The passage from one stress system to
another can be gradual thanks to the existence of secondary stress. Imagine a
language in which the location of primary stress is lexically determined (and so
cannot always be predicted from the segmental and morphological composition of
the word), but which develops secondary stress on the root-initial syllable,
playing a demarcative function (indicating the beginning of the central
morpheme). Now the tricky thing about stress is that it is RELATIVE. Speakers
use a combination of phonetic cues (pitch, duration, loudness and segmental
features) to make certain syllables perceptually prominent. Usually,
a syllable is considered to be stressed if it is more prominent than any
adjacent syllable. Primary stress (within a given domain) requires some
extra prominence, e.g. a change of pitch in addition to greater loudness,
increased duration and full vocalism. If a neighbouring syllable, or any other
syllable in the same word, receives more prominence, e.g. in connected speech,
or for the sake of emphasis, it may "steal" stress from its original
locus.
For example, English <thirteen> is
end-stressed if pronounced in isolation, but English has the so-called Rhythm
Rule (or Iambic Reversal) that modifies the stress pattern of certain words to
ensure rhythmic stress alternation ("strong-weak strong" rather than
"weak-strong strong"). As a result, in the phrase <thirteen points> the
main stress is on <points> and the secondary stress is on
<thir->, while <-teen> loses its rhythmic prominence.
The Germanic "tendency" I alluded to would
have consisted in placing obligatory emphasis (manifesting itself as a burst of
initiatory energy) on the root syllable. This was at first superimposed on the
inherited stress pattern, leading to a tug-of-war between two rival stress loci
(if there WAS a mismatch, i.e. if the original stress was non-initial) and to a
gradual (word-by-word, speaker-by-speaker) increase of preference for the
initial syllable as the predictable location of PRIMARY stress.
Stage I (pre-Germanic):
*paté:r
Stage II (after Grimm's Law):
*faþé:r
Stage III (just after Verner's Law):
*fadé:r
Stage IV: free variation between *fáde:r
and *fadé:r, with the lexically listed stress pattern gradually phased out in
favour of predictable root-initial stress (vive la simplicité!).
Stage V (and beyond): *fáde:r
(consistent initial stress) > *fáder > ['fad&r] (the second syllable
is gradually deprived of all phonetic cues that could be interpreted as
correlates of stress, such as vowel length or "full" quality).
Far from being speculative, this is a
frequently reenacted scenario. The question is only whether Proto-Germanic
(defined as the most recent common ancestor of the known Germanic languages)
represented Stage IV or V. It doesn't matter much, since the internal dynamics
of the process makes Stage V follow Stage IV almost inevitably, so if the
incipient shift was not completed before the fragmentation of Proto-Germanic,
the daughter languages would have carried it out nevertheless.
Piotr
----- Original Message -----
Sent: Monday, July 30, 2001 3:35 PM
Subject: Re: [tied] kuningas <-> knyaz
Piotr:
>The TENDENCY towards initial stress may have
begun in Proto-Germanic to be parallelly developed in the daughter languages.
Anyway, the time window is pretty narrow, though its absolute dating is
difficult (from ca. 100 BC to ca. AD 200).
How might this "tendency" for
initial stress develop? Either the stress is initial or it's not. I need
clarification on that one.