[I was going to post this at the end of the series, but I thought perhaps it
would be helpful, if anyone is paying attention at all, to state in more detail
the soundlaws behind the reconstructions as given in the last two postings.
Beware that this chapter is only a first draft version, and isn't finished
yet...]
- Appendix: Soundlaws
The pre-PIE phonological system which I take as point of departure for the
developments leading up to PIE as we "know" it, is the following:
Short vowels: *a * i *u
Long vowels: *a: *i: *u:
Diphthongs: *ay, *aw, *a:y, *a:w
Semivowels: *w *y
Nasals: *m *n [*N]
Liquids: *l *r
Labials: *p [*b] *bh
Dentals: *t *d *dh *s *z
Velars: *k *g *gh *x [*gh]
Uvulars: *q *G *Gh *X *R
Glottals: *h *?
The existence of *N (*ng) may be postulated in order to simplify the structure
of such forms as *hásxung- "blood" (> *h1ésh2ar(gW)) [or to make Nostratic
connections for the word easier, e.g. with Kartv. *z-isxl], but the entity
behaves in all respects as a consonant cluster *ng, which is how I'll notate it.
The voiced labial *b was extremely rare (perhaps absent), for which an
explanation is required, but for now I'll write the stops in the traditional
fashion (*t, *dh, *d), without using alternative glottalic notations [none of
the different variants of the glottalic theory is entirely convincing].
The presence of *z is demonstrable only for the nom.sg. (animate) marker. In
all other cases, I'll write *s. Likewise, there is no need to distinguish
between *x, *X and *gh (we usually cannot tell which was which), so I'll use
only *x. The same goes for *R, which can be recognized in the causative and
some thematic formations, but not in lexical roots. The distinction between *h
and *?, too, is mostly beyond recovery, so I'll use only *h.
The other essential symbols are - (morpheme-break) and ' (accent).
Roots had the structure (s)(C)CV(:)C(C)V, and suffixes (C)(C)VC(C).
Input Rules:
*** Root/Compound rule
A noun or verb can consist of a root (+ endings) or of a root + suffix (+
endings).
*** PD/HD rule
A noun can have the main stress on the root or on the suffix. Suffix-stressed
nouns are exclusively of animate (masculine) gender, root-stressed nouns mostly,
but not exlusively, inanimate.
*** Final vowel loss
At the most remote stage, pre-PIE roots and suffixes always ended in a vowel
(*-a, *-i, *-u) [in the nominal forms having the function of absolutive,
genitive and ergative, respectively], which was then lost.
In the Auslaut, the effects on the preceding consonant were:
*-Ca *-Ci *-Cu
----------------------------------------------------------------------------
*-ti > *-t^ > *-y *-tu > *-tW > *-sW
*-ka/*-qa > *-x > *-h2 *-ki/*-qi > *-x^ > *-h1 *-ku/*-qu > *-xW > *-h3
*-tku > *-dhW > *-tW
*-ghu, *-Ghu > *-ghW
*-gu, *-Gu > *-gW
*-pa > -?? *-pi > ?? *-pu > *-xW = *-h3 (?)
*-bha > *-m *-bhi > *-m *-bhu > *-mW > *-w (?)
*-si > *-s^ *-su > *-sW
*-za > *-z > *-s *-zi > *-z^ > *-s^ *-zu > *-zW > *-sW
*-xi > *-h1 (?) *-xu > *-h3
*-hu > *-h3 (?)
*-na > *-n > *-r *-ni > *-n^ > *-y *-nu > *-nW > *-mW > *-w
*-mu > *-mW > *-w (?)
*-li > *-l^ > *-y (?)
The resulting word could still end in a vowel, if it originally ended in
[absolutive] *-aa (*-a:) [> *-a], *-ia/*-ya [> *-i/*-y] or *-ua/*-wa [>
*-u/*-w]. These are the thematic, i- and u-roots. If there is no further
suffix following, the -a, -i and -u behave as separate morphemes (except for -i
and -u in the svarita rule).
*** Collective rule
A collective is made from a simple noun by shifting the stress one syllable
towards the end, and lengthening the vowel in that syllable. In the NA, the
marker *-x (**-ka ?) is added. Root nouns make collectives of the shape
*R-ú:(-x)
*** Long vowel rule
Root and suffix may not both contain a long vowel. If this happens (e.g. in the
formation of a collective), the unstressed (first) long vowel becomes short.
*** Stress-shift rule
When an ending containing a vowel is added, the stress shifts one syllable to
the end.
*** Svarita rule
The vowel of an immediately posttonic suffix or desinence, ending in a
consonant, is lengthened, if that syllable and the preceding morpheme are not
heavy (i.e. do not contain a long vowel or end in *-CC).
Most input forms as given in the chapters on nouns are already processed by teh
rule sgiven above. The subsequent rules, in approximate chronological order,
are:
*** Fricative rule
When voiceless stops collide with another consonant (before zero-grade,
obviously), the stop becomes a fricative (laryngeal):
*Ck > *Ch2
*kC > *h2C
*pC > *h3C
This does not seem to apply to *t (e.g. *tk > *th2), but see the "instrumental
rule" below for *t > *h1. Perhaps tt > tst can be fitted in here. [Note also
assimilation at a distance in Indo-Iranian "nail", "navel"]
*** R-metathesis rule
When *R- is prefixed, it becomes an infix, except when the first consonant of
the stem is *r.
*** First labialization/palatalization rule
*u(:) (not *w) labializes a following (in the case of velars/uvulars, a resonant
may intervene) consonant as follows:
*u(:)p > *u(:)pW (> *u(:)kW)
*u(:)t > *u(:)sW
*u(:)k, *u(:)q > *u(:)kW
*u(:)g, *u(:)G > *u(:)gW
*u(:)gh, *u(:)Gh > *u(:)ghW
*u(:)m > *u(:)mW (?)
*u(:)n > *u(:)mW
*u(:)s > *u(:)sW
*u(:)z > *u(:)zW (?)
*u(:)x > *u(:)h3
*u(:)h > *u(:)h3 (?)
The effect apparently applies unconditionally to velars/uvulars, elsewhere
*only* if *ú(:) is stressed.
*i(:) palatalizes at least a following *n (*-in- > *-in^-).
*** Thematic lengthening rule
Before voiced segments, a root-final (thematic) vowel *a becomes *a:
*** Vowel reduction rule
*a > *& (always)
unstressed *a: > *a
unstressed *i: > *i
unstressed *u: > *u
*** Vowel breaking rule
*& = *&
*i > *^& (except *iy > *&y, and *-i)
*u > *W& (except *uw > *&w, and *-u)
*i: > *^e: (except *i:y > *e:y, and *-i:)
*u: > *Wo: (except *u:w > *o:w, and *-u:)
*** Second labialization/palatalization rule
*i (> *^&), *u (> *W&), *i: (> *^e:) and *u: (> *Wo:) have the following effect
on a preceding consonant (provided it's in the same morpheme):
labializations:
*k, *q, *g, *G, *gh, *Gh > *kW, *kW, *gW, *gW, *ghW, *ghW
*x > *h3
*m > *mW
palatalizations:
*n, *l > *n^, *l^ (> *y)
*w and *y have the same effect, and additionally (Anlaut only!?):
*pw- > *pW [> *kW]
*bhw- > *bhW [> *ghW]
*tw- > *sW-
*sw-,*zw- > *sW-
In other cases, *^&, *W&, *^e: and *Wo: become *&, *& , *e: and *o:, except
morpheme-initially, where they remain as *y&, *w&, *ye: and *wo:.
*** e-Rule
Stressed schwa becomes *e, as does unstressed *a (< *a:). *a: and *o: merge.
Svarita-lengthened vowel are shortened.
stressed svarita unstressed
& é -- &
a -- -- e
a: ó: o --
e: é: e --
o: ó: o --
*** Brugmann's law
In closed syllables, *ó: > *ó
*** Vowel colouring rules
e > a in the neighbourhood of *x, *q, *G, *Gh
e > o in the neighbourhood of *h3 (*R)
-i > -u before *w or labialized consonant
*** Laryngeal monophthongization rule
Before a laryngeal *ey, *ew etc. become *i, *u:
VyH > iH
VwH > uH
*** ye/we rule
*y&, *ye > i
*w&, *we > u
This applies not only when *y&/*yé and *w&/*wé derive from old *i and *u, but
also when from *ya and *wa.
*** Szemerényi lengthening rule
A vowel followed by one consonant and a fricative (*s (*sW, *s^), *x (*xW, *x^))
and then word end (*-VCF#) is lengthened (*e > *e:, *o > *o:, *i > *i:, *u >
*u:).
*** ww/yy rule
In *-Vwu#, *-Vwm#, *-Vyi#, the vowel is lengthened and *w or *y is lost.
*** Amphidynamic rule
In heavy PD nouns as well as i-/u-stems, the stress shifts from the suffix to
the ending, if that contains a long vowel.
*** Rasmussen initial accent rule
If there is a pretonic *e, it acquires the accent, and any following non-high
vowels are reduced to *&.
*** Zero grade rule
*& > zero (except where too heavy consonant clusters would arise [specify
further]).
*** Final fricative rule
In -V:CF#, the final fricative is dropped if C = r, n, s. In compound nouns and
-s, also when C = m, y, (l)
*** Instrumental rule
*-ét# > *-éh1#
*-étC- > *-éh1C- (?)
*** Rewriting rules:
x = h2
h = h1
k, g, gh = k^, g^, g^h
q, G, Gh = k, g, gh
z = s
<Some other rules, which I haven't worked out and organized into the
chronological scheme yet:>
stative lengthening rule
rules of reduplication
haplologies (Abl/Ins pl. -atatu, 2pl. pp swesw > susw > usw, etc.)
mn-rule (root has rounded V mn -> n, else mn > m)
tl/tr-rule (root has l/r -> tr, else tl)
laryngeal deletion / accentuation with R-infix
Delabialization of clusters ccW -> cc
Labialization dissimilation (1pl. pp mwesw > mesW [but mweswz > mwesz > musz >
mu:s "mouse"], mWei- > wei-) NOT for h3, kW, etc.
Thematic vowel and /i/ (%% -> ye, i%/%i -> % (?))
<And some other rules which I have forgotten to include...>
=======================
Miguel Carrasquer Vidal
mcv@...