This work contains a discussion of redundancy rules and the role they play in the phonological component of a generative grammar. Phonological redundancy rules were first given a clear theoretical foundation in the morpheme structure rules used by Halle (1959) to predict redundant phonological information in morphemes. We will be concerned in what follows with reformulating and extending this original theory of morpheme structure rules. We will give particular attention to the formal nature of morpheme structure rules and to the use of blanks in representing redundancies, and will give a solution to a long-standing problem in this area. In spite of this improvement in morpheme structure rule theory, we will find other motivations for introducing a new formal device, morpheme structure conditions, to replace morpheme structure rules. We will attempt to show that, both in the types of statements which can be made about the structure of morphemes and in the formal nature of these statements, morpheme structure conditions are to be preferred over morpheme structure rules.

A method is described which allows us to plot, easily and rapidly, the positional relations among linguistic elements which have fixed order relative to each other (such as affixes). The relationship of this method to connectivity matrices is dicussed, as well as the relation to Nida's and Hays's methods of positional analysis.

The doubling of intervocalic glides in Proto-Germanic has been attributed to different causes: preceding accent, following accent, and various laryngeals. Rather than introducing factors which have left no traces in the historical languages, it is preferable to take into account the inherited ablaut system and its productivity. The starting point of the Germanic innovation consists in the replacement of the prevocalic zero-grade -u-, -i- by -uṷ-, -ii̭-, a development which has parallels in other languages. The relation between the zero-grade -uṷ-, -ii̭-and the full-grade -euṷ-, -eii̭- is analogous to the relation between -uT- and -euT-, -iT- and -eiT- (T = obstruent). Hence strong verbs with roots in -eu, -ei are to be considered as the chief source of forms containing a geminated glide.

OE brecÞa in Beowulf 171 is taken by editors and commentators as either nominative singular of a masculine n-stem or nominative plural of a feminine ō-stem. Syntactic considerations suggest and morphologic observations prove that it is a masculine n-stem. OE -brecÞ occurs only in composition and is a different formation.

The interchange of / and v (u) in German spelling is found in the oldest written records of the Old High German period. The reason for the introduction of v (u) as a symbol for /f/ in Old High German lay in the Latin spelling practices of the Merovingian period, and in the Vulgar Latin features which they reflected.

A re-analysis of English syntax is proposed, in which the forms have and be in their use as ‘main verbs’ are eliminated from the base and reintroduced by transformational rules. The rules are shown to be minor additions to rules required on independent grounds. Syntactic arguments are given first, then answers to possible semantic objections. It is further suggested that the use of these forms as auxiliaries can be explained in the same way. The resulting base rules are shown to be somewhat closer to a reasonable set of universal base rules than the corresponding rules of earlier analyses. The paper concludes with a discussion of analogous constructions in a number of different languages.

The establishment of abstract underlying forms, along with the recognition of certain basic rules such as contraction of successive identical segments, apocope, etc., greatly simplifies the formation of Spanish plurals, as well as accounting for divers other phenomena such as stress position and word-final assibilation.

Several linguists have contended that categorical perception is a central or even a defining property of language. The statements of this ‘polarity principle’ for language usually imply descriptions of two kinds of behaviors: detecting that two speech sounds are different, and identifying each of them. This article spells out the implied descriptions, then presents evidence that they are valid. It also shows, however, how categorical perception can be obtained for various non-speech continua, including hue, loudness, and azimuth. Thus, it casts doubt upon a popular theory that attributes categorical perception of speech to articulation. Instead, the article concludes that the description of language provided by the polarity principle is, under certain conditions, equally valid as a description of any behavior, human or infra-human—in other words, that the polarity principle is merely the formulation in the domain of language of some very general laws concerning the behavior of organisms.

In certain spoken Tibetan compounds, there is a medial p which does not occur with either member of these compounds in isolation. This raises several questions which we attempt to answer in this article: (1) Are there descriptive grounds for assigning this p to one of the member morphemes? (2) Written Tibetan represents to some degree a much earlier stage of Tibetan. Does Written Tibetan support the reconstruction of morphemes in p? (3) Is there evidence outside Tibetan to support such reconstructions?