[comp.edu] Another thread gone tangential

dlindsle@blackbird.afit.af.mil (David T. Lindsley) (12/14/90)

In article <9237:Dec1315:28:3190@kramden.acf.nyu.edu> brnstnd@kramden.acf.nyu.edu (Dan Bernstein) writes:
>In article <1811@blackbird.afit.af.mil> dlindsle@blackbird.afit.af.mil (David T. Lindsley) writes:
>> The widespread irregularity of English verbs makes learning any underlying
>> structure difficult -- more difficult, I'd contend, than in other Western
>> languages.
>
>Hardly. In fact, verb structure is simpler in English than in any other
>language I know of. The verb ``catch'' has just five forms: infinitive
>(to catch), present third person singular (he catches), past (I caught),
>perfect (I have caught), and participle (I am catching).
>
>catch/catches/caught/caught/catching is a *complete* conjugation. What
>other language is so simple? To learn a verb in English you must learn
>just five words.

But there is a notable lack of rules to follow when attempting to generate
the latter four forms from the first (whereas, in Latin, for example,
this is a trivial task for the majority of verbs).

>> In addition, it is difficult to tell gerunds from participles,
>
>A gerund is a noun. A participle is not. That's about as simple as a
>language rule can get.

Simple in theory, yes. But when you get into complex-compound sentences
with multiple subordinate clauses, it's not so easy to tell the
difference -- precisely because there are "just five words".

> [lots of other stuff deleted]

This is getting tiresome.

1. As has already been pointed out, this thread ("the structure or
   lack thereof in English") does not belong in comp.edu.
2. The article I posted (which seems to have started this thread)
   pointed to a lack of structure as *one* cause for a problem
   teaching/learning writing skills.

I was raised bilingually, and in fact spoke another language fluently
language before starting to learn English. Perhaps I am biased as
a result, but it has been my experience/observation that people
who learn English as a second language have less problems with
English grammar than "native-speakers". Consistently.
(I mention this to avoid any accusations of having an axe to grind.
I don't, and in any case, that would be an _ad_hominem_/cirmustantial
fallacy, without any bearing on the argument.)
Now the only reason I can see for this is that English lends itself
less to formal/rigorous definition than (at least some) other
languages do. I have discussed the topic with scholars of language(s),
and I have not found one who disagrees with me. (Not till this thread
started, anyway, and I'm still not sure...)

So I would say that English is unstructured, relatively speaking. And
I do believe that's justified.

>You're talking about these ``problems'' of English as if the subject
>line were ``How to program a natural-language parser.''

If the syntax is (relatively) undefined, I would think it would
necessarily be (relatively) difficult to express oneself clearly.
Thereby making it more difficult to learn to write.
This is *not* intended as a judgment of the English language; it is
merely observation. Observations which have a positive side, I might
add: if I'm correct, that makes English capable of more nuanced
communication than "more structured" languages.

>easy to teach generative parsing to a computer. Sure, linguists have
>only been somewhat successful at formal models of natural languages.
>But people will continue to read. And when I write
>
> [examples deleted]
>
>people will understand me perfectly in each case.

You begin by discussing "formal models of natural languages", then
use semantics to prove your point?


Dave L			dlindsle@blackbird.afit.af.mil
#24601

Opinions. Mine. (Sorry, the words don't come any smaller.)
"If you don't succeed at first, transform your data!"

brnstnd@kramden.acf.nyu.edu (Dan Bernstein) (12/14/90)

In article <1814@blackbird.afit.af.mil> dlindsle@blackbird.afit.af.mil (David T. Lindsley) writes:
> In article <9237:Dec1315:28:3190@kramden.acf.nyu.edu> brnstnd@kramden.acf.nyu.edu (Dan Bernstein) writes:
> >catch/catches/caught/caught/catching is a *complete* conjugation. What
> >other language is so simple? To learn a verb in English you must learn
> >just five words.
> But there is a notable lack of rules to follow when attempting to generate
> the latter four forms from the first (whereas, in Latin, for example,
> this is a trivial task for the majority of verbs).

You deleted my assertion to the contrary. In English it is a trivial
task for the majority of verbs to generate the mere five forms. There
are perhaps a dozen different verb types, each based on the spelling of
the word, to cover almost every case.

In French, a verb has dozens of forms. There are at least two hundred
different types, only loosely correlated with spelling; there are more
exceptions than in English.

There is no objective basis for claiming that French verbs are more
regular than English verbs.

> >> In addition, it is difficult to tell gerunds from participles,
> >A gerund is a noun. A participle is not. That's about as simple as a
> >language rule can get.
> Simple in theory, yes. But when you get into complex-compound sentences
> with multiple subordinate clauses, it's not so easy to tell the
> difference -- precisely because there are "just five words".

No!

You deleted my points about how people parse. It is irrelevant that a
word may have several meanings. A reader will read the most probable
meaning that he's expecting based on the previous text. If that meaning
doesn't fit, he'll back up and try again; this makes the text a bit more
difficult to read.

You keep talking about features of English which really are problems for
computer parsing. They are NOT problems for human readers. Your direct
versus indirect object, for instance: people don't need to decline
``bus'' in order to understand ``catch the bus'' and ``get on the bus''
equally well. Once you expect a direct object, you read a direct object
without any trouble.

  [ native speakers supposedly have the most trouble with English ]
> Now the only reason I can see for this is that English lends itself
> less to formal/rigorous definition than (at least some) other
> languages do. I have discussed the topic with scholars of language(s),
> and I have not found one who disagrees with me. (Not till this thread
> started, anyway, and I'm still not sure...)

Several years of copy editing experience, since you asked.

> So I would say that English is unstructured, relatively speaking. And
> I do believe that's justified.

Not on any objective basis. English is both more regular and more
structured than any other language I know of.

> >You're talking about these ``problems'' of English as if the subject
> >line were ``How to program a natural-language parser.''
> If the syntax is (relatively) undefined, I would think it would
> necessarily be (relatively) difficult to express oneself clearly.
> Thereby making it more difficult to learn to write.

Kee-rist. Anyone who learns to write by writing and getting criticism
will do perfectly well in any language. And just because it's hard to
program a generative parser doesn't mean that the syntax is vague.

> You begin by discussing "formal models of natural languages", then
> use semantics to prove your point?

Yes. No linguist in the last fifty years has tried to formally model
natural language without taking semantics into account.

Now you've really convinced me that you're a computer scientist who's
dabbled in computer parsing and had a tainted view of English ever
since.

---Dan

robert@cs.arizona.edu (Robert J. Drabek) (12/15/90)

Bernstein and Lindsley write (for my point it doesn't matter who is who
here) :
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> catch/catches/caught/caught/catching is a *complete* conjugation. What
> other language is so simple? To learn a verb in English you must learn
> just five words.

JUST 5?  I have to learn 5 words to speak ONE meaning?  Yikes!
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> But there is a notable lack of rules to follow when attempting to generate
> the latter four forms from the first (whereas, in Latin, for example,
> this is a trivial task for the majority of verbs).
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> You deleted my assertion to the contrary. In English it is a trivial
> task for the majority of verbs to generate the mere five forms. There
> are perhaps a dozen different verb types, each based on the spelling of
> the word, to cover almost every case.

A trivial task for most.  But those dozen (he said that, I didn't) do
make life difficult for older second-language learner.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> In French, a verb has dozens of forms. There are at least two hundred
> different types, only loosely correlated with spelling; there are more
> exceptions than in English.

I never considered French to qualify as a natural language. :-)
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
  Not on any objective basis. English is both more regular and more
  structured than any other language I know of.

Let's see.  Why don't we try Chinese, Mandarin specifically.  There is
only one form of all verbs.  Conjugation is totally absent.

Also, Spanish is much more more regular than English, in case you only
want to consider Western languages.

--
Robert J. Drabek                            robert@cs.Arizona.EDU
Department of Computer Science              uunet!arizona!robert
The University of Arizona                   602 621 4326
Tucson, AZ  85721

swsh@ellis.uchicago.edu (Janet M. Swisher) (12/15/90)

In article <261@coatimundi.cs.arizona.edu> robert@cs.arizona.edu
(Robert J. Drabek) writes:
>Bernstein and Lindsley write (for my point it doesn't matter who is who
>here) :
>- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
>> catch/catches/caught/caught/catching is a *complete* conjugation. What
>> other language is so simple? To learn a verb in English you must learn
>> just five words.
>
>JUST 5?  I have to learn 5 words to speak ONE meaning?  Yikes!

Not one meaning--five related meanings.  Don't be so obtuse, silly.
'I caught' does NOT mean the same thing as 'I am catching.'

Doesn't the Subject line of this thread have any meaning to you?
Follow-ups to sci.lang, and this time I mean it.  Sci.lang folks,
please forgive me.

1/2 :-)



--
Janet Swisher			Internet: swsh@midway.uchicago.edu	
University of Chicago		Phone: (312) 702-7608
Academic and Public Computing	P-mail: 1155 E. 60th St. Chicago IL 60637, USA
"This whole world's wild at heart and weird on top."  -- Lula

staff@cadlab.sublink.ORG (Alex Martelli) (12/15/90)

brnstnd@kramden.acf.nyu.edu (Dan Bernstein) writes:
	...
>Yes. No linguist in the last fifty years has tried to formally model
>natural language without taking semantics into account.

You know Dan, you keep making absolute assertions, and they are SO
grating.  Leonard Bloomfield only died in 1949 - that's barely over
THIRTY years, NOT fifty - and I would like you to now try and defend
the position that HE "took semantics into account"!!!  Indeed, just
to avoid injecting my personal bias I would like to quote from the
Encyclopaedia Britannica - "Bloomsfield's followers pushed even
further the attempt to develop methods of linguistic analysis that
were not based on meaning.  One of the most characteristic features
of post-Bloomfieldian American structuralism, then, was its almost
complete neglect of semantics".  I guess you'll have to grasp at
that "almost", won't you?  Or would you prefer to argue that the
structuralists, with their "discovery procedures" to be mechanically
applied to texts, were not trying "to formally model natural language"?
Please note that post-Bloomfieldian structuralism takes us fully
into the fifties, and that's definitely NOT outside "the last fifty
years".
Not that I would *AGREE* with Bloomfield and followers, mind you, but
baldly asserting, in practice, that they did not exist, and indeed
*dominate* American linguistics for a long period WELL within your
chosen 50-years limit, is tantamount to Orwellian history-rewriting.
-- 
Alex Martelli - CAD.LAB s.p.a., v. Stalingrado 45, Bologna, Italia
Email: (work:) staff@cadlab.sublink.org, (home:) alex@am.sublink.org
Phone: (work:) ++39 (51) 371099, (home:) ++39 (51) 250434; 
Fax: ++39 (51) 366964 (work only), Fidonet: 332/401.3 (home only).

brnstnd@kramden.acf.nyu.edu (Dan Bernstein) (12/20/90)

In article <553@cadlab.sublink.ORG> staff@cadlab.sublink.ORG (Alex Martelli) writes:
> brnstnd@kramden.acf.nyu.edu (Dan Bernstein) writes:
> >Yes. No linguist in the last fifty years has tried to formally model
> >natural language without taking semantics into account.
> Leonard Bloomfield only died in 1949 - that's barely over
> THIRTY years, NOT fifty - and I would like you to now try and defend
> the position that HE "took semantics into account"!!!

It's tempting to argue that he did---after all, his structuralism shared
every essential aspect of ``features'' in generative grammar, and those
are what I mean by semantics. But you're right. Insert either ``sane''
or ``respected'' into my original statement wherever it makes you happy.

---Dan