ATA time when nearly every nook and corner of the globe has been made accessible and when international relations are becoming from year to year more general and more intimate, it seems only natural that the idea of a universal language, as a means of common intercourse for all mankind, should have been revived, so as to be hailed in many quarters with delight and enthusiasm. “Revived,” I said, for we must not imagine that the present generation is the first to embrace a similar idea. Every one of us is familiar with the story of the tower of Babel, as told in Genesis, chapter 11, beginning with the statement “And the whole earth was of one language and of one speech.” This condition was contrary to the will of the Lord, especially after the people had begun to build a city and a tower whose top was to reach unto heaven. “So the Lord scattered them abroad from thence upon the face of all the earth: and they left off to build the city. Therefore was the name of it called Babel: because the Lord did there confound the language of all the earth.”

It is a truism of diachronic linguistic theory that most historically transparent forms are explainable as compromises between sound shifts (the norm) and a wide range of analogical adjustments (the deviation from the norm). In practice, however, scholars have so far made only modest advances in their attempts at analyzing associative interferences with predictable linguistic change. Specifically, if numerous examples have been collected, often at random, of the extension or complete generalization of an inflectional or derivational morph (say, the marker of a number, a gender, or an agent), little has been ascertained, at least in the Romance domain, on the opposite phenomenon, the preservation of a close-knit paradigmatic pattern at the heavy cost of a break in the consistency of a sound correspondence. Though Sapir, and possibly others before him, eloquently outlined the problem of morphological resistance to a sound shift (Language 196–204, ed. 1921), the precise exemplification and even the technique of analysis remain to be provided on the Romance side—a side credited with controlling a particularly suitable portion of experimental data. The present article is meant to mark just one step, long overdue, in this direction.

This paper presents evidence that semantic notions—such as presupposition, speaker’s and hearer’s beliefs about the world, and previous discourse—must be taken into account in a complete treatment of the distribution of some and any in conditional, negative, and interrogative sentences. Syntactic conditions alone will not account for the fact that, in certain sentence types, the two forms occur with different meanings.

[The existence of visarga as a phonetic reality has been both accepted and denied. Evidence in the Prātiśākhyas points to the phonetic existence of several spirants written ḥ. These spirants, jihvāmūlīya and upadhmānīya, are shown to be allophones of visarga, which in turn is interpreted as a subclass of the s phoneme.]

Within the Vedic verbal system there is a series of second singular forms made by suffixing -si to the full-grade root, which differ from the present type ási ‘thou art’ in that they function primarily as imperatives; e.g. yákṣi ‘sacrifice’. This archaic formation is found frequently in the Rigveda, where approximately 150 forms occur from 23 roots (§2); in other Vedic texts -si imperatives occur only for roots which have such forms in the Rigveda and predominantly in Rigvedic mantras; e.g. maghónām ‘bring safely across the gift-giving of the patrons’ (RV 8.103.7, 9.1.3); vakṣi yákṣi ca ‘bring the gods and sacrifice’ (.5.21.6, 6.16.2, 8.102.16); see Bloomfield, Concordance sv. párṣi, .

A recently published paper by Gordon H. Fairbanks discusses a question pertinent to the glottochronologic counts of Morris Swadesh and others: On what phonological basis are words to be correlated for lexicostatistical purposes? The most desirable basis is undoubtedly that of determining cognates on the basis of phonemic correspondences. But if this is the only basis, glottochronological methods must be largely inapplicable at present to such groupings as Hokan and Penutian, where few phonemic correspondences have been established. It is apparent that Swadesh does not always limit himself to such a basis. To take an example from his more recent work, his relating of words for ‘man’ in languages of both Americas, ranging from Kwakiutlan *wism to Yamana win, can hardly be based on knowledge of a regular correspondence between Kwakiutlan *w and Yamana w. The basis must then be general phonetic similarity, rather than correspondence. We may apply the term correlate to pairs connected either as cognate or as phonetically similar. The question then remains, what degree of phonetic similarity is necessary to establish a pair or set of correlates? To judge from Swadesh's list of cognate words for ‘man’, which includes such diverse items as Lokono uadili, Caxinaua ina, and Quechua runa, the degree need not be great. The exact degree appears to be determined more by intuitition than by any objective factor.

The dichotomy between exocentric and endocentric constructions seems well established in linguistics. Its importance in syntactic analysis is obvious. In this connection it is enlightening to refer back to Bloomfield's view, in which English syntax, for example, emerges as strongly exocentric, since the more common of his two favorite sentence types calls for the exocentric actor-action construction. The characterization is further supported by the frequency of his other exocentric constructions (relation-axis, clause-subordination, phrase-subordination).

1. Setting of the problem. When in 1937 Douglas Chrétien and I published Quantitative Classification of Indo-European Languages (Lg. 13.83–105) it aroused a little flurry of attention but no follow-up, except an extension of the method by ourselves to include Hittite (Lg. 15.69–71 [1939]). For some years I had looked on our essay as a dead limb of effort, especially since my coauthor came subsequently to doubt the fruitfulness of the method on the ground that most of the values found by us lacked statistical significance.