There is
conflicting evidence over whether the accusative singular ending of uters in
Proto-Indo-European was -n or -m. To simplify the problem for
illustrative purposes, ancient Greek had -n or (from syllabic-n) -a
(with no native words ending in-m); Sanskrit had -m (with some words
ending in -n as the result of the loss of a final consonant or consonants); and
Latin had -m (with the particle an and the occasional word ending in
-n because of paradigm leveling; cf. pre-Classical sanguen [n.]
"blood" with its gen. s. sanguinis). On the face of
it, the PIE ending is hard to determine.
Naturalness theory and marking theory say that in the
word-final (not elsewhere in a word) position, the consonant -m is less marked
than -n--which can change to -m. (The reasons for this are
extensive but include sound distributions and sound changes generally found; the reader
can take the writer's opinion or check the matter out.) But syllabic-n is
found in languages in which syllabic-m is missing (or rare, as in English, where
it arises out of syllabic-n by assimilation); so, all else being equal, the
presence of syllabic-m implicates the presence of syllabic-n in a
language, and this implicationality means that syllabic-n is the less-marked.
All else being equal, the more-marked is prone to change to the less-marked
(assimilation is a marked environment that may reverse what is more or less
marked)--unless the item itself is so marked that it requires being markered.
However, the accusative singular is one of the least marked case inflections.
If the issue were not complicated by the syllabic
nasals, we could say that the Greek situation is older, Sanskrit and Latin exhibiting
changed sounds. But the consonant-stems are older than the thematic or vowel stems,
other than the short-i-stems--which, at least in Latin, have become confused with
them--though in a neatly implicational manner, as shown by this writer and W. Würzel.
Consonant stems have PIE syllabic-m--which became -em in Latin.
(That this syllabic-m derived from an earlier unstressed vowel plus apical
nasal consonant is probable but not very relevant here, especially when the problem is
being simplified for illustrative purposes.) This line of reasoning helps us
with the Greek and Sanskrit a bit less than one might expect on the face of it, since,
e.g., Greek and Sanskrit syllabic-n became [a]. But if we assume that an
early syllabic-m changed to less marked syllabic-n (and then to -a
in Greek and Sanskrit), the picture is consistent with marking theory.
How then, did Greek get consonant-n on the thematic
(vowel) stems? The answer is less obvious. Among various possibilities
to consider is the possibility that consonant-n was original for consonant-stems,
and that unstressed vowel + n changed to syllabic-m the way syllabic-n
changes to syllabic-m in open and happen in English, without
affecting the -n that remains before a vowel--as in op(m)ning,
happ(m)ning. But two considerations make this
hypothesis problematic: (1) Syllabic-m is assimilated from less-marked syllabic-n
in English only after labials stops--not, e.g. after //s//--whereas PIE
lacked (except in a very few borrowed words) one order of labial stops--what used to be
postulated as *b. (2) Unlike the English situation, no vowel followed the
ending in PIE to preserve it in the underlying form, though--if vowel-stems got an -n
prior to the assimilation of syllabic-n to syllabic-m--the analogy of
the -n would easily convert the reflex of syllabic-m to that of
syllabic-n. This hypothesis is not necessary, in view of the reflex
of a syllabic nasal is "a" in Greek and leaves us wondering how an
original syllabic-n could become syllabic-m in any way except by
assimilation. (Sanskrit is more complicated, but in ways that do no violence to the
hypothesis being comment on. I will omit discussion of the possibility that either
nasal consonant changed to still less-marked agma [n]--the least-marked word-final consonant of the three and the most-marked
syllabic-nasal of the three; and then through some process became a dental nasal.)
The situation that is most compatible with the facts given
is to suppose that unmarked -m occurred word-finally in PIE; when it became
syllabic, it became unmarked -n in some lects; and this apicality was
taken over by the vowel-stems in the earliest Greek--the result of trying to combine a
vowel plus a syllabic-n. At least in early Indo-European, the sonorants
were ambiguous as to syllabicity--becoming non-syllabic or syllabic according to general
principles that have long since been investigated. When PIE syllabic-m
became vowel + m in early Italic consonant stems, there was no (assimilatory or
other) reason for it ever to change to (vowel +) more-marked final -n.
Final -m was then found generally, but final -n can sometimes arise
through loss of, e.g., a final -(nt)s. This at least explains what did
happen--Latin and Sanskrit reflecting the original situation and Greek getting
"n" through natural changes. There may be other explanations compatible
with the facts and with pan-language naturalness theory, but this one does not sound
far-fetched.
Other ideas on the matter that presuppose the naturalness
framework would of course be welcome.
The failures of reconstruction--or reconstruction "theory"--are enormous. Aside from the pervasive ignoring of the rôle of markedness, one may mention:
Listing basic words, the most borrowable of all items, and other surficial things like word order, while ignoring STRUCTURES. (Anyone cognizant of the English spoken in any country where it is a major language will see that the idea that basic words are not borrowed--the tiresome cant not only of the glottochronologists but also of respectable historical linguists--is utter nonsense; counterexamples abound.) Counting beans is poor substitute for analysis.
Structures are what matter most, since they borrow less readily. Note how the English present anterior m odality parallels the older French passé indéfini; and in connection with markedness, note how this marked modality is dying out EXCEPT in marked modalities like the infinitive. It is easy to show that the English exochronous modality replaced a number of uses of the old subjunctive (a Germanic and French a mode lacking in current English), as I show in the volume cited and other places. But aside from proofs from marking theory and just plain old logical grammatical analysis, isn't it more than weird to call speaks and makes a "present" in "She speaks Spanish but she's speaking English" and (on Saturday night when the factory is closed) "This factory makes lightbulbs"--not to speak of examples like "He arrives at eight tomorrow" or (what is not parallel with the foregoing, however parallel it may look to the undiscriminating observer) "When it arrives, be sure to let me know."
Treating calques etymologically rather than in terms of the system they are part is a losing approach. Thus, English has five calques on French equivalents of for--in front of, like French en face/front de; instead of or in place of, like French au lieu de; because of, like French à cause de (calqued in German as aufgrund von). And what about on behalf of? I have elsewhere shown an important number of "basic" Germanic words that function like the French words they calque; cf. caseless who? with caseless qui? (Every student of the history of English knows that whom was invented--whether by appending Latin accusative -m [A-S had -ne] or not--just as every user of English knows that it is so foreign that even grammarian columnists not seldom misuse it (in examples like "the one whom I thought did it" and "gave it to whomever asked for it").
I have shown in my Essays on time-based linguistic analysis (Oxford University Press, 1996) that English double negatives and many other phenomena--but particularly expressions like "John and myself did it" (less educated "Him and me did it") and "to she and I"--are structurally French. (There is nothing--except, irrelevantly, etymology--Germanic about these examples.) When one pays attention to structures, etymology takes a backseat. Many things decried as incorrect by grammarians who are besotted with the idea that English is Germanic (however obviously English typologically differs from Germanic) are correct in French. And if they are correct in French, what grammatical principle rules out the possibility of their being correct in other tongues?
To understand the phenomenon just discussed, it is necessary to grasp the implications of W. Mayerthaler's principle of reversals in marked categories and marked environments. In chap. 10 of the volume just cited, I show that only with this principle can we understand the switchings of early Modern English shall : will or current English be gonna : will in English grammatical analysis, as well as the be- and get-passives and much else.
Calling Proto-Indo-European a "tone" language when it is notable for a phenomenon that "tone" languages lack (unless they have inherited it from an earlier phase of their development)--viz., vowel reduction--is not very convincing.
Instead of going on and on with demonstrations that historical- reconstruction "theory" is in a really parlous state, I will just conclude by observing that the above comments represent a minimal list of amendments to the "theory" necessary for it to be able even to pretend to be half-adequate.
