Zellig S. Harris

Distributional Structure

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 distributional structure and independently of it.

From Phoneme to Morpheme

The following investigation1 presents a constructional procedure segmenting an utterance in a way which correlates well with word and morpheme boundaries. The procedure requires a large set of utterances, elicited in a certain manner from an informant (or found in a very large corpus); and it requires that all the utterances be written in the same phonemic representation, determined without reference to morphemes. It then investigates a particular distributional relation among the phonemes in the utterances thus collected; and on the basis of this relation among the phonemes, it indicates particular points of segmentation within one utterance at a time. For example, in the utterance /hiyzkwikǝr/ He’s quicker it will indicate segmentation at the points marked by dots: /hiy. z. kwik. Ər/; and it will do so purely by comparing this phonemic sequence with the phonemic sequences of other utterances.

Co-Occurrence and Transformation in Linguistic Structure

This paper defines a formal relation among sentences, by virtue of which one sentence structure may be called a transform of another sentence structure (e. g. the active and the passive, or in a different way question and answer). The relation is based on comparing the individual co-occurrences of morphemes. By investigating the individual co-occurrences (§ 1.2; § 2) we can characterize the distribution of certain classes which may not be definable in ordinary linguistic terms (e. g. pronouns, § 2.6). More important, we can then proceed to define transformation (§ 1.3), based on two structures having the same set of individual co-occurrences. This relation yields unique analyses of certain structures and distinctions which could not be analyzed in ordinary linguistic terms (§ 3). It replaces a large part of the complexities of constituent analysis and sentence structure, at the cost of adding a level to grammatical analysis. It also has various analytic and practical applications (§ 5.7), and can enter into a more algebraic analysis of language structure (§ 5.2, 4, 6) than is natural for the usual classificatory linguistics. A list of English transformations is given in § 4. The main argument can be followed in § 1.11 (Co-Occurrence Defined), § 1.2 (Constructional Status), § 1.3 (Transformation Defined), § 2.9 (Summary of Constructions), § 3.9 (Summary of Sentence Sequences), §5 (The Place of Transformations in Linguistic Structure).1

From Morpheme to Utterance

This paper presents a formalized procedure for describing utterances directly in terms of sequences of morphemes rather than of single morphemes.1 It thus covers an important part of what is usually included under syntax. When applied in a particular language, the procedure yields a compact statement of what sequences of morphemes occur in the language, i. e. a formula for each utterance (sentence) structure in the language.

Discourse Analysis: A Sample Text

This paper offers an example of how connected discourse can be formally analyzed in such a way as to reveal something of its structure. The method used here was described in a previous paper, ‘Discourse Analysis’, Lg. 28 (1952), 1–30. It consists essentially of the following steps: given a particular text, we collect those linguistic elements (morphemes or sequences of morphemes) which have identical environments within a sentence, and we call these equivalent to each other; thus, if we find the sentences A F and B Fin our text, we write A=B and say that A is equivalent to B or that both are in the same equivalence class. We further collect those linguistic elements which have equivalent (rather than identical) environments, and we call these also equivalent to each other; if we find the sentences A F and B E, and if A=B (because B F occurs too), then F is secondarily equivalent to E, and we write F=E. (Note that in the sentence A F, A is the environment of F, and Fis the environment of A.) This operation enables us to collect many or all of the linguistic elements or sections of any particular text into a few equivalence classes. For example, if our text consists of the sentences1 A F: B E: C G: B F: M E: A G: N E: N G: M H, we set up two classes: one class to include A, B (because of A F and B F), C (because of A G and C G), M, and N (because of B E and M E and N E); the other class to include F, E (because of B F and B E), G (because of A F and A G), and H (because of M E and M H).2

Morpheme Boundaries within Words: Report on a Computer Test

For the science of linguistics we seek objective and formally describable operations with which to analyze language. The phonemes of a language can be determined by means of an explicit behavioral test (the pair test, involving two speakers of the language) and distributional simplifications, i. e. the defining of symbols which express the way in which the outcomes of that test occur in respect to each other in sentences of the language. The syntax, and most of the morphology, of a language is discovered by seeing how the morphemes occur in respect to each other in sentences. As a bridge between these two sets of methods we need a test for determining what are the morphemes of a language, or at least a test that would tentatively segment a phonemic sequence (as a sentence) into morphemes, leaving it for a distributional criterion to decide which of these tentative segments are to be accepted as morphemes.

The structure of science information

The organization of information within science can be investigated in a principled way through analysis of science language. The restricted use of language in science enables description of the informational structure of science and of particular subfields, with strong similarities to structures in mathematics and programming languages. This result rests on decades of research into the relation between form and content in language, based on an information-theoretic approach to the structure of information. Examples are provided from immunology and the social sciences. Practical applications include storage of science information in databases, indexing the literature, and identification and resolution of controversy.

Simultaneous Components in Phonology

This paper1 investigates the results that may be obtained when phonemes, or utterances in general, are broken down into simultaneously occurring components: as when the English phoneme /b/ is said to consist of voicing plus lip position plus stop closure, all occurring simultaneously.2

Morpheme Alternants in Linguistic Analysis

The purpose of this paper is to suggest a technique for determining the morphemes of a language, as rigorous as the method used now for finding its phonemes. The proposed technique differs only in details of arrangement from the methods used by linguists today. However, these small differences suffice to simplify the arrangement of grammars.

Linguistic Transformations for Information Retrieval

This paper discusses the application to information retrieval of a particular relation in linguistic structure, called transformations.1 The method makes possible the reduction of a text, in particular scientific texts, to a sequence of kernel sentences and transformations, which is roughly equivalent in information to the original text. It seems possible to determine the division into kernels in such a way that each adjusted kernel will carry about as much information as is likely to be called for independently of the neighboring information in the article. A text may therefore be stored in this form (perhaps omitting, by means of formal criteria, any sections which are unnecessary for retrieval), and its individual kernels may be retrieved separately. Since the carrying out of transformations depends only on the positions of words in a sentence, and not on knowledge of meanings, it seems possible that at least part of this operation can be performed by machine; the more so since the method does not require any judgment about the subject matter, or any coding of the concepts of a particular science.