ACKNOWLEDGMENTS* In addition to my indebtedness to the work of Roman Jakobson, to be mentioned in the notes, I would also like to thank Fred Householder and Charles F. Hockett for making helpful critical comments on the earlier version of this paper.
1. I am indebted to the work of Roman Jakobson for directing my attention to the importance of implicational universals. Notes
2. See, for example, the remarks of R. B. Braithwaite, Scientific Explanation (Cambridge, 1953), concerning scientific laws. "The one thing upon which everyone agrees is that it always includes a generalization, i. e., a proposition asserting a universal connection between properties" (p. 9).
3. That is, empirically, not logically, implied. All languages are observed to have the characteristics in question. It should be added that universals in the sense of nondefinitional characteristics, if found only in language, do have the additional logical property of implying as well as being implied by the definitional properties.
4. Some of the ideas regarding the basic order typology are found in nineteenth-century linguistic literature. For example, the relation between genitive position and prepositions versus postpositions and the hypothesis that some languages favor the order modifier-modified and others the opposite order is already a familiar notion in R. Lepsius' introduction to his Nubische Grammatik (Berlin, 1880).
The most systematic treatment is that of W. Schmidt in Die Sprachfamilien und Sprachenkreise der Erde (Heidelberg, 1926) and in several other works. Schmidt's basic conclusions may be summarized here. Prepositions go with nominative-genitive order and postpositions with the reverse order. The nominative-genitive order tends to appear with verb before nominal object and genitive-nominative with object-verb. Schmidt says nothing of subject-verb order so that types I and II as treated in this paper are not distinguished. Further, nominative-genitive is associated with noun-adjective and genitive-nominative with adjective-noun. This last correlation, particularly the latter half, is much weaker than the others. Schmidt gives figures based on a world sample which show good general agreement with the results from the thirty-language sample utilized here. It should be added that Schmidt's chief interest in this topic is as a vehicle for the interpretation of culture history. His results there verge on the fantastic.5. Siuslaw and Coos, which are Penutian languages of Oregon, and Coeur d'Alene, a Salishan language, are exceptions.
6. The manner in which each language has been assigned can be determined from the data of Appendix I.
7. For details, see Appendix II.
8. Iraqw, a southern Cushitic language, Khamti, a Thai language, standard Persian, and Amharic.
9. The single case where it does not hold seems to be Amharic, which has SOV GN, and AN, but is prepositional.
10. However, Householder informs me that in Azerbaijani, and in most types of spoken Turkish, it is allowable to have one modifier, especially a dative or locative noun phrase after the verb.
11. Languages of type I—Berber, Hebrew, Maori, Masai, and Welsh; II—Thai, Yoruba; III—Burmese, Burushaski, Japanese, Kannada, Nubian. For Yoruba, see further note 12.
12. In the following languages the affix or particle follows: II—Finnish, Guarani, Malay, Maya, Serbian; III—Basque, Burmese, Japanese, Kannada, Nubian, Turkish, Quechua. It precedes in Yoruba, but may be accompanied by a final particle.
13. The question word is first in Berber, Finnish, Fulani, Greek, Guarani, Hebrew, Italian, Malay, Maori, Masai, Maya, Norwegian, Serbian, Welsh, Yoruba, and Zapotec.
14. Again, this only holds for literary Turkish, according to Householder. See note 10.
15. Auxiliary precedes verb in Finnish, Greek, Italian, Masai, Maya, Norwegian, Serbian, Swahili, Welsh, Zapotec. Auxiliary follows verb in Basque, Burushaski, Chibcha, Guarani, Hindi, Kannada, Nubian, Quechua, Turkish.
16. For details, see Appendix I.
17. Languages with adjective-noun and adverb-adjective order are Burushaski, Finnish, Greek, Hindi, Japanese, Kannada, Maya, Norwegian, Quechua, Serbian, Turkish. Languages with noun-adjective and adverb-adjective order are Basque, Burmese, Chibcha, Italian, Loritja. Languages with noun-adjective and adjective-adverb order are Fulani, Guarani, Hebrew, Malay, Swahili, Thai, Yoruba, and Zapotec. Languages with noun-adjective and the rule that certain adverbs precede and certain follow the adjective are Maori and Welsh. Berber, Masai, Masai,Nubian, and Songhai—no data.
18. Languages with adjective-marker-standard are Berber, Fulani, Greek, Hebrew, Italian, Malay, Maori, Norwegian, Serbian, Songhai, Swahili, Thai, Welsh, Zapotec. Languages with standardmarker-adjective are Basque, Burmese, Burushaski, Chibcha, Guarani, Hindi, Japanese, Kannada, Nubian, Turkish. Both constructions are found in Finnish.
19. Languages with common noun-proper noun are Greek, Guarani, Italian, Malay, Serbian, Swahili, Thai, Welsh, Zapotec. Those with proper noun-common noun are Basque, Burmese, Burushaski, Finnish, Japanese, Norwegian, Nubian, and Turkish.
20. The relational expression precedes the noun in Basque, Burmese, Burushaski, Chibcha, Japanese, Kannada, Turkish. The noun precedes the relational expression in Berber, Fulani, Greek, Guarani, Hebrew, Hindi, Italian, Malay, Maori, Masai, Maya, Norwegian, Quechua, Serbian, Songhai, Swahili, Thai, Welsh, Yoruba, Zapotec. Both orders are found in Finnish and Nubian. In Finnish the construction with the relational expression preceding the noun is in imitation of literary Swedish (personal communication of Robert Austerlitz).
21. The exclusively suffixing languages are Basque, Burmese, Chibcha, Finnish, Hindi, Japanese, Kannada, Loritja, Nubian, Quechua, Songhai, Turkish.
22. The reason for specifying order is that there are instances of neutralization of number agreement in which the order of the item is not involved. For example, in classical Greek the neuter plural goes with a singular verb without regard to order.
Additional note: The following facts were learned too late to be included in the paper. According to information supplied by Einar Haugen, Norwegian has both genitive orders. Note that Norwegian had been the only exception in the sample to the generalization on p. 64. In a discussion at the International Congress of Linguistics at Cambridge in August 1962, it was pointed out that Papago, a Uto-Aztecan language, is I/Po. This is therefore an exception to Universal 3. From Mason's data it should probably be assigned to type 7 of Appendix II.
*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. 104-107.