117
other changes of quality type, I—^-alternation shows greater variation, ablaut
forms being o-forms or «-forms; the remaining types show greater regularity,
ablaut forms invariably conforming to the default rule given above. Obviously,
among changes of quality type, I—>U-altemation must be assigned a special
status. In fact, given well-known criteria of markedness (Greenberg 1966),
these observations indicate that I—»U-altemation must be considered the
unmarked change of quality type.
With regard to quantitative ablaut, there are two types to be distin-
guished, viz. shortening, which provides forms of normalised length, on the
one hand, and lengthening, which provides forms of marked length, on the
other. The junction between change of quality type and change of quantity
type turns out to be straightforward. The unmarked change of quality type
(I—^-alternation) is coupled with shortening; otherwise change of quality
type is coupled with lengthening.25 Remarkably, there is not a single stem that
shows 1—»U-alternation combined with lengthening, nor a single stem that
shows another change of quality type combined with shortening. We find it
confirmed that I—»U-alternation is opposed to all of the remaining changes of
quality type. As it is the unmarked type of qualitative ablaut and combines
with normalisation of length, it may be referred to as
simple ablaut.
Of course, lengthening as well as shortening can apply only if a stem be-
longs to the class of stems that are subject to quantitative ablaut at all. If a
stem from this class shows I—>U-alternation. then the ablaut form is invariably
short. Unless the base is short anyway, there will be a change of quantity type.
E.g., the stems FLIESSL (long base, /fli:s/) and SPRECHL (short base, /jpre5/)
show I-^U-alternation; as the vowels are followed by voiceless fricatives,
these stems are subject to quantitative ablaut; in this case, either both base and
ablaut form are short (/fpre^/—>/JprD9/) or a long base alternates with a short
ablaut form (/flits/—>/fbs/). In summary, ablaut forms due to simple ablaut are
short if possible.26
25 This holds also of the four stems in sonorants that are subject to quantitative ablaut, viz.
n e h m
l ,
k o m m
l, FALLl , HALTl . E.g., NEHM
l
with a long base and I—»U-ablaut shows
shortening (
nehm—>nomm),
falll
with a short base and A—»I-ablaut shows lengthening
(/a//—►//«/). As noted above, there is one stem (
biet
1) that is exceptionally exempted from
quantitative ablaut. BlEr1" shows I—>U-alternation
(biet—*bot) but no shortening, which
would be expected on account of the post-vocalic voiceless obstruent (cf. Grimm 1870:
903). Incidentally, the preference for short vowels is generally rather weak in the case of/t/
as compared to other voiceless consonants (d’Alquen 1979: 196, with references). On
SIED
l
see note 23, supra.
26 «//'-alternation is coupled with shortening, too. Hence
t ret l
and
n e h ml
show /e:/—>/i/-
alternation as they are from the class of stems that are subject to quantitative ablaut.
Long—►short-alternation with
ge bl
(Curme 1922: 253) or even LES
l
is now considered
non-standard; see the pronouncing dictionaries and cf. also Paul (1917: 229): colloquial
118
All of the remaining changes of quality type (I—>A, A->I, A—>U, U—>1,
U—>A) trigger lengthening, i.e., ablaut forms are long if possible. If a stem that
is subject to quantitative ablaut shows one of these alternations, then the
ablaut form concerned is invariably long. Unless the base is long anyway,
there will be a change of quantity type. Hence all of the following ablaut forms
are long: sprach (from sprech by I—»A-altemation), briet (from brat by A—>1-
altemation), schuf (from schaff by A—^-alternation), rief (from ruf by U—>1-
altemation), and kam (from komm by U—»A-alternation); the corresponding
bases are partly short, partly long.
As a result, stem vowels of ablaut forms due to these alternations are, as
a rule, so-called point vowels, that is, vowels of extreme qualities, located, as
it were, at the points of the vowel triangle (thus maximally differentiated
vowels in Martinet’s sense).27 These are /i:/, /u:/ and the a-vowels, /a:/ and /a/
(which, in Standard German, are differentiated only in terms of length, Kohler
1995: 170). This regularity overrides even strong phonotactic preferences (or
aversions, cf. Curme 1922: 14-15), producing stem forms like hielt /hi:lt/,
base: halt, and wuchs /wu:ks/, base: wachs, which have long vowels although
the stem terminates in a consonantal cluster; cf. also wusch /wu:j7, base:
wasch, which has a long vowel before /J7. It is crossed only by the ban on long
vowels before /if, which rules out quantitative ablaut; only in this extremely
rare case, A—>I-ablaut produces an ablaut form in a non-point vowel (cf.
f a n g
l :
/faiy—»-/fir)/).
If ablaut vowels are point vowels, this will tend to maximise formal con-
trasts between base and ablaut forms. Further, as noted above, lengthening
usually produces ablaut forms of marked length; thus in these cases the means
available for ablaut are, as it were, made use of in full. Hence I refer to this
type, covering all changes of quality type except I—»U-altemation, as full
ablaut.
Vowels of ablaut forms due to I—»U-altemation are invariably non-point
vowels (viz. /o/, /o:/, /u/); thus the ablaut forms concerned cannot be confused
North German. Secondary stem forms that are formed by means of umlaut do not show
quantitative change.
27 Cf. Martinet (1957: 264): “[vowels that occupy] the corners of the vocalic triangle”. As
has often been noticed, point vowels have a privileged position in vowel systems (cf., in
particular, Jakobson 1941), and it is assumed that the preference for point vowels is related
to their favourable articulatory and auditive properties (cf. Stevens 1972). Apparently,
ablaut forms that show point vowels are optimal if reliable differentiation and salience are
‘desiderata’.