Is phonemic procedure arbitrary? It can be, but I submit that it need not be. If one defines a phoneme as ‘one of an exhaustive list of classes of sounds in a language’1, one admits an endless variety of treatments and the choice of one or another treatment is arbitrary. The ideal of exhaustiveness is not in itself sufficient to define a unique scientific procedure: But if we also take the ideals of simplicity and self-consistency, we have the basis for a non-arbitrary method, particularly in view of the essentially systematic nature of phonetic totalities. The aim of a phonemic analysis may then be stated as: the smallest number of elements having the largest number of permutations and the most systematic relationships. To attain this ideal, it is necessary to consider always the totality2 of phenomena in the given language. A formulation as to any detail must be made on the basis of all pertinent and comparable data anywhere in the totality of the given phonetic system. In the present paper I propose to demonstrate this method in application to a selected type of phenomenon, long consonants, in different phonetic totalities.3

Despite all that has been written on the subject, it is still an open question whether the prevailing types of strong preterites of class VII in NGmc. and WGmc., as Ole. hét, hlióp, helt, fekk, lét, blét, etc., have a common origin with Gothic reduplicated preterites of the types haíháit, áiáuk, haíhald, faífāh, lailōt, haíhōp, etc., or descend from IE unreduplicated formations, in particular lengthened-grade aorists. In recent decades this problem has come to be involved with another one, namely the origin of the pt. pl. stem in classes I-V of the strong verb, above all in classes IV and V for Germanic as a whole, and probably most scholars who accept unreduplicated formations as the source of the forms in the category first mentioned do so for the forms in the latter categories also. It is not proposed in this paper to restate the arguments that have been put forward in support of this view,1 or in refutation thereof,2 but rather, tentatively assuming its general correctness, to reconstruct in chronological order the various developments which the theory seems to require if it is to be brought into detailed correspondence with the known NGmc. and WGmc. forms which it purports generally to explain.

Catalan abaltir ‘to put to sleep’ has been derived by Brüch1 from the assumed t-preterit of Gallic *adbalo ‘to perish’; but Meyer-Lübke2 regards this etymology as improbable inasmuch as we do not know that Gallic had a t-preterit. Spitzer3 has suggested the derivation of abaltir from *expavitare ‘to scare’, a notion which Meyer-Lübke pronounces phonetically and semantically unacceptable.

[1. Ir. āsaim (NIr. fāsaim) ‘grow’ : Goth, wahsjan, etc. 2. W blin ‘tired’ < IE *mlēno- or *mlīno- from the root *mel- ‘rub, grind’. 3. Ir. breo ‘flame’ : OE brīwan, brīw, etc. 4. W brig‘top, crest’ : Grk. . 5. Ir. bun, W bon ‘bottom’ from the root *bhen- ‘strike’ in Goth, banja, etc. 6. W clwyf ‘ wound, disease’ : Skt. klībá-. 7. Ir. crēcht ‘wound’ : MLG schrā, ON skrā, etc. 8. Ir. fennaim ‘flay’ : Skt. pari-vās-, etc. 9. Ir. gāba ‘danger’ : Russ, zabóta, OE cēpan. 10. W gosod ‘put’ derived from the root *st(h)ā- ‘stand’. 11. W hawdd ‘easy’, gwahodd ‘invite’ : Skt. sādhú-. 12. W hogen ‘girl’ from the root *seu-q- ‘suck’. 13. Br. miz ‘expense’ from OFr. mise. 14. Br. piz ‘stingy’ semantic borrowing from Fr. chiche. 15. On Ir. serrach ‘foal’. 16. W tor ‘tear’ : Lith. tárpas, etc. 17. W trwch ‘ broken, maimed’ : Lith. trúkti.]

[Continuation of the work published in Language 6.1-24 (1930); 9.47-71, 171-91 (1933); 10.149-201 (1934).]