a preceding sentence in a particular way. This situation does not hold for natural languages, where the truth-value of logic is not kept constant through successive sentences, and where the types of succession are more varied.Google Scholar
  22. 20.
    Such relations as that of active to passive, or buy to sell, are essentially substitutability relations (§3.4), i. e. they show that certain elements have similar environments (e. g. partially inverted ones). The fact that they may appear in neighboring sentences is a serial relation (§3.3) which is a secondary characteristic of certain substitutabilities. Relations like that of active to passive are different from the essentially serial relations of successive intervals of a discourse, discussed at the end of §3.5.Google Scholar
  23. 21.
    The fact that a discourse contains several or many occurrences of a given substitution class, often in parallel positions, brings out a rare relation in linguistics: the order of occurrence of various members of the same class. Something like this comes up in compound nouns, or in successions of two or more adjectives (sometimes with preferred order). Usually, if two members of a class occur in one domain, their order is not regular (e. g. in most cases of N and N); but in compound nouns, for instance, certain members are frequent in the first N position, and others in the second.Google Scholar
  24. 22.
    There will be a few exceptions where the passive is not obtainable. And if we try to elicit the active on the basis of the passive, we run into the difficulty of distinguishing between by of the passive (The letter was finished by Carl) and by as preposition (The letter was finished by noon).Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 1970

Authors and Affiliations

  • Zellig S. Harris
    • 1
  1. 1.University of PennsylvaniaUSA

Personalised recommendations