1. Introduction and summary. A phrase-structure grammar has been written which generates exactly the set of sentences generated by a fairly large transformational grammar written by Noam Chomsky. The phrase structure version of Chomsky's grammar is included in the appendix. It is written in an abbreviated notation which is explained below.

The cluster *ns appears to have had a varied development in Hittite. Forms occur which indicate that *n was lost before s, while in others the outcome was almost certainly [nts] (written nz). No one has yet succeeded in satisfactorily determining the conditions of the distinction or distinctions which apparently must be made. In its entirety the problem is quite complex, for analogy is likely to have played a part in producing the attested results, and the number of otherwise reliable etymologies is so small that the forms may be divided on more than one basis depending on which of several conceivable instances of analogy one prefers to accept. However, one group of forms containing *ns about which it is apparently possible to make a specific statement at present is that in which the combination arose secondarily through derivation within Hittite.

In his article ‘Problems of Armenian phonology III’, Lg. 38.254–62 (1962), Werner Winter treats the metathesis of IE consonant clusters in Old Armenian. To his example for the development of initial IE *tr-, Arm. erek' 'three' : Lat. trēs, I should like to add Arm. ert'am ‘go’ : Lat. in-trā-re ‘go in, enter’, trā-ns ‘across’; Skt. tr-ya-te ‘rescue’, cf. tára-ti ‘pass’; Gk. trā-ns ‘piercing’; from IE *trā-. Morphologically, erVam parallels other Arm. -am verbs, as gam ‘come’ < *- : Lat. vā-dō ‘go’, OHG. wa-tan ‘wade’ and perhaps ham ‘stand’ < * : Skt. aor. á-gā-t ‘went’.

1. The most distinctive feature of the verbal system of a Slavic language is undoubtedly its verbal aspect. Stated simply, a particular Slavic verb has either perfective (P) aspect, which mean that a commitment as to completion has been added to the verb's basic meaning, or it has imperfective (I) aspect, in which is no such implication. To express the meaning ‘to write’ in Russian, a choice has to be made between the two aspectual partners pisat' I 'to write, to be in the process of writing' and napisat' P to write, to get written, to complete the writing (of)'. There are also morphological implications to aspect, in that in certain constructs the selection of a verb form of a particular aspect is obligatory; for example, the periphrastic future of Russian permits only the use of an I verb, as in ja budu pisat' 'I will write, I will be writing' (*ja budu napisat' does not occur). Exceptions to this typical aspectual pairing, such as Russian obeščat' 'to promise', which functions as either an I or a P verb, testify by their very oddity to the pervasiveness of the aspectual dichotomy.

At the outset I make an assumption: if it can be shown that there is a dependence between the function of language as a means of communication and the regularity of phonetic change, then we are permitted to construct a hypothesis that this function of language explains the regularity of phonetic change and not the reverse. It would be more reasonable to say, if it is true, that the use of language as a means of communication forces regularity on phonetic change than to say that the regularity of phonetic change facilitates communication between speakers of a language.

1. Certain conclusions about the nature of linguistic change and genetic relationship are implied in the work over the past fifteen years in the area now known as lexicostatistics. This paper attempts a critical examination of some of these conclusions.

The availability of electronic computers for text analysis makes it possible to consider logical and statistical operations on a scale far beyond what has been practical before. This paper will propose a form of statistical semantic analysis applicable to either a written or a spoken corpus, and, with suitable modifications, to conversation.

1. Introductory. Tf we regard language activity, in its formal aspect, as the repeated selection of one from among a number of possible events—or, so as to formulate it more simply, of items from sets of items which are “at risk” under certain conditions, the total selection situation (items and conditions) varies along a cline. At one end of the scale, there are a very large number of sets, in very complex interrelations, but with very few items in each: this is grammar. At the other end, there are few sets, in simple relations, but with large inventories of items making them up: this is lexis. There is no sharp line between them, though the extremes are clearly distinct. Since patterning of the first type lends itself to much greater abstraction, we can make the theory of grammar much more powerful than the theory of lexis. So the linguist tries to bring as much as possible of linguistic form into his grammatical statements—though there is a theory of lexis too, which has a part in the total description.'