Contributions by Steve Long

[ECOLING at aol.com]

This message is about positive contributions
which Steve Long has made to our field, in discussions on this list.
Most of those contributions have to do with the PRESENTATION
of the results of comparative-historical linguistics to those who may
or may not be actively involved in the field, so we should be grateful
to Steve for bringing these to our attention.

I do not like the fact that the title of the message is "ad hominem",
to the extent that it mentions a particular member's name,
but since this message is both about the content
and about respectful treatment for a member of our list
who has made positive contributions,
I cannot think of a better title.

***

I find the recent discussions on the IE list disturbing,
because of the failure to come to grips with real issues
raised by participant Steve Long,
because of the treatment of him as if he were an ignorant child,
by others assuming he could not possibly have a valid point
when he questions the way things have been done for years.

Quite on the contrary, judging purely from the logic of the reasoning
presented and my specialist knowledge as a trained PhD linguist (*see below),
I believe Steve Long has won the argument on several points.
The rest of this message refers ONLY to those several points,
not to other areas in which I might agree he lacks knowledge.
It is our obligation to treat people with respect and to try to find the
positive and valuable in what they contribute.

What perhaps first frustrated people with Steve Long was his not assuming
that some kinds of innovations could go only in one direction,
or that there was actually sufficient data and analysis available to
demonstrate where the root of the IE tree should be placed,
or that there were innovations on the branches of the tree which were
not labeled (the "stem"), or other matters similar to these.
However, in several of these areas, his persistence has nevertheless been
valuable, by pointing to weak links in our publicly presented reasoning,
potentially also in the reasoning itself,
however inconvenient this may be to professionals.
It is important that professionals in any field
should be able to explain their actual reasoning without inadvertently
omitting or hiding important links.  Communication with non-experts in
whatever fields is very healthy for this reason, among others.

Steve Long has also done rather well to persist despite the treatment he has
received.
That in no way means that I think he is right about everything,
nor that he is equivalent to a trained professional in various fields.

His persistence in questioning what he is being told does NOT display
lack of respect for those he is conversing with (unless one means that
he is supposed to display respect simply by accepting what he is told).

The many red herrings that have been introduced also put him at a
considerable disadvantage, asking him to produce examples which were
not even needed, when he does not have the expert knowledge to
get the best examples, and then focusing on what he gets wrong.
These deflected the discussions from his original points.

***

Here are the items on which I think Steve Long's contributions
have been to the point, or his questions have implied a correct
understanding, one different from what we usually take for granted.
I have succeeded in learning from the discussions provoked by him,
AND IN THAT RESPECT FROM HIM, as well as from the others
on this list.

Point 1.

Family Tree representations can be much improved.

Steve Long's contributions have shown that the types of family tree diagrams
which we are accustomed to using in our tradition
DO NOT adequately represent even the locations of innovations
on which such trees are by all accounts theoretically based,
not even whether innovations occur or do not occur on particular branches!
I have noted details in a previous message,
suggesting what we might do about this.
I will certainly think twice before producing or reproducing further
family trees which do not show such information!

I have certainly learned some specific examples of
how communication by all of us professional linguists
(myself included, as I have used family trees)
is at times ineffective or objectively misleading or even false.

***

Point 2.

Parent and daugher languages can indeed in theory co-exist,
exactly as Steve Long said.

I use here the definition of distinctness of "languages" preferred by
most linguists, including the experts on this list, that is, fuzzily,
"forms of speech which are mutually unintelligible".
Using that definition, Steve was right.
The fact that there is no sharp line is of no concern here,
because we can work with clear cases not the fuzzy boundary.
Since this definition has been advocated by participants on this list
in other contexts, it is certainly not logical to change the definition
away from this one in trying to defeat Steve Long's points.

As noted explicitly also by one other member of this list, Peter Gray
the discussion seems to have degenerated into the circularity
of definition.  I quote Peter Grey's message here because it is so
a propos, and expressed in a balanced manner:

>Does the whole debate reduce to a matter of definition?
>
>Imagine a situation where a group of spakers of a language go and settle
>elsewhere, where substrate and other factors make their language change
>swiftly, while those who stayed at home enjoy a very much slower rate of
>change.    After some years, and political upheaval, we can see a situation
>where the settlers are deemed to be speaking a different language from that
>which they brought with them years before.
>
>I suspect that [one contributor] is saying that the stay-at-homes are also
>speaking a different language, by definition;  while some others are saying
>that if the changes are few enough, it should be defined as the parent
>language.   And of course, both sides are right - which is why there is so
>little understanding on both sides.   The debate is ultimately based simply
>on our definition of what a single language is.

If we now apply the definition to a situation of one mother M and two
dialects A and B, where one dialect A has changed so little in say 100 years
that we have no excuse for saying M and A are different languages,
by our standard definition of different vs. same language,
we must by that definition say they are the same language (and NOT merely
for political reasons, one of a zillion red herrings in this discussion),

but the other dialect has changed enormously, so that we are FAR PAST
the point of having to say M and B are distinct languages, there is virtually
no comprehension at all between them, then in such a situation

Steve Long is correct that a parent M=A
     (A is the "same language" as the parent M)
can co-exist with a daughter B
     (a "different language" from A / M).

Now if we wish to deny this conclusion,
I can only see doing it by changing the definition of "same language",
though that would be cheating,
or by admitting explicitly that in this context we
DO NOT MEAN THE SAME THING as we did in other contexts
by "same language" and "different language".
(In other words, that no matter how small the difference is between
M and A, in this context of discussion we will insist they are not the
same language. Probably because our symbol system, our family tree,
is influencing how we think about reality, is substituting itself for the
reality which it is supposed to faithfully represent!)

It is simply off the point to discuss how common or rare the occurrence is.
The point is, by the usual definitions, it can of course occur.
And surely has occurred numerous times, if we choose our languages
and precise time points precisely.

Here is one response which is off the point:

>Sure.  But your position appeared to be that absence of change in a
>living language was a serious possibility, and that's what I was
>objecting to.

It is off the point because it changes the topic to a straw man.  It uses a
phrase
"absence of change" which is relevant only if it means "absolute absence
of change" (which all of us would regard as vanishingly rare, but which
is then more clearly irrelevant to Steve Long's point), or if it means
"absence of the kind of change which makes it a distinct language",
which then immediately reveals the failure in the logic of the responder,
because we know that one dialect CAN change slowly enough to
produce jut such a situation.

We are not talking about branches on a family tree, in these discussions,
because they are not languages, they are only symbols.  We are talking
about actual languages, and our symbols should be made to conform to reality.
Though we may operate by convention as if there are always great changes
on BOTH branches of any split in a family tree, and a sufficiently
great time depth that daughter and mother are not LIKELY to be the same
language any more, that is not explicit in our family trees, and is even
known to be false in those cases in which our understanding is that one branch
does NOT have significant innovations of all of its descendents in common,
prior to the next split.

Here is another response which is off the point:

>Sorry, but I definitely do not agree, if by this you mean that a speech
>variety indistinguishable from that of Caesar could have coexisted, as a
>mother tongue, alongside a variety of Latin so changed as to be
>incomprehensible to Caesar.

Because the word "indistinguishable" is either intended in an absolute sense,
thus attempting to win the argument by use of circurlar definitional use
of words rather than by substance,
or is not linked to our standard definition of what makes two dialects
"distinct languages", in which case our answer is that the situation
described above could indeed have existed,
a speech variety understandable to Caesar without substantial difficulty,
alongside a speech variety mostly incomprehensible to Caesar.

This Point 2, on which Steve Long is I think correct by simple logic,
is closely related to Point 1, that our traditional Family Trees
do not adequately represent whether significant innovations (those leading
to language splits?) have or have not occurred on particular branches.

We can choose to find positive things in what a member of our list
has contributed, INCLUDING ways to improve what we do,
or we can be defensive, changing the topic of discussion,
changing definitions, or simply stonewalling.  I find all of the latter
quite a waste, aside from being ineffective in the long run.
Failing to improve our tradition when we can simply weakens us.
There is no reason to bite off our own noses in this way.

As an addendum, here again is an instance in which the discussion
was deflected to talk about whether the SAME GROUP OF PEOPLE
was speaking both a parent and a daughter language at the same time!

>From Steve Long:

>What's so difficult about believing "Latin" was a living recognizable
>language at the same time an early Italian language was developing among
>some Latin speakers?

Reply by one list member:

>-- because the very fact of developing "Italian" features means that they're
>shedding the features which make us refer to what they're speaking as
>"Latin".  "Birth of Italian" is the _same thing_ as "death of Latin".

The gap in the logic here is the use of "they" to refer to a single group of
people, where Steve Long's scenario referred to two different groups
of people, explicitly so where he referred to "some" Latin speakers,
by implication of course distinct from "other" Latin speakers.

***

Point 3.

Computer algorithms may have their conclusions partly built into the data.

Steve Long's contributions have led to acknowledgements in the most recent
messages, when the "characters" were actually specified for the first time
within
these discussions, that the "characters" used in the UPenn tree for the IE
languages
include in several cases specifications of innovations vs. retentions,
and thus may build into the data a part of the conclusions, or at least
strong biases
towards them.  That was one of Steve's early assertions, and in that
he has been shown to be at least partly correct.  If this had been
acknowledged
forthrightly from early on, if the "characters" had actually been listed in
the
discussion then, we would have had far fewer words expended, and people
would have become less frustrated with each other.
(That is not to say the judgements of innovations vs. retentions included in
any of those "characters" are wrong, simply that they are already there,
they are not generated independently as the result of the computer algorithm.
It can be a very difficult task to investigate whether there has in effect
been
any distorting bias resulting from this, but the potential is clearly there
on the surface.)

Given this point, it becomes much more difficult to determine what it is that
the
UPenn algorithm has contributed beyond the results of decades of work by
very accomplished traditional experts in Indo-European linguistics.
That is not to say it has not contributed a lot, perhaps primarily in making
judgements more explicit and self-conscious in some contexts, which is
certainly a positive.
But to the extent that assumptions become hidden behind
the magic of the infallible computer,
that side of things can be a negative as well.

The real question which remains for me (judging only by the recent
discussions here and the public presentations I have seen) are whether
the choice of WHICH "characters" to use does not already imply what
conclusions will be drawn.  With a deterministic algorithm
(as opposed to a partly randomizing one), that should of course be the case.
So our attention really needs to be directed to those choices.
Especially since the addition of one "character" changed the family tree
as it affected the positionof Celtic and Italic.

The choice of what data to begin from, here as in almost any scientific field,
can be an even more subtle influence on thinking than
including as data decisions on what is retention and what is innovation.
It is hard to escape, because our intuitive understandings (partly
conclusions)
do of course affect what data we consider to be relevant,
and if we have an assumption we are not aware of,
we are likely to make mistakes in estimating what is relevant data.
True of all human beings, of course even experts in comparative linguistics.

***

Point 4.

Additional information has been used to shape the UPenn IE tree.

Recent responses have acknowledged that geographical location and some other
factors other than the linguistic-structure "characters" have been used
to determine the root of the IE tree.

***

Point 5.

The tenuousness of the Celtic-Italic grouping (whether ultimately right or
wrong)
has been highlighted.

***

Sincerely yours,
Lloyd Anderson
Ecological Linguistics

I am a researcher, son of two researchers,
and a PhD linguist who in parts of my life
does historical and comparative linguistics.
My dissertation was on why generative-grammarians'
accounts of vowel harmony,
while satisfying some desires for neat symmetry,
might be imposing that on data which was much more complex,
which reflected successive layers of historical accretions
and processes of change gone only partly to completion,
and which therefore could not simply be used as evidence
for internal psychological structure (the generative "rules"),
as if the entirety of the phonology and inflectional patterns etc.
of the language were generated synchronically.

So obviously I have the greatest of respect for the abilities
and experties of comparative-historical linguists.
Exactly the same ability to think informs my conclusions
that Steve Long has raised some important points and has been
correct about several of them.