102
by-word basis only because membership in declensions is not always predict-
able.)
The thesis of complete synchronic unpredictability of ablaut clashes with
the results of traditional grammar; furthermore, it does not take into account
the existence of generalisations that were covered by the rule systems in early
generative grammar (Ross 1967; Wurzel 1970). To be sure, recent treatments
still recognise the existence of (sub-)regularities; an example is: if the past is in
o, then the past participle is in
o, too (Wunderlich & Fabri 1995: 254, note 16;
256). But more often than not such observations are relegated to footnotes or
ignored altogether. The reason may be that known regularities apparently do
not ‘make sense’: from a synchronic point of view they appear to be arbitrary.
However, the present paper is intended to show that the apparently so
confusingly multiple gradations of German strong verbs are but manifestations
of a rather uncomplicated system; as I will show, ablaut in German is charac-
terised by strict regularities and a simple form-function relation. In what fol-
lows I shall discuss first (in Section 2) the form of ablaut (What is ablaut?).
The rest of the investigation is concerned with the function of ablaut (What is
ablaut for?) and moreover the form-function-relation
(How does ablaut
work?). The rest of the present section serves to address questions of concep-
tual background and terminology.
1.3 Morphological models
The notions used so far have been taken from the traditional word-and-
paradigm-approach (to use the division of ‘models’ of grammatical description
well-known from Hockett 1954), for which the distinction between words and
forms of words is pivotal. Within such an approach it is the formation of forms
of verbs that provides the dominant perspective on ablaut (as far as morphol-
ogy is concerned). Consequently, gradation classes are conceived of as classes
of verbs. A verb such as
s p r e c h e n w
is classified as a member of the fourth
class of strong verbs. Forms of verbs (here:
spreche, sprichst, ...) are charac-
terised in terms of morpho-syntactic classifications (with respect to person,
number, mood, tense and voice), and on this basis they are arranged into
paradigms. As an outgrowth, often ablaut alternations are recorded by listing
verb forms, namely, the so-called principal parts. Given this starting point, the
internal make-up of word forms comes into focus only in the second place.
But, of course, verb forms are dissected into stems and endings. Moreover,
complex stems are assumed and their formation is investigated. Paul (1917:
189) refers to ablaut, reduplication and suffixation as means of tense stem
formation. Nevertheless, stems are usually characterised by recourse to the
functions of word forms of which they are parts (e.g., as
present tense stem,
past tense stem), not in terms of their formal make-up.
103
On the other hand, it is precisely the occurrence of certain stems that co-
determines to which categories a verb form belongs. Consider past tense stems
(past stems, for short). From a morphological point of view, past forms of a
verb are what they are for the reason that they are built on past stems. Hence
one would want to know what it is that makes a stem a past stem (but taking
recourse to its occurrence in certain verb forms would obviously lead into a
circle).6 What, then, are the formal (expression-related) properties that distin-
guish past stems?
It was with the advent of structuralism that such questions became a spe-
cial focus of attention. The analysis of past stems that are, traditionally speak-
ing, formed by adding tense suffixes did not seem to pose problems for an
item-and-arrangement-morphology. A past stem such as lobt (of the weak verb
LOBENw)
could be analysed as combining two morphemes, lob and /, the latter
having the meaning ‘past’; hence the stem lobt is a past stem because it con-
tains a morpheme with the meaning ‘past’.
However, the problems encountered by such an approach are notorious
(cf. Robins 1959, Matthews 1970, 1972), and it was through the treatment of
internal inflection that they became particularly obvious. Past stems of strong
verbs such as band (of
BINDENW)
would not lend themselves to segmentation
easily, and various treatments were discussed as to how this could be dealt
with (see Harris 1942, Bloch 1947, Nida 1948, and, again, Hockett 1954). The
basic problem is that a form such as band simply does not contain a past tense
marker comparable to the ‘dental suffix’ in lob-t. Since band differs from the
present tense stem bind only with respect to vocalism, the stem vowel would
be the only reasonable candidate. But there are other strong verbs that have
stems in a that happen to be no past stems (but present and/or past participle
stems as in the case of
FANGENW
or
HÄNGENW),
and these verbs in their turn
have past stems (fing, hing) that show vowels which appear in present or past
participle stems of the former ones or still other verbs (cf., e.g.,
b i n d e n
w ,
present stem bind)
REITENW,
past and past participle stem ritt). Thus even if
you convince yourself that a stem such as ritt may be segmented (say as r...tt
+ /), it would still be questionable to address the vocalic ‘infix’ as a tense
marker.7
The rival approach, item-and-process, as presented by Hockett, would
conceive of markers as being constituted by differences between forms (cf.
6 Note that the problem of circularity, which has been made to stand out clearly by Lieb
(1983: 173, 178f.), is not restricted to ablaut — characterisations such as ‘the genitive
ending -(e).s' pose similar problems. For further discussion, see Section 5, infra.
7 Matthews (1970: 107) formulates this argument using vocalic suffixes of Italian nouns;
the problem is not peculiar to internal inflection.