Genetic Descent

[David L. White]

[ moderator edited ]

>> Whether TK meant "anything goes" or not depends on how strong a reading we
>> give to their statement.

> LT:  But why should we impute to them a position which they have never
> asserted or endorsed?  Is this fair?

DLW:  My point is that their formulation includes possibilities that do not
actually occur, _in their own evidence_, which I presume they can reasonably
be expected to be familiar with.  Their failure to note this is surely a
failure of some sort, and whatever we choose to call it, it is fair to call
it something not good.

>> It should be noted that whether finite verbal morphology gets borowed at all
>> is a matter of definition,

> LT:  Sorry; I can't agree.  If language A gradually replaces most of its
> inherited verbal morphology with morphology taken from language B, while at
> the same time retaining much of its inherited lexicon and grammar, then
> there is no definition about it: the language has borrowed verbal
> morphology, and that is the end of the matter.  T & K's stance is that we
> should consider this a possibility and look to see if there are any
> examples of it.

DLW:  There aren't, save by hypothesis.  To both assume and conclude that
the "gradual replacement" scenario has happened is entirely circular.  It is
indeed a matter definition, or at least of theoretical stance.  We have one
simple fact:  there are no attested mixed finite verbal morphologies.  This
can be viewed either as reflecting a mystic pricinple that "There is no
partial borrowing of finite verbal morphologies" or as indicating that
"Genetic descent can always be traced through finite vebal morphology (where
applicable)".   Or we can 1) assume that in the past there were mixed finite
verbal morphologies, so that their absence in the present (and the attested
past) is therefore simply a misleading coincidence, and then 2) triumphantly
conclude that mixed finite verbal morphologies are possible, QED.  This is a
sub-case of the "it was different in those days" view of language change, by
which all sorts of bizarre things happened in the past, in situations we are
not able to observe, in this case mixed languages lurking under every rock,
a la Steve Long.  You quite rightly decry this sort of approach in a very
recent posting.  Good luck, by the way, answering Steve Long's objections to
what might be called "mono-descent", in view of what you have recently said
concerning the supposed applicability Tiger Woods' genetic descent to this
matter.

> LT:  Thomason now concludes that Laha, at least, is indeed an
> example of this very thing.  Whether she is right in this is a matter for
> empirical investigation, not a matter of definition.

DLW:  Unless there is something in the facts that can prove that a partial
Malay finite verbal morphology existed at one point (not likely, since Malay
has no finite verbal morphology) it remains fundamentally a matter of
theoretical stance.

> LT:  Well, Laha is on the table.  The chief investigator, Collins, has
> concluded that Laha has indeed borrowed almost all of its verbal morphology
> from Malay.  What is there to discuss, apart from the possibility that
> Collins may be in error?

DLW:  That, even if true, is not proof that a partial Malay finite verbal
morphology existed in Laha at some point in the past.  And "borrowing" of an
entire verbal morphology may just as well be taken as descent from the
supposed source of this.  Consider the case of Mednyj Aleut.  If Collins
claims (with good evidence) that Laha at one point had part of its finite
verbal morphology from Malay, that would be different, and Laha would indeed
be a mixed language.  But we are, to use DGK's phrase, playing ping pong in
the dark without access to the original article.

>>> LT:  If so, let me ask this: in what respect does Michif fall
>>> short of being a mixed language?  It looks to me like a paradigm case of a
>>> mixed language.  If we encountered, or imagined, a real mixed language,
>>> what features would it have that Michif lacks?

>> DLW:  The finite verbal morphology of Michif is all Cree.  Therefore, by
>> the Davidian standard, it is Cree.  That also answers the scond question.

> OK.  My friend Homer is a specialist in nouns.  He loves nouns, and he
> regards them as paramount in languages.  According to Homer, the ancestry
> of a language is determined by the origin of its nouns.  Therefore, by the
> Homeric definition, Michif is French.  End of discussion. ;-)

> Why is Homer's position more arbitrary than yours?

DLW:  You seem to be repeatedly missing some basic points here.  Not only is
there a good case to be made for the proposition that verbs are more
fundamental ("higher") than nouns, but there are some mixed nominal
morphologies, whereas there are no mixed verbal morphologies.  The cases are
not parallel.  Among other things, "Homer's" standard would not yield a
clear verdict in cases of mixed nominal morphology.  He would also have to
say why nominal morphology is to be regarded as the most "framish" part of
the frame, and therefore the most reliable indicator of genetic descent,
when the evidence suggests very strongly that this honor goes to verbal
morphology.

[on Takia]

>> If the finite verbal morphology (morphemes, not categories) is Takia, it is
>> Takia. That may well seem arbitrary, but it is not.  Verbs are more basic
>> than nouns, and what is more basic is more indicative of descent.

> LT:  I'm sorry, but I cannot agree that verbs are more basic than nouns.
> This seems to me an arbitrary fiat.  Anyway, I've never seen a language
> without nouns, and that suggests to me that nouns are about as basic as
> anything can be in languages.

DLW:  It is not a matter of what is or is not basic in some binary sense,
but of what is more basic in the sense of being higher in the tree.  Nouns
depend on verbs, not the other way around.

> LT:  But I'll discuss this in terms of what follows.
> A little quibble first, though.  Arguably, I possess no characteristic more
> "basic" than my biological sex, which is male.  Does it therefore follow
> that I am more closely related to my father than to my mother? ;-)

DLW:  Vague analogies with biology are not really relevant here.

> LT:  The position that the verb is the head of the sentence is embraced in
> certain syntactic theories, especially in most dependency grammars and in
> some relational grammars.  But it is rejected in other approaches,
> including other relational approaches and all Chomskyan approaches.  The
> Chomskyan view takes the head of the sentence to be an abstract element,
> and not the verb.  I am not a Chomskyan, but I don't think it can simply be
> taken for granted that the verb is the head of the sentence.

DLW: It is not being taken for granted.  It is in effect part of what is
being posited:  "if we assume that verbs are more basic than nouns .."  By
they way, the Chomskyite "INFL" (if they are still doing that:  I know
little and care less) is historically descended from the inflection of the
verb, so their position too would have to be that verbal inflection (in this
case finite) is more basic.

> LT:  There is also a fascinating counter-argument here.  Consider this
> example:

>   'That Martians are green is well known.'

> Now, the subject of 'is' is plainly the complement clause 'That Martians
> are green'.  Fine.  Now, what is the head of this clause?  Is it the verb,
> as David seems to be implying?  Then the head of the subject is 'are'.  But
> -- whoops -- this is plural, and yet the whole clause takes singular
> agreement.
> Oh, dear. ;-)

DLW:  This is not so fascinating, or problematic.  It is fairly easy to say
that the subject of the verb is "that", which would explain why it cannot,
in such cases, be deleleted.

>> We may also note that things that might be held to apply to
>> an entire proposition, like negation, are typically attached to or
>> associated with verbs,

> LT:  It is true that sentence operators, such as negation and tense, are most
> typically associated with verbs.  But this is not true without exception.
> For example, in Sanskrit -- I am told -- the negative marker can appear
> anywhere in a sentence at all, and need not be attached to the verb.  (OK;
> I'm ready for flak from the Sanskritists on this list. ;-) )

DLW:  You do not have to wait for flak from Sanskritists.  What you say is
true not only of Sanskrit, but also, marginally, for English:  "I see not
any good cause ."  To use "no" would be more normal, but "not any" seems
possible.  In any event, it is not relevant.  I said "typically", not
"universally".   And we do not, for example, find markers of tense attached
to subject nouns.

>> and that though there is a type of language,
>> polysynthetic, in which the verb and its mandatorily associated elements
>> form a complete sentence, there is no corresponding type of language
>> where the subject noun and its mandatorily associated elements form a
>> complete sentence.

> LT:  It is true that, in *some* languages, a sentence may consist entirely of
> an affixed verb.  But it is equally true that, in *some* languages, a
> sentence may consist entirely of nominal elements, with no verb-form present
> at all.

DLW:  There is no _type_ of language where the subject noun and its
mandatorily associated elements (which would include tense) form a complete
sentence.  And except for cases of BE-deletion (which only occurs in the
present), adjetives that serve as (and are marked as) verbs, and NPs uttered
when context makes clear what the higher VP would be, I venture to doubt the
validity of your assertation.

>> Thus upon examination it is not arbitrary to regard
>> vebal morphology as more basic than nominal morphology.

> LT:  Ah, no -- for several reasons.  I've just mentioned two: it is not an
> obvious truth that the verb is the head of the sentence, and it is not true
> that verbs are universally more indispensable than nominals.  But there is
> more, much more.

> What David White is arguing, successfully or not, is that the *lexical verb*
> plays a central role in sentence structure.  Even if we grant this, it *does
> not follow* that verbal morphology is somehow more central than any other
> morphology, such as nominal morphology.  Verbal morphology is not the same
> thing as a lexical verb.

DLW:  This is a point you yourself seem to miss, in attributing to me
beliefs about lexical verbs which were intended to apply only to verbal
morphology.

> LT:  Moreover, David's arguments here seem self-contradictory.  Consider a
> language with native verbal morphology but with many borrowed lexical verbs
> -- such as English or Basque.  In such a language, it is the (borrowed)
> lexical verb which serves as the sentential head, in David's account, and
> which determines the syntactic properties of the sentence.  At the same
> time, the native verbal morphology merely follows the requirements of the
> lexical verb in a wholly passive manner.  But now consider David's
> position: the (borrowed) lexical verb is the most central element in the
> sentence, and its properties determine the structure of the sentence,
> including the associated morphology.  Yet it is the native verbal
> morphology which is somehow (I haven't followed this).

DLW:  No kidding.

> LT:  .more central in identifying the genetic origin of the language,
> because this morphology is now what is "central".  Have I missed something?
> This line of thinking seems wholly inconsistent to me.

DLW:  Let me back up (or off) and attempt to restate what I am saying.
Finite verbal morphology is the most "framish" of the frame.  As the frame
is reduced, under severe external influence, finite verbal morphology is the
last to go.  (If it does, then genetic descent has truly been lost.  But
this is astronomically rare.)  To put it another way, when genetic descent
is backed into a corner, finite verbal morphology is that corner.  Yes, this
is all "by hypothesis".  But the hypothesis is, of course, intended to serve
a useful prurpose or two, and has not been invented out of sheer
counter-trendy perversity (though I am surely guilty of that too).

The main "useful purpose" served is to explain why we find mixed nominal
morphologies but not mixed (finite) verbal morphologies.  People will borrow
nominal morphology ("alumni", "crises") before they will borrow verbal
morphology.  (Note that nobody even bothers to worry about the "proper"
preterites of Romance verbs in English, for example.  They just will not
intrude so much foreign-ness into "the frame".)

Another "useful purpose" is to provide a clear standard for assigning
genetic descent in doubtful cases.  (See just above for why the standard is
_not_ arbitrary.  I am getting tired of hearing about how it supposedly is.)
I note that no answer has been given to my assertion that TK might as well
be tossing a coin in some of their assignments.  I presume this is because
there is no valid answer.  There is indeed little to be said for a
theoretical framework that in one case of "Language A morphology, Language B
words" assigns the language in question to Language A, but in another case
assigns the language in question to Langage B.  If things like this are TK's
main selling points, I hope they give up sales.  Be that as it may, with a
clear standard, we can avoid not only coin-tossing but also finding mixed
languages (or worse, semi-mixed languages) under every pre-historic rock.
Thus we have an answer to the "No Proto" crowd, which is more than TK can
manage.

I will give a quite relevant example of what I mean by this.  If genetic
descent is to be traced through non-sound features (I notice you are no
longer actively claiming this), then why isn't Celtic a mixed language, so
that there was never really any such thing as Proto-Celtic?  The non-sound
convergence of (Insular) Celtic to Semitic, or something like it, is quite
striking, and extends even down to the level of fairly minor details like
the Northern Subject Rule and Interior Possession ("You killed my horse"
rather than "You killed to me the horse")  I suppose you and others could
claim that the level of Semitic-seeming non-sound resemblances in Celtic,
though it is "a lot", is not "enough", but who is to deny the Steve Longs
and other No-Protos of the world when they come back and say "Yes it is."?

> LT:  It seems that David is telling us that, at one and the same time, the
> borrowed lexical verb is paramount, while the native verbal morphology is
> also paramount.  And I can't follow this.  Surely it has to be one or the
> other, at best.

DLW:  I said nothing about lexical verbs being "paramount" in genetic
descent.  You made that one up yourself.  The argument about verbs
(presumably lexical verbs) being heads (not the same thing) was made in
support of the arguments about the primacy of verbal morphology in the frame
of language, and therefore in genetic descent.  It was not an argument about
the primacy of lexical verbs.  Two different primacies.  In my example with
"necesitates", the verb is the head (more than any of the nouns, anyway),
but the ending is part of the frame, in a neo-Swadeshian sense of being the
most basic morpheme.  Speaking of waxing Neo-Swadeshian, note that if we
included "preterite" (more or less) as a basic meaning and took "/o/-grade"
as its phonic instantiation, the IE-ness of Germanic would be evident on
that basis alone. On a smaller scale, the Germanicness of all Germanic would
be evident from dental preterites alone.

> LT:  Finally, let me advance a counter-argument.  As David's own 'tank'
> example illustrates, it is commonplace in languages for verbs to agree with
> nominal arguments -- that is, for intrinsic features of nominals (person,
> number, gender, maybe others) to be copied onto verbs.  But it is virtually
> unknown -- perhaps entirely unknown -- for verbal features to be copied onto
> nominals.  Is this not a splendid argument that nominals are more autonomous,
> more central, more "basic", and that verbs are merely the slaves of nominals?
> ;-)

DLW:  Not really, no.   Polysynthetic languages have been described as
having their nouns dependent on verbs like planets orbiting the sun.  I
think that is a good analogy.  Thus in Swahili if you were to say the
equivalent of  "The crocodile ate the book", the sentence would have (I
forget the ordering) a noun meaning 'crocodile', a noun meaning 'book', and
a verb meaning (more or less) "he ate it".    Class markers would bind the
element meaning 'he' to 'crocodile' and the element meaning 'it' to 'book'.
(Theoretically if 'crocodile' and 'book' were in the same class as there
would be ambiguity.)  The verb is a complete sentence, with its associated
nouns clearly (I would think) dependent on it.  If, by your argument, the
class concords indicate that the verb is to best regarded as dependent on
one of the nouns, which is it?  Are we going to diagram the sentence as
having its verb under two nouns?  I hope not.

 By the way, pronouns are often cliticized and effectively bound to the
verb.  Thus it is that "look up the answer" is possible whereas "look up it"
(with the same meaning) is not.  I note that TK's only two examples of
pronouns supposedly having been borrowed involve English supposedly
borrowing pronouns from other Germanic languages (Norse, Dutch) that had not
necessarily (far from it) diverged to the point of mutual
incomprehensilibility.  This is surely a suspicious coincidence, and I do
hereby officially venture to doubt that there are any examples of personal
pronouns being borrowed in cases of unequivocal mutual unintelligibility.
(Other than silly things like the joke usage of "moi" in English.)  So prove
me wrong.

Dr. David L. White