Developments in linguistic theory have shed important light upon the theory and practice of translation, resulting in the recognition that translating is basically not a process of matching surface forms by rules of correspondence, but rather a more complex procedure involving analysis, transfer, and restructuring. Such linguistic procedures as transformation and componential analysis provide far more satisfactory bases for translation than have existed in the past. At the same time, the theory of translation is able to provide linguistic science with new insights into structure and with improved methods for testing hypotheses.

The chief views requiring the Assumption’ of proto-language uniformity are evaluated. The conclusion is reached that the ‘assumption’ can usefully be replaced by (1) a hypothesis regarding the relation of the reconstructed phonemic system to that of some occurrent idiolect, and (2) an assumption regarding the interrelevance of non-irreconcilable reconstructions. In the course of developing the argument, the article touches on the law of the regularity of phonetic change, and proposes a definition of the comparative method.

Various proposals have been advanced concerning what has been referred to as ‘refutation of the nil hypothesis’ in language comparisons, i.e. that the sound-meaning correspondences found for any two particular languages are within the range of what one would expect by sheer coincidence. Swadesh concludes that at 95% confidence level, four or more cognates in a list of 100 basic vocabulary items justify accepting a hypothesis of genetic relationship. H. K. J. Cowan estimates that the existence of three or more cognates is sufficient at the same level. Both these proposals are based on mathematical formulations, Cowan’s the more sophisticated. Approaching the problem from the empirical side, 100-item lists in 21 languages are here used to arrive at a conclusion that the presence of more than two solid CVC sound-meaning correspondences in languages believed to be unrelated raises a strong probability that more than chance is involved.

IE *kugh- violates the rules for the consonant structure of Indo-European roots; the development of IE *kugh-ti- to Gmc. *hug-di-, doubtful for other reasons as well, must therefore be rejected. But from the Germanic root *hug- we would expect the *-ti-abstract *huh-ti-; thus *hug-di-, the form actually functioning as abstract to *hug-, must be an analogical formation. *-hug~di-z is built on the stem of the preterite and past participle of *hug-, which in Germanic followed the pattern of the ē-verbs.

Medieval Spanish had a complex system of strident obstruents. Slightly divergent paths of evolution of these segments have resulted in the major phonological differences between modern Castilian and Latin American dialects. These widely studied sound changes are here re-interpreted as changes in the sets of phonological rules of Castilian and American dialects. The role of the theories of markedness and linking, especially with respect to the features [distributed] and [strident], is discussed in some detail, and seems to provide insight into puzzling problems. In passing, it is seen that, contrary to popular belief, there has been no merger of [θ] and [s] in Latin American dialects as [s]; rather, Castilian diverged from other dialects in replacing [s] with [θ].

Three different reflexes of Proto-Slavic *g and the reflex of *di̭ changed to fricatives in early Slavic. A hitherto unnoticed sequential constraint in Common Slavic, along with the geographical limitations of these changes, permits us to date them relative to the loss of syllable-final obstruents in CS1., to the fall of the jers, and to one another. Typological analysis shows that the source of the phonetic changes was a pronunciation rule implying phonemically tense vs. lax obstruents in CS1. The relative chronology established is significant for several problems of chronology within Slavic and for its relation to Baltic; and the set of concentric isoglosses defined throws light on the pattern of expansion of early Slavic.

The thesis of this paper is that a ‘mirror image’ convention should be incorporated into linguistic metatheory. This convention may be interpreted as a notational device; it serves to collapse rules that are significantly similar in that they are mirror images of one another, in a sense that can be made fairly precise. Part I gives a preliminary formulation of the convention and discusses syntactic evidence; Part II, to be published in Lg. 45:4, will deal with lexical and phonological evidence.

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.