LINGUIST List 7.592

Sun Apr 21 1996

Disc: Grammatical Gender

Editor for this issue: Ann Dizdar <dizdartam2000.tamu.edu>

Directory

1. Joel M. Hoffman, 7.590, Disc: Grammatical Gender
2. Richard DeArmond, Re: 7.573, Disc: Grammatical Gender
3. Alexis Manaster Ramer, Re: 7.590, Disc: Grammatical Gender
4. "Douglas Moore", Disc: Theory of Gender
5. Robert Beard, Re: 7.590, Disc: Grammatical Gender
6. Rachel Lagunoff, Re: 7.587, Sum: Grammatical gender

>> I am not sure who suggetsed that gender is arbitrary, but
>> that is plainly untrue.
>
>I don't think so. Gender _is_ arbitrary. If you disagree, please let us
>know how you account for the distribution of gender among the names of
>plants, animals, and inanimate objects. All you can possibly say is that

I would go one step further, noting that even among words for
"person," gender seems somewhat arbitrary. The examples from Romance
of "person" (which is fem.) have already been noted. In Biblical
Hebrew, there are two words for person: iS and nefeS (S="sh"); the
former is masc., the latter fem.

In most languages that I'm familiar with, the labels masc. and fem.
are actually somewhat misleading. What we really have is category A
and category B (and sometimes category C). Often category A is the
default, with B used for exceptions. Thus, A is used for people when
their sex is unknown or immaterial.

But the cultures in which many of these languages were and are used
often give grossly unequal roles to men and women, and sometimes
equate "men" with "default people," unfairly relegating women to
deviants from the norm. In these cultures, not surprisingly, men fall
into category A (default) and women into category B (exception). But,
it seems to me, to fault the language for reflecting the culture,
rather than faulting the culture, is misguided anger.

-Joel Hoffman
(joelexc.com)

Message 2: Re: 7.573, Disc: Grammatical Gender

>Date: Tue, 16 Apr 1996 22:19:56 BST
>From: billsoftdom.com (Bill Holowacz)
>Subject: Disc: Grammatical Gender
>
>
> At the risk of taking a very simplistic position on the
>grammatical gender discussion, I'd like to repeat something which I
>think everyone heard at least one in their linguistic studies:
>grammatical gender is ARBITRARY. The terminology which descibes
>language is ARBITRARY.
> There is nothing as beautiful as language. Instead of undoing
>what has taken thousands of evolution to produce in the name of
>"political correctness," why not change the terminology? Instead of
>the labels of masculin, feminin and neuter, why not group a,b and c,
>or 1,2,and 3. I really can't see how changing the "gender" of a few
>nouns will substantially better of position of women in society. Why
>don't you all direct your energy into something concrete and useful.
>Bill
>
>billsoftdom.com
>------------------------------------------------------------------------

 Yes, we choosing 'a', 'b', and 'c' would be arbitrary and
politically correct in English. But, I ask, are cultures with overt
grammatical gender agreement getting into political correctness? A
Spanish speaking friend of mine has no intention of trying to
eliminate gender agreement in order to be politically correct in
Spanish.

 Suppose we define these genders, which are grammatical as
opposed to semantic, as F, M, and N, following a suggestion by Bob
Beard (Journal of Slavic Linguistics Vol 3). And we retain Masc and
Fem as grammatical genders that directly correspond to natural sex. I
turns out that most Fem nouns are F, most masc nouns are M, but -Fem,
-Masc nouns can be M, F, or N. The corresponding relationships are
not arbitrary. Only the -Masc and -Fem ones are, though there is a
certain amount of predictabiblity, even if not logical--abstract nouns
tend to F is some languages. We don't want to lose the ability to
capture these redundancies. F, M, and N aren't really totally
arbitrary; they are a cover up for the innate sexism in these kinds of
languages.

Dick.

Message 3: Re: 7.590, Disc: Grammatical Gender

I am glad that Stefan Kaufmann like my friend Hartmut's contribution,
to this discussion, and it pains me he liked mine so little. But the
simple fact is that Geisel 'hostage' is not neuter and we were talking
about neuters. It is not even feminine, which is the other class we
touched on. So it is not relevant to the discussion. My claim was
that German does not allow neuter nouns, except derived ones (
diminutives, compounds whose head is neuter, and the like) to denote
subsets of the set of adult human beings which contain men and women
equally. I did say I was willing to contemplate the possibility of
one or two individual exceptions but did not know of any, and I still
do not. Hartmut's Opfer 'victim' comes closest, but I now realize I
should not have accepted that as a counterexample either, because I am
talking quite emphatically about nouns denoting (proper) subsets of
adult humans, hence, nouns like names of professions, ranks, age
groups, ethnic groups, people from particular localities, people with
a particular name, etc. Thus all names of nationalities and
professions and so on, as far as I can think, are masculine or
feminine, but not neuter. And it inconceivable that it should be
otherwise--for a German speaker, and that is why no one has proposed
saying *das Ingenieur, except my new friend Sean Witty, who is
presumably not a German speaker.
 
Nouns like Opfer are a horse of a different color, because you cannot
partition the set of humans into victims and non-victims (recent
psychobabble notwithstanding). Everybody is a victim of something,
whether of the slap on the behind shortly after birth, the shock of
birth itself, the aging process, etc., etc. The serious point behind
my somewhat frivolous examples is that nouns like victim are entirely
different semantically from nouns like engineer or Spaniard or
colonel. Nouns like Opfer also very rarely are neuter, maybe this the
only one in fact, but nouns like engineer etc. never seem to be--in
German that is.
 
The question of feminine nouns is a little different because there are
some semantically epicene feminines in German, like Person, Figur,
etc., but it is worth while noting that they too overwhelmingly or
maybe even completely fall into a very distinctive semantic class,
which I find it a bit hard to characterize but which is also quite
distinct from nouns denoting professions, ethnic groups, etc.
 
To conclude, i do not see what this whole ruckus is about anyway.
What I said, which is that, in case of nouns denoting human beings,
gender is not arbitrary is what is common knowledge. That is
precisely why masculine is called masculine, etc. Otherwise they
would be nouns class A, etc. Even if there are exceptions, that does
not change the basic fact, and surely the fact that exceptions are not
ALWAYS sufficient to totally throw out a generalization is one of the
half- dozen or so good ideas of Chomsky's which I would think all
linguists by now accept. The case before us is I think such a one:
even if there are occasional exceptions, they are either (a)
unproductive and hence tell us nothing about the synchronic grammar of
the language, or (b) confined to certain special classes, such as all
diminutives being neuter in German, which means they are not really
counterexamples to anything but rather examples of a minor
subregularity. And I still know of no clear counterexample anyway.
 
Alexis MR

On Tue, 16 Apr: billsoftdom.com (Bill Holowacz) wrote:

 "At the risk of taking a very simplistic position on the
 grammatical gender discussion, I'd like to repeat something which I
 think everyone heard at least one in their linguistic studies:
 grammatical gender is ARBITRARY. The terminology which descibes
 language is ARBITRARY. There is nothing as beautiful as
 language...."

Everyone hearing something stated at least once in their linguistic
studies doesn't make that something right. That a process as profound
and widespread as grammatical gender should be simply understood as a
consequence of arbitrariness has never seemed to me as a very
satisfactory kind of theory. One could just as well describe
grammatical gender as MYSTERIOUS. This terminology has the merit of
being a bit closer to the truth - notably that we simply don't
understand this phenomenon. At one time, the laws of physics could
also have been correctly seen as MYSTERIOUS, and even "beautiful."
However, to have described them as ARBITRARY would have been an error.


Since no one has ever proven the hypothesis that the grammatical laws
of gender are arbitrary, whatever being arbitrary may mean, I would
like to take this opportunity to propose a totally different
hypothesis. I propose that linguistic gender is FUNDAMENTAL. It is
reflected in the surface structure of some languages as grammatical
gender but I propose that it is also present in all languages from the
base up. This is my assertion. Such an assertion is no less scientific
than describing the phenomenon as being ARBITRARY.

Now, whilst the argument for the ARBITRARY hypothesis is very thin
indeed -- not much more than an admission of ignorance -- I would like
to take this opportunity to put together the elements of a story that
may provoke some second thoughts on this question. 

Some First Thoughts
===================

My first face to face contact with the importance of grammatical
gender was when I lived in France and was obliged to become fluent in
French. French has possibly the most pronounced form of grammatical
gender of any language. First of all, it has no neuter. Secondly,
unlike any of the other Latin languages, the gender construct is
rather universal in scope and even applies to the proper names of
countries. Most other languages merely refer to the names of countries
as always having neuter or, at most, masculine gender. French's
insistence in gender typing even country names has not always been
without difficulty. Canada was one of the hardest cases and remained a
problem for some time. It has been reported that about two thirds of
the native speakers were using it as a feminine noun and the other
third were using it as masculine. Maybe it was the numbers that
countered or maybe it was Voltaire's unequivocal usage of Canada in
the feminine, but at any rate, the final acceptance by the Academy was
that the noun was indeed feminine and Canada officially became "la
Canada." 

To some, this might be seen as just another collaboration of the
social consensus theory where grammatical gender is just an arbitrary
grammatical convention voted on by the masses, and in the case of
France, rubber stamped by the Academy. However, there remains the
nagging question of what factors were at work to make Canada appear as
"more feminine" than masculine to its native speakers and literary
greats.

If the French language strikes one by its pronounced bi-polar 
grammatical structure heavily based on gender, one can also notice
this same tendency in the cultural domain. Whilst the anglophone will
tend to be happy with a concept if he can find a name or label for it,
the natural French cultural tendency is to express a fundamental
concept in terms of its opposition with another. We see this in
structural linguistics, which finds its natural terrain in a country
like France. The famous opposition 'la langue/le langage' is a good
example and is notoriously difficult to translate well into English.
In philosophy, another example is 'la connaissance/le savoir.' In both
these examples and many others, the opposition expresses a meaning,
but also flows across grammatical gender lines in the form of a
feminine and a masculine noun. This may start one thinking that gender
just might be a little more than simply the affair of a strictly
syntactic construct.

Now all of the above is quite anecdotal and merely has the status of
subjective opinion. People may even object to these opinions. This is
of little importance. What does matter is whether or not one can
arrive at any sort of methodological handle on the gender construct.

I've spent quit a lot of time and effort on this gender problem over
recent years. My first concerted effort was in the guise of my MANA
computer language which was marketed by my ill fated software company
Gender Systems Pty. Ltd. MANA was a multi-paradigm language
integrating the three basic programming paradigms: the logical,
functional and procedural. I naively coined the logical paradigm as
being of feminine type, the procedural masculine and the functional as
neuter. In retrospect, this exercise was more a declaration of intent
rather than any sort of result which could formally characterise the
role of gender in language.

My next project was my "Metaphysics of the Computer" book. If this
work has any value, it is more likely to be for its ironic humour -
particularly on the gender question - than for any scientific value.
In this exploratory book, gender was posed as some kind of allusive
but universal, fundamental systemic organisational principle. This I
believe to be correct, but there was no rigorous formulation of the
construct in the book as I did not understand it properly at that
time.

It was only fairly recently that I finally believe to have discovered
what I claim to be the essential characterisation of the gender
construct. I am presently writing this up, but will take time off to
attempt to explain some aspects of it here.

Some Most Outrageous Postulates
=============================

First of all, what I have to say may not immediately appear to be of a
linguistic nature. Rather, I start by considering the object of
language rather than language itself. As for language, I make no a
priori distinctions between natural languages, the more fundamental
computer languages, and the language of Nature - the genetic code. In
fact, what I am after is a _generic_ language - one which is not
problem oriented in any way, a language for all seasons, a language
which is applicable to any context but which in no way whatsoever
privileges any particular context. 

A fundamental postulate is that the genetic code is such a language. I
postulate that the genetic code is a generic language. Moreover, I
postulate that the generic language concept when posed in its most
fundamental form, becomes a problem to which to which the solution is
absolutely unique. It may thus be that Nature already has tumbled upon
this solution. From my point of view, I see it as our task to also
arrive at the solution of this riddle by a process of reverse
engineering, starting from first principles. One must reverse engineer
generic language which may indeed be synonymous with reverse
engineering the genetic code.

In passing, in order to see where this all fits into the big picture,
I complete what is now fast becoming a most outrageous story by adding
that the genetic code may only be one instance of generic language and
that the exact same principles may be applicable to the formal
language necessary to describe, prescribe, unify, and explain even the
most fundamental laws of physics. In other words, I argue that the
genetic code -- if it is truly generic -- should also be applicable to
the inanimate as well as to the animate. Generic language is the
mother of all languages, so to speak.

As far as natural languages are concerned, and also the more 
fundamental of the computer languages, I claim them to be context
specific languages and hence non generic. Nevertheless, the natural
languages would appear to be much more generic than present day
artificial, computer languages, and hence closer to the ultimate
generic language that I'm inferring. The final postulate here is that
natural languages are based on the same organisational principles as
generic language but are also subject to additional imperatives such
as expressability and the specificity of the natural problem
environment in which they evolved and are evolving. Concerning
expressability, it is clear that a spoken language explicitly based on
a four letter alphabet, such as appears to be the case for the genetic
code - leaves a lot to be desired. In addition, it would suffer from
being excessively more verbose than strictly necessary to serve its
immediate purposes. However, it could very well be that the same
fundamental organisational principles underlying the truly generic
version remain the same in the non generic, domain specific natural
and computer languages as to the generic case.

Now, I realise that these may seem like very outrageous postulates.
Underpinning it all, is the postulate that all of these linguistic
phenomenon, be they in the area of natural language, certain computer 
languages or the language of the genetic code, all are or can be based
on a common underlying set of organisational principles. In a
nutshell, what I am saying is the very antithesis of the popular
competing theory - that claiming the inherent arbitrariness of it all
- de Saussure's claim concerning the arbitrariness of signs at the
sub-morphemic level, extending right through to the arbitrariness of
grammatical structure, including grammatical gender. The position I'm
developing goes completely contrary to all of this.

My position is that it is the arbitrariness theory which is the truly
outrageous postulate. My schema, introduced above, is directly anti-
symmetric to it. In opposition to a flat earth theory I propose a
round earth theory. What appears as outrageous in one historic
conjuncture can become its opposite in another and vice versa. 

Consequently, it may be worth spending a few minutes at least in
contemplating the feasibility of what I am saying and perhaps just
start wondering whether it could turn out to lead to something quite
significant. Admittedly, my story appears an outrageous one and not
everyone's cup of tea...yet. 

The Gender Construct
====================
Now to the gender construct. We start by considering the objects of
language. The language concerned is to be of generic vocation and so
we must consider objects from the most generic point of view possible.
In other words, our reasoning must not depend on any subjectively
determined contingent factors whatsoever.

Objects have properties of some kind which make one object different
from the other and ultimately distinguishable from each other. What we
are interested in are generic objects. What exactly is formally meant
by a generic object will not be developed here, but it suffices to say
that generic objects must have generic properties. The question is,
"What is the most generic property of all that is necessary, adequate
and sufficient to guarantee that at least two objects are different
from each other?" 

Now all of the present day sciences, including mathematics, are based
on the premise that objects are rendered different dues to a
difference in their attributes. Two objects are defined to really be
the one object if their attributes - this includes spatio-temporal
co-ordinates - are identical. They are different if and only if they
differ in their attributes. This has the consequence of assimilating
difference to distinguishability. Two objects are different if they
are distinguishable (by attribute comparison).

The construct that I present here, goes against this popular 
assumption. I define difference is such a way that it is a quite a
seperate notion to distinguishablity. The idea is that by starting
with a more fundamental notion of difference, one which doesn't assume
distinguishability, the distinguishablity problem becomes a tractable
problem to be solved. How do we distinguish? That is the problem. If
difference and distinguishability are assumed to be one and the same,
the problem can't even be posed, let alone solved. 

Thus, in a nutshell, what is the definition a difference between two
objects that doesn't even assume distinguishability? What is the most
elementary difference possible? Let me answer this question as
follows.

Consider one entity which has for its sole specificity that it 
possesses an attribute of a certain kind. Now consider a second entity
which is none other than this attribute. The specificity of the
attribute is none other than it is an attribute and that it is the
only one possessed by that entity. In other words, we start with two
entities - one _has_ an attribute, and the other entity _is_ that
attribute. (A very Stoic approach, this one)

I will say that these two entities are of different _type_. The entity
which _has_ the attribute, I will say is of type feminine, and the
entity which _is_ this attribute is of type masculine. Two entities
which differ in this way, will be said to differ by _gender_. 

Given this definition we can easily see that two entities which differ
only by gender - one being of feminine type and the other of masculine
type- are in fact indistinguishable. Distinguishability relies on
attribute comparisons, and this is impossible here, as there is only
one attribute in play - one is and the other has it. Nevertheless, two
entities that differ by gender are clearly different. I call this kind
of difference - a _generic_ difference relation.

One can see that this kind of typing is quite unlike the approaches of
Russell or Church. Gender is an absolutely _relative_ type, where the
type of one entity is determined uniquely by its relativity to the
other. All typing used in our present forms of logic, mathematics and
science are of this nature. They are all based on an absolute typing
system where difference and distinguishability are assumed to be
synonymous (only quantum mechanics has some exceptions to this rule) 

Gender Can Become Tractable
=====================
In computer languages that employ pointers and references there is the
notion that each entity has two sides two it. It can be treated as a
value or a placeholder for a value. These are some times called right
and left values (rvalue/lvalues) respectively. This is a concrete
example of a relative type of the gender kind. Gender is the most
elementary and "generic" form of this "value" and "placeholder" kind
of semantics. I point this out to provide some indication that this is
not all just a word game. The gender construct is potentially quite
tractable, even though a very difficult problem to handle. 

It is a shame that I have to rely on such specialised examples to
illustrate my point. I guess that all I can say on this matter is that
a practical, concrete experience on implementing modern computer
languages is the best way of getting a tractable and concrete grasp on
this apparently very slippery gender construct. In computer languages,
like C and C++, the placeholder entities are indicated and typed by an
anonymous "wildcard" token '*' whilst the value entities seem to be
generally typed by a token written as 'void' Complex expressions can
be built from these two "feminine" and "masculine" tokens, even though
not all combinations make sense in these very primitive languages. In
my proposed generic language, all combinations are permissible and the
whole language is built from these two token primitives. (they have a
more vast form of semantics than the computer languages mentioned) The
structural linguist Jacobson would probably have liked all that -
everything is built from binary oppositions from the ground up.

The Four Elementary Genders
=====================
I use the '*' and 'void" tokens from computer languages as merely
illustrative. If, instead I use a'#' and 'M' notion, I get the four
elementary combinations #M, M#, MM and ##. These are the first
determined forms of gender relations. Thus an alternative to a pure
M/# based system, we get the a kind more determined feminine #M,
masculine M# and "neuter" MM. The fourth combination of two anonymous
wildcard tokens '##' is so non determined that it is hard to put a
name to it. 

These four binary combinations make up the four "letter" of my 
proposed generic language. The postulate is that they correspond to
the G,A,T and C "letters" that are known to make up the generic code.
I am presently working on obtaining some evidence for this by trying
to predict, from first principles, the known elementary properties of
the genetic code (predicting stop codons and the so called degeneracy
in the coding of amino acids)

Well, there is a sketch of my most outrageous story concerning the
nature and role of gender in generic language. I extend the postulate
to say that this structure is what underlies also the natural
languages. It is going to take some time and effort for the story to
become a convincing one, but at least it can claim to provide what
might be the only present alternative explanation of gender than that
based on the arbitrariness hypothesis. Another part of my story is
that all of this boils down to a revamping of ancient Stoic logic -
but that is another story again.

Final Remarks
==========
The above, if correct, and if the gender construct I mention is 
synonymous with that appearing in the grammatical surface structure of
some languages, it does give some intuitive insights to the role of
grammatical gender.

Intuitively (only), nouns of masculine gender behave a little like
transitive verbs and the feminine like intransitive verbs. For
example, In French one usually says
 Il fait la bicyclette
and not
 Il fait le velo
The latter, though permissible appears bordering on slang and has an
implicit familiarity and would have some kind of extra situational
nuance. Remember that, according to my theory, gender is a relative
type based on difference and not on distinction. There is no ultimate
distinction between these two phrase but there is always a very subtle
difference. That;s why it appears "arbitrary." The same applies to
 J'aime mon petit velo. 
Which I think is more direct and neutral than
 J'aime ma petite bicyclette.
Once again the latter version has an extra subtle nuance, but 
this time in the femine, that is missing in the first "more 
natural" case. There is no distinction between the two 
concepts expressed, but there is a difference.

Food for thought. Gender, I believe, is very subtle construct -
the most subtle in fact.

The PC people should tread with extreme care. They may be unconscious
participants in an historic farce of quite Quixotic proportions.

Douglas Moore
=======================================================
Dr Douglas Moore mailto: dougleopard.lat.oz.au 
http://www.cs.latrobe.edu.au/~doug/index.html

Message 5: Re: 7.590, Disc: Grammatical Gender

 The discussion of Gender would benefit from an updating of
terminology. 'Gender', as it has been used in the discussion up to
now has carried three meanings:

 1. Natural gender, as found in nouns referring to sexed
objects like Russian _student_ and _studentka_,

 2. Declension (Noun) class which refers to the lexical class
to which all declinable nouns belong,

 3. Agreement class, which even indeclinable like Russian
_kofe_ (M[as]-agreement) and _taksi_ (N[eu]-agreement) possess.

All three must be distinguished because the number (count) of Natural
genders does not correspond to the number of declension classes and
the Declension and Agreement classes are asymmetrical (Declension I
has M- and N-agreement, Declension II, M- and F-agreement, etc.).

 Confusion arises over the fact the Declension classes are
often used to express Natural gender, as in Russian _student_ and
_studentka_. However, this relationship is and arbitrary like all
grammatical relationships; other grammatical considerations may
override it, such as the Neutralization of all German Diminutives.
Moreover, it is complicated by the marked-unmarked relationship, which
derives from the fact that binary semantic categories are doubled at
the grammatical level, so that in Slavic, for example, nouns express
four different mixtures of 'feminity' and 'masculinity':

[+Feminine, -Masculine] sestra 'sister' (cannot be masculinized)
[-Feminine, +Masculine] brat 'brother' (cannot be femininized)
[+Feminine, +Masculine] student 'student' (can be femininized:
studentka)
[-Feminine, -Masculine] lisica 'fox' (grammatically F-agreement
and Declension II but the
reference is to male & female foxes, i.e. generic)

The so-called 'common gender' nouns belong to the fourth class, too. 

 Unless these correlations are wholly fortuitous, then not
only are the relations between Declension class and Agreement, on the
one hand, and individual nouns, on the other, is certainly arbitrary.
While the relation of Natural gender to a noun is not arbitrary, the
relation of Natural gender to both Declension classes and Agreement
functions is. All these relations are accounted for if Declension
classes and Natural gender are lexical categories and Agreement a
syntactic (inflectional) one. See 'The Gender-Animacy Hypothesis'
_Journal of Slavic Linguistics_ 3.59-96 for details.

 While this is not wholly to the point of the discussion,
whether we should tinker with the language in order to make it reflect
appropriate male-female relations in the "gender" system, the
experience of the Russians after the 1917 Revolution might be. The
same issues were raised by the (socialist) femininists of the
time. The originally decided that the stature of the feminine forms
should be raised to a grammatical stature equal to that of the
masculine forms, i.e. the women's magazines themselves used, and
encouraged others to use words like 'professorsha', 'doktorsha',
'biolozhka' when referring to women. They even added to catalog of
feminine forms so that a male biology major was a 'biolog' and a
female major was a 'biolozhka'. This attempt was stillborn.
Apparently, the disrespect attached to the feminine forms of the
professions was so deeply ingrained in both male and female
consciousness that the hopelessness of the situation was immediately
apparent. On the other hand, there was nothing to do with the
functions of Declension class, Agreement, and Natural Gender in
Russian. Result: these terms are used identically in 1996 as they
were in 1916.

- Bob

- ---------------------------------------------------------
Robert Beard Bucknell University
Russian & Linguistics Programs Lewisburg, PA 17837
rbeardbucknell.edu 717-524-1336
Russian Program http://www.bucknell.edu/departments/russian
Morphology on Internet http://www.bucknell.edu/~rbeard
- ---------------------------------------------------------

Message 6: Re: 7.587, Sum: Grammatical gender

Some comments (and a query) on Alan R. King's comments on English singular
'they':

>(17) [ARK]
>It seems clear that the use of "singular they" is not an invention of
>feminists (although possibly some feminists might wish to adopt or favour
>this options among those existing), and also that it is not peculiar to
>America.

Yes!

> I distinctly remember noticing as a young boy (over thirty years
>ago) in northern England that my even younger brother, perhaps six years old
>at the time, sometimes used "they" to refer to a single individual when not
>knowing their (!) sex. . . .

>The trouble is, it is doubful whether this mechanism is available
>(in spontaneous English at any rate) when the speaker manifestly knows the
>sex of the referent, e.g. (with "they" coreferent with "friend"):
>
> ?I told a friend of mine and they said they're interested.
>
>and certainly it isn't where the speaker and hearer both clearly know:
>
> *I told Lucy and they said they're interested.

While sentences like the first one, when the speaker has some specific
individual in mind, are rare (and perhaps unacceptable to some
speakers), they do exist (I've documented naturally-occurring
examples). The context can be when the speaker wants to conceal the
indentity of the person, or when the sex of the person is irrelevant.
Here's a real example that a fellow graduate student uttered to me one
day:

 "Someone told me it was going to rain today, and I believed them."

It is true that singular 'they' is impossible with name antecedents,
even when the name could be attributed to a person of either sex.
Thus, the following is out, even if the speaker wants to conceal the
sex of the person referred to:

 *Chris told me they would be here, but they haven't showed up.

QUERY: IF ANYONE FINDS SUCH SENTENCES ACCEPTABLE, OR HAS READ OR HEARD
INSTANCES OF SINGULAR 'THEY' WITH NAME ANTECEDENTS, PLEASE LET ME
KNOW! Thank you!

Rachel Lagunoff
lagunoffucla.edu