LINGUIST List 5.1077

Mon 03 Oct 1994

Disc: The Teaching of Syntax (Part 1)

Editor for this issue: <>


  1. Richard Hudson UCL, teaching of syntax
  2. , Teaching of syntax wars
  3. high coup, The teaching of syntax
  4. Tim Stowell, teaching syntax

Message 1: teaching of syntax

Date: Fri, 23 Sep 94 10:13:06 +0teaching of syntax
From: Richard Hudson UCL <>
Subject: teaching of syntax

Maybe the focus on GB in this debate is a red herring, and
there is a more general problem that we all face, regardless
of our particular theoretical persuasions (if any). How do you
get from 0 to the frontiers of theoretical research in one
year (or one term, or whatever the time available may be)?

The problem may affect GB syntax in a particularly acute way
simply because this is the only theory that produces a
constant stream of introductory textbooks, each more up-to-
date than the last (in this department we've gone through
Radford 1, Radford 2, Haegeman 1 and Haegeman 2 in the last 10
years). The assumption underlying these developments is clear:
that an introductory book, for absolute beginners, must be
able to take them up to the frontiers. Why should we assume
that this is either possible or desirable? Pushing the
frontiers of research forwards inevitably lengthens the
distance students have to cover, so the only way to cover that
distance is by reducing the time for sightseeing - and more
seriously, for consolidating the students' understanding by
looking at data.

The problem stems from a structural fact about linguistics in
UK and USA, which is that it's not taught (yet) in schools, or
at least not to a significant extent, so university students
have to start from scratch. Just think how different
everything would look if we took undergraduates who had been
studying language (including grammar) seriously for 10 years
before they reached us. They'd have spent plenty of time
looking at data by then, and would be ready for all the
theories we could tell them about.

Let's lay off GB, and focus on the problem we all share.

Dick Hudson
Dept of Phonetics and Linguistics,
University College London,
Gower Street,
London WC1E 6BT
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Teaching of syntax wars

Date: Fri, 23 Sep 1994 09:57:52 Teaching of syntax wars
From: <>
Subject: Teaching of syntax wars

>From the perspective of someone who considers himself both a
syntactician and a sociolinguist, the problem of this debate seems to
come down to simple stereotyping. In other words, "non-GBers" have a
stereotype about "GBers" (as bad teachers), prompting "GBers" to take
the criticism personally and close ranks. I've had syntax mostly in
GB, and I've had excellent through adequate teachers (thankfully no
bad teachers), with varying amounts of "data" and "theory," but mostly
it's been a balance. Most important, the data/theory ratio does not
correlate with what I learned, or how good the teacher was (if they
can even be separated -- data can be a very theoretical enterprise,
but that's another topic). I know we like to think of ourselves as
dispassionate scientists, but the the fact is we're humans, and we
have all the emotions that humans have (good and bad). Teaching, it
seems to me, is very personal and individual. Generalizations based on
a few postings do not a study make, so until someone undertakes such a
study, please try to learn from each other's teaching experiences
rather than criticizing based on stereotypes.

Scott Fabius Kiesling | "The surest way to corrupt a
 | youth is to instruct him to
Georgetown University | hold in higher esteem those | who think alike than those
1711 Mass. Ave., NW Apt. 521 | who think differently."
Washington, DC 20036-2136 | -Friedrich Nietzsche
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 3: The teaching of syntax

Date: Fri, 23 Sep 1994 10:19:04 The teaching of syntax
From: high coup <>
Subject: The teaching of syntax

I've been following the thread on the teaching of syntax, and it seems
to me that there has not been enough discussion of the *possibility* of
teaching syntax in a theory-neutral way.
The third is to do a mini- history of linguistics. The fourth is to
teach as up-to-date a theory as possible. I have seen four approaches to
teaching introductory syntax. One is a true attempt at being
theory-neutral. It is an almost purely inductive approach. The second
is to teach "old" syntax, e.g., _Aspects_ theory, with lots of individual
transformations, and then go on to a separate advanced course dealing
with one's favorite theory. The rationale here must be something like
the useful fiction of phonemes in an introductory linguistics course when
one "knows better". The third is to do a mini- history of linguistics.
The fourth is to teach as up-to-date a theory as possible.

For me, the problem of a totally theory-neutral approach is the selection
of the data. I believe that a theory, for good or ill, partially
determines which data one examines. Teaching a theory that one knows
isn't used any more is very frustrating for the teacher (or at least
for me). Doing the mini-history of linguistics is much more meaningful
for the teacher than for the students. So I do the fourth option but
bring in as much real data as I can to get the students to think about
analyzing language as a system. I don't mind doing some of the other
things in the context of a lower-division introductory linguistics
course, but in a course focusing on syntax, I think it is necessary
to expose students to fairly current thinking, within whatever theory one

Susan Fischer
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 4: teaching syntax

Date: Fri, 23 Sep 94 09:59 PDT
Subject: teaching syntax

We have all recently received an informal poll on GB syntax teaching
which is apparently leading some people to conclude that an impartial
survey has revealed GB syntax to be badly taught in comparison to other
subjects in our field. But as any expert on polls will tell you, this
is a bit like Rush Limbaugh asking his TV audience whether any of them
think Clinton is a liar and a crook and then, after receiving many
affirmative replies, reporting the poll results as evidence that Clinton
is probably more of a liar and a crook than most other politicians.
My point is that if we want to find out how various subjects are taught,
we don't ask people whether they agree that one subject in particular is
taught badly in some way; we ask a neutral question about ALL subjects and
then see what we find. Anyway, what follows is a more detailed response
to the specific charges raised against GB syntax teaching in the recent
LINGUIST posting...

Most of the complaints voiced against the style of teaching GB in
the recent LINGUIST postings (too much theory, not enough concern
for data, forcing the data to fit the theory) have a familiar ring
to them; they are the same complaints that used to be raised in the
1960's against generative grammar in general. The only old
complaint that didn't resurface was "they only talk about English"
or "they analyze every language to make it look like English".

Perhaps the English-only complaint has been soothed by the GB
research industry on Romance causatives, Romance clitic climbing,
serial verb constructions in various languages, Wh-in-situ in
Chinese/Japanese/Arabic/etc., subject and/or object null pronouns
("pro-drop") in Italian/European Portuguese/Korean/ etc., head-
internal relative clauses in languages X,Y, and Z, experiencer
predicates in Modern Irish, verb raising and V-2 in Dutch and
German, pronoun shift in Irish and Icelandic, Topicalization and
Subjacency islands in Japanese, resumptive pronouns in Yoruba and
Swedish, agreement and stem classes in Bantu, VP fronting in
Amazonian languages, Scrambling in Hindi, Ergativity in Lardil et
al, Clitic left Dislocation in Modern Greek, to name a few.
Apparently, though, none of this "theory" about other languages
involves any "data".

It should be obvious that I don't agree that GB syntax classes
leave out the data. In fact even a cursory glance at virtually any
elementary syntax text will show that it's packed with data. Once
can't help feeling that the real objection is not to the absence of
data analysis but rather to the presence of theory. Furthermore, I
think that more new syntax data has been discovered within the
context of the theoretical "GB syntax" literature than anywhere
else; generally speaking, concern with theory stimulates the
collection and organization of data, rather than impeding it.
However, to be fair to the people whose opinions have been
highlighted in the recent postings, the actual complaint is not
that data are ignored, but rather that beginning students are
taught to understand theory before they are trained to analyze
data. I will therefore direct my comments to this specific

If one believes that Linguistics is a scientific enterprise, it is
hard to understand what kind of science some of our non-GB
colleagues think we should be emulating. How many professors of
elementary physics or biology classes instruct their students to
walk into a lab and start playing with the apparatus in search of
interesting data before they are taught any theory? I don't know of
ANY (natural) science that encourages its beginning students to
analyze data in a purely inductive fashion without any set of
theoretical assumptions to act as a guide, though admittedly some
social science research seems to operate in this mode. It's been
said many times before, but evidently it needs to be said again:
scientific data derive their interest chiefly from the relationship
they bear to scientific theories, and one can't evaluate them in
this respect without first understanding the theory.

The too-much-GB-theory objection has also been expressed in two
different guises: (a) it's fine to teach theory, but one should
also teach pretheoretical data-organizing techniques; (b) it's fine
to teach GB theory but one should be ecumenical and teach other
theories too.

Concerning (a), it's hard to disagree in principle, since it is
obviously important for students to learn how to organize data,
construct paradigms, etc. I do try to teach students how to do
this, though in my experience most good students don't need to be
taught it; they view it as common sense. However, there are
thousands of ways that one could, in principle, organize bodies of
data into different categories, and students can waste a lot of
time and paper extracting generalizations from limited bodies of
data that are true of those data but misleading about the language.
Anyone who has taught introductory linguistics classes has had the
experience of encountering weird answers to problem sets that
happen to work because the data set provided didn't have enough
examples to rule out every conceivable strange alternative
solution; people usually talk about training students to develop a
linguist's intuition about what the right kind of solution is, if
more than one solution is compatible with a certain body of data.
But where does a linguist's intuition come from? I think it comes
from three places: (1) old chestnuts from traditional grammar or
naive versions of information theory, e.g. languages with lots of
(overt) morphology don't need to be so strict about syntactic
restrictions on word order, etc. (possibly true as a vague tendency
but mostly useless as a predictive guide for detailed phenomena);
(2) predictions made by particular linguistic theories that one may
or may not believe; (3) raw intuitions that can't easily be
attributed to (1) or (2). Whenever I teach students about a
particular phenomenon, I usually try to give them a general idea
about (1), which usually doesn't take too much time. Clearly (2)
involves teaching theory (the only issue is WHICH theory or
theories). (3) arguably shouldn't be taught at all.

Concerning (b) -- why teach only GB theory -- the issue is socio-
politically loaded but I think a strong case can still be made for
teaching beginning students a single coherent and at least partly
compatible framework of theoretical assumptions. Most of us teach
syntax courses within the context of a program of instruction that
also includes training in phonetics, phonology, semantics, and
often also historical linguistics, psycholinguistics, and
computational linguistics. At my university (UCLA), there are just
two ten-week graduate courses in syntactic theory that are
required of all graduate students, plus one additional ten-week
course that is required of students who will end up specializing in
syntax or semantics. The syntax literature is VAST, and twenty or
thirty weeks is barely enough time to scratch the surface, even if
the course were to contain nothing but old EST and GB classics. But
graduate students rebel whenever we suggest adding to the list of
required courses, so we are left with our meagre 20 or 30 weeks.

I sometimes get the impression that many people think that they can
keep more or less up-to-date in GB-style syntax by reading Noam
Chomsky's latest paper and browsing through an orthodox textbook
every few years. It ain't so. The fact is that "GB theory" isn't a
theory at all-- it's a research tradition involving a series of
commonly-agreed-upon notational conventions and an accumulated body
of analyses and theories within the context of which new theories
can be expressed. In virtually every domain, there is no single "GB
theory" of X; there are many competing theories of X, fighting it
out in the marketplace of ideas. To be properly educated, a student
must learn not only the "standard" theory about small clauses,
constraints on Wh-movement, or whatever, but also various competing
theories about the same phenomena. The fact that these theories may
be expressed in terms of the same set of notational conventions
does not mean that they are the same theory, or even different
versions of the same theory. They are different theories, period.

Thus, to talk of "GB theory" as a monolithic entity is not much
different from talking of "20th Century Physics theory" or
"Molecular Biology theory" in the same terms. Anyone who considers
this remark imperialistic or arrogant with respect to other
frameworks should have the experience of sitting on an abstract
selection committee for a major linguistics conference sometime. I
have done this a few times, and I was (initially) surprised to
discover that the overwhelming majority of syntax abstracts
submitted are written in the theoretical vocabulary of GB, even
when the conference is held at a university whose faculty is
generally regarded as not being a bastion of orthodoxy. This even
applies to the LSA (based on my one experience back in 1981-82).

When we train graduate students, we have a responsibility to
prepare them to be research scholars in the field who will be
respected and taken seriously by established scholars working
elsewhere. If we spend our time teaching them a little bit about
everything, but not enough about anything to turn them into experts
in some area, then they will not produce publishable results and
they will have a hard time getting a job when they graduate. If
time were not an issue, I'd love to have three or four years of
required syntax courses, which could survey every major framework,
including lots of pre-, post-, non-, and anti-GB stuff. But time IS
an issue, and graduate students have to be allowed to grow up and
join the field. We can't expect them to learn everything we learned
when we were students, or they'll have no time left to learn or
discover stuff that we don't know. So we have to make choices, to
distil from what we know just what we think they need to know to
function in the field. Sometimes this means cutting out work done
in alternative frameworks, though even the most orthodox
departments usually have other courses designed to meet the need of
educating students about this kind of work.

Anyone who has actually gone through a graduate syntax course
sequence can probably remember experiencing a long period of
confusion and frustration, of having to learn reams of facts,
paradigms, analyses, generalizations, rules, principles,
constraints, filters, and so on. Some of it fits together, some of
it doesn't (or seems not to). In short, there's too much to learn
and it's hard to assimilate. Even graduate students who excel in
these courses and love the material they are exposed to are still
babes in the woods at the end of their first year, because there is
so much to learn. If we were to include in entry-level syntax
courses an introduction to every major current notational framework
as well, we would simply compound the problem and probably scare
off everyone. No doubt this leaves some students feeling that
they've been given a distorted picture of the field, but I can't
see a better way of doing it, given our educational mission and the
time constraints we are operating under.

--Tim Stowell (UCLA Lingusitics) (IML7TIMMVS.OAC.UCLA.EDU)
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue