1. Introduction*
The tentative nature of the conclusions set forth here should be evident to the reader. Without much more complete sampling of the world's languages, the absence of exceptions to most of the universals asserted here cannot be fully assured. As indicated by the title, attention has been concentrated largely, but by no means exclusively, on questions concerning morpheme and word order. The reason for this choice was that previous experience suggested a considerable measure of orderliness in this particular aspect of grammar. In the body of this paper a number of universals are proposed. A large proportion of these are implicational; that is, they take the form, "given x in a particular language, we always find y." When nothing further is said, it is understood that the converse, namely, "given y, we always find x," does not hold. Where the two sets of characteristics are binary, the typical distribution in a tetrachoric table is a zero as one of the four entries.1 From the point of view of scientific methodology, there is nothing to apologize for in such results, and this is so for two reasons. First, the lowest-level laws as described in manuals of scientific method take precisely this form.2 Second, what seem to be non-universals about language are in fact tacitly implicational since they are implied by the definitional characteristics of language.3 Further, to assert the definitional characteristics themselves is obviously tautologous.
It is perhaps worth while to point out that a number of universals of the second type -- that is, those implied by the definitional characteristics of language -- although not usually formally stated in this paper, are in fact involved in the notion of the general comparability of languages in the grammatical sphere which underlies the specific statements found here. For example, a whole series of universals in the usual sense are assumed in such a statement as the following: If a language has verb-subject-object as its basic word order in main declarative clauses, the dependent genitive always follows the governing noun. It is here assumed, among other things, that all languages have subject-predicateconstructions, differentiated word classes, and genitive constructions, to mention but a few. I fully realize that in identifying such phenomena in languages of differing structure, one is basically employing semantic criteria. There are very probably formal similarities which permit us to equate such phenomena in different languages. However, to have concentrated on this task, important in itself, would have, because of its arduousness, prevented me from going forward to those specific hypotheses, based on such investigation, which have empirical import and are of primary interest to the nonlinguist. Moreover, the adequacy of a cross-linguistic definition of 'noun' would, in any case, be tested by reference to its results from the viewpoint of the semantic phenomena it was designed to explicate. If, for example, a formal definition of 'noun' resulted in equating a class containing such glosses as 'boy', 'nose', and 'house' in one language with a class containing such items as 'eat', 'drink', and 'give' in a second language, such a definition would forthwith be rejected and that on semantic grounds. In fact, there was never any real doubt in the languages treated about such matters. There is every reason to believe that such judgments have a high degree of validity. If, for example, someone were to dispute the specific assignment of order type of a genitive construction given in this paper, it is quite clear on what evidence such an assignment would be accepted or rejected.
For many of the statements in this paper, a sample of the following 30 languages has been utilized: Basque, Serbian, Welsh, Norwegian, Modern Greek, Italian, Finnish (European); Yoruba, Nubian, Swahili, Fulani, Masai, Songhai, Berber (African); Turkish, Hebrew, Burushaski, Hindi, Kannada, Japanese, Thai, Burmese, Malay (Asian); Maori, Loritja (Oceanian); Maya Zapotec, Quechua, Chibcha, Guarani (American Indian).
This sample was selected largely for convenience. In general, it contains languages with which I had some previous acquaintance or for which a reasonably adequate grammar was available to me. Its biases are obvious, although an attempt was made to obtain as wide a genetic and areal coverage as possible. This sample was utilized for two chief purposes. First, it seemed likely that any statement which held for all of these 30 languages had a fair likelihood of complete or, at least, nearly complete universal validity. Second, less reliably, it serves to give some notion of the relative frequency of association of certain grammatical traits. In this respect, of course, it is not to be taken literally. On some questions I have gone well outside the sample.
The main section of the paper, which follows, is concerned with the establishment of universals on the basis of the empirical linguistic evidence. These are presented with a minimum of theoretical comment. The final section is exploratory, seeking to discover what general principles may exist from which at least some of the generalizations of the earlier sections might be deduced. For convenience of exposition, the universals scattered though the text are repeated for cross reference in Appendix III. The theoretical section is far more speculative and uncertain than the sections devoted to the universals themselves. In a certain sense we would prefer to have as few universals as possible, not as many. That is, we would like to be able to deduce them from as small a number of general principles as possible. However, the establishment of a relatively large number of empirical generalizations must, on the whole, come first. For one thing, it would be embarrassing to deduce a particular universal from what seemed like a valid general principle, only to discover that the generalization was not empirically valid.
*Joseph H. Greenberg, "Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements", In: Joseph H. Greenberg (ed.). Universals of Language. London: MIT Press, pp. 73-75.