the meaning of "genetic relationship"

[bwald]

----------------------------Original message----------------------------
We have been discussing the concept of "mutual intelligibility", an
interesting topic that seems crucial to Isidore Dyen's concept of
"language" (technically, hololect = chain of mutually intelligible pairs of
*dialects* -- or, after his last clarification, perhaps *idiolects*, or,
preferrably, in my view, *lects*, in the sense suggested in the 1970s by CJ
Bailey) and ultimately, in his view, to the concept of genetic
relationship.  About MI (mutual...), D writes:
 
>What I
>mean by mutual intelligibility for these purposes is being able
>communicate with each other in their native dialect. This definition
>produces a first (or native) language dialectology in which a language is
>unitary.
 
As I have been saying in previous messages on this topic, MI between
dialects is a matter of degree.  It seems to me that a quantitative leap, a
lowering of MI, will occur in moving from a bunch of lects within a single
(socially-delimited) dialect to another.  That leap is similar but less
than moving from one hololect to another.  Basically, it all seems to be
about moving from one vocabulary to another.  It is true that some
syntactic constructions can be misunderstood, or not understood across
dialects, and some phonology can take getting used to even if vocabulary
and syntax are familiar, but where there is shared vocabulary (recognised
as shared vocabulary) there will be some degree of MI.  Thus, if a German
says to a monolingual English speaker under most circumstances "komm hier",
the English speaker will understand, and, in fact, not even realise that
s/he was spoken to in German.  So there is some MI between German and
English.  I don't suppose (maybe I'm wrong) that D expects there to be a
chain including English and German in a single hololect.  I'm not sure this
is important.  There is certainly the quantitative leap I spoke of above.
Presumably, the extent of the hololect is an empirical matter, so maybe E
and G do belong to a single hololect although MI is minimal.  I mention the
above to indicate that I understand D's distinction between a "hololect"
and what we usually call a "language" (socially defined).
 
D continues:
 
The term dialect is difficult; the term itself is used in such a
>variety of ways, that in effect the definition a language given above
>really appeals to the intuition in the matter of dialects. However it is
>assumed that no two individuals whose speech-types are in the same
>language are the same. Of course the term dialect is also applied to
>collections of speech-types
>that share a particular feature or some collection of features. [Here
>'feature' is used for some relevant speech phenomenon, not in the
>meaning it has in phonology.] In any case I believe it doesn't matter much
>which definition of duialect you use.
 
That's why I mentioned various terms like "dialect", "idiolect" and "lect"
at the beginning of this message.  Empirically there is a great difference
between a (social) dialect and an (asocial) idiolect.  Speakers of the same
dialect are much more mutually intelligible than speakers of different
dialects.  Idiosyncratic speakers are no more intelligible within their
dialects than outside of them on whatever points they're idiosyncratic on.
The shared speech patterns of a single dialect even goes beyond what seems
necessary for MI.
 
Another rhetorical problem came up; by rhetoric I mean the way people
express their points; it is a linguistically interesting thing about
choices at the discourse level of linguistic analysis.  D objected to my
use of the word 'dogma', as follows:
 
>I should add that I find the term 'dogma' a pejorative term.
 
Of course I am aware that 'dogma' is sometimes taken to be pejorative, but
in a discussion of linguistic complexity a while ago I wrote on this list
that 'dogma' is not always a bad thing.  In context I was saying that the
dogma that all languages are of equal complexity is just that, since we do
not know what we are talking about when we make that claim.  (Not that it's
wrong, just that we don't know what "equal complexity" means, or how to
measure or conceptualise *global* complexity with respect to human
languages -- and the notion that if something changes in a language,
something else MUST change in order to "restore" that "constant" complexity
is vacuous.)  But I said the dogma is *good* because it alerts us to this
problem when we hear somebody say or suggest that language A is more
complex than language B.  And I went on to say the dogma allows us to
recognise an issue, so that the least we can respond to a claim of
"inequality" in complexity is: "what do you mean?"   And it invariably
turns out that they mean something much more specific than "global
complexity", e.g., "Chinese" is "complicated" because of its writing
system, etc etc.
 
Now, the "dogma" I was referring to in D's message is that languages don't
mix -- and, I'm not sure because he didn't respond -- languages don't
converge over an area (how could they without mutual intelligiblity -- and
if they're separate hololects how can they be mutually...?), and maybe
dialects don't either; they just either don't do anything or get more and
more different from each other.  OK, it sounds like I'm being sarcastic,
but he really didn't respond and I can't anticipate how he deals with this.
Anyway, again I don't think the dogma is totally misguided, but there is a
better point in D's message to pursue to this, where he suggests that I
will destroy the comparative method with my assumptions (or maybe it's:
without assumptions like his.)  Let's move on to that.
 
>But if languages
are permitted to mix, that is, if a language boundary between two
languages is permitted to dissolve, then the kind of inferences that we make
regarding the past hstory of a language must take the possibility of
mixing into account.
 
Yes.  Why not?  But NB, the language "boundary" between two languages never
totally dissolves.  It only dissolves on certain points.  Note, for
example, Gumperz's classic study of Marathi and Kannada in Kupwar.  More or
less the same grammar (including phonology) but different lexicons.
 
>The consequence is that the hypothesis of a
protolanguage becomes unavailable unless the possibility of mixture can be
ruled out.
 
Why?  Mixture doesn't have to happen.  But basically we ignore
mixture/convergence according to the comparative method and concentrate on
what can be accounted for without it.  That's fine.  But to then say "case
closed", nothing else can happen, or ever does, and languages can't mix,
that's like saying "you can't cross the street against the light.  I mean,
you CAN'T.  Try it, you'll see you can't!"  (But you can -- if you don't
get run over.)
 
In any case, there are two situations that test the limits of the
comparative method (the one on which the genetic hypothesis is based), 1.
two languages with the same grammar and different vocabularies; 2.  two
languages with the same vocabulary and different grammars.  These are
matters of degree, but at the extremes, 1. is reflected in the Kupwar
Kannada-Marathi situation, among others (with few historical linguists
exploring them instead of claiming they are "rare"), and 2. is reflected in
such things as creolisation etc.  The issue is very meaningful for the
comparative method and reconstruction.  Thus, for example, we have the case
of Hittite, where some scholars, e.g., Lehmann, argue that it reflects a
much more archaic grammar of IE than other surviving branches, while others
argue that many of these features are due to convergence with non-IE
languages of the area (which have similar features).  The "dogma" against
mixing does not help resolve this issue on way or the other.
 
>That is the function of the assumption. In a first language
dialectology applied through time, once a language has been
formed, its being disjoint cannot be destroyed. If you are willing to give
that up, I would say you are giving the power of the comparative method.
 
A strange use of the word "assumption" to me.  I'd say the comparative
method works for what it works for, and it doesn't deserve any more power
than that.  Now, the idea that the disjoint(ed)ness of a language once it's
formed can't be destroyed is another matter.  But I'm reading "language"
here as individual's linguistic system formed before a critical age.  More
on that later.
 
To get beyond the English-German hololect, D writes:
 
>The view is available that zero mutual intelligibility occurs.
 
It is almost nonsensical to deny this, though it is not as obviously true
as it seems (e.g., unidentifiable languages heard on the radio convey even
less than those witnessed performed in public, according to my personal
experience).  The point remains that MI develops somehow under contact
situations.  D wants to disregard this by insisting that there is a
strictly circumscribed item, call it a native (monolingual?) dialect
(sociolinguists often call it a "vernacular") and that is the only thing
that counts for *genetic* relationship.  But note if you rule "mixing"
(accomodation) out by definition, you simply choose to ignore the variety
of ways in which languages change, both synchronically, and (relevantly)
diachronically.  What about areas where people generally grow up
multilingually?  What effect does that have on "genetic" relationship?
(Interestingly, it varies, in Bantu East Africa genetic relationships can
be demonstrated by conventional comparative methods, even for areas which
are highly multilingual, but tree classification of Bantu languages in such
areas and in general, is, for the most part, a hopeless mess.)
 
I wrote:
>>one might be tempted to assume that only through mutual intelligibility
>> can dialects influence each other and changes spread from one dialect to
>> another -- and there is no doubt some truth to this.  ....
 
D responded:
>Without being disrespectful, let me suggest that you have used the term
>'assume' above in the sense of 'conclude' or 'infer', not in its ological
>sense. I don't object to l/ay terminology; I am not a logician.
 
1. huh?
2.  what's "l/ay terminology"?  Is it the vernacular?
3.  I thought I meant "assume" (although a logical process leads to that
assumption by ignoring
      certain facts).   Elsewhere, D seems to use "assumption" to mean
"postulate" (noun).  To me an
     "assumption" can be like a "guess", and a guess is often based on a
rational process, as
     suggested in the word "guestimate".  In any case, assumptions, like
guesses, can be wrong.
 
Finally, D brings up a point which has long intrigued me, and we have some
common ground:
 
>I believe that you have begun to touch on the very important question that
>deals with the time at which an individual can be said on the average to
>be in control of his native language. If I suggest at the end of the first
>decade of his life, I imagine I might attract some disbelief. For certain
>purposes however, it strikes me as being not an unreasonable expectation.
 
This is relevant to the "vernacular" which D insists is the object of
genetic classification (to the exclusion of any other objects, it seems).
As for disbelief, I think a "critical age" between 10-12 is widely accepted
-- for PHONOLOGY (some put it even earlier, with some good evidence).  But
not for much of syntax and vocabulary.  Since MI seems to depend crucially
on vocabulary, there is some explaining to do here.  At the same time, it
may reduce "genetic relationship" largely to phonological evolution,
something which would not surprise me in view of traditional practice,
though not traditional belief.  Structures that may be acquired well after
the age of 10 are also subject to linguistic evolution, however, as is any
non-universal feature of language (whatever the universal features of
language turn out to be). Therefore, there is more to be said, esp about
syntactic evolution.
 
P.S.  I get the sinking feeling that in the end D will say everything he
was saying was ONLY relevant to the point of the term GENETIC relationship,
and that everything I said could be true but not to the point.  Whether or
not genetic relationship is a big deal in the totality of ways in which
languages can change might be considered a separate issue, the issue of
*internal* change.  We note that even with respect to borrowing there have
been linguists like Jakobson who have proposed that languages can only
borrow what could also result (?spontaneously) from internal change (what
is consistent with its structure before the borrowing, or some such
notion).  That remains an interesting (!) idea, not always easy to
distinguish from vacuity, and intended for borrowing beyond the
arbitrariness of the lexicon.