Volume 27, Issue 3

With the language constellation of Norwegian/German, a methodology is developed to analyse features of linguistic interference within a given material (e.g. a continuous reading sample). Every particular text segment is given a pattern of binary traits, defined in terms of the present problem, so as to point out and to specify the points of interference. These patterns of traits are checked by means of a listening test, i.e. in a communicative situation. As shown by the statistical treatment of the results, the following conclusions are drawn. No particular trait alone does give a sufficient prediction of interference, whereby such effects are indicated by certain selective combinations of characteristics. This concept is further confirmed by analyses of speaker reliability and correlative connections of the traits.

Typological considerations both of phonemic structure and of phonemic change call for a restatement of the system of primitive Indo-European stops. The two series so far taken to be voiced and voiceless are reinterpreted as glottalized and voiceless aspirate, respectively. The languages believed to have undergone major consonant shifts (Germanic, Armenian, Hittite) turn out to be very conservative as far as their consonant systems are concerned.

Generative phonology and, to some extent, generative transformational linguistics in general, is criticized for its extreme process model, intuitively unacceptable underlying abstractions, its lack of synchrony and exaggerated praise for one’s own achievements coupled with overstated criticism of other orientations. To the author’s mind, what is good in the generative transformational theory is not new and what is new is not good.