LINGUIST List 5.1272

Fri 11 Nov 1994

Qs: French Creole Aspect, Proper names, Mon-khmer, Parsing

Editor for this issue: <>

Directory

1. Vincent DeCaen, French Creole Aspect
2. Dounce Valdivia Lourdes-IIF, Proper names & quantifiers
3. Vincent DeCaen, mon-khmer tense-aspect
4. Ted Pedersen, Right to Left parsing?

nov 8,

It is not uncommon for a language to encode nonperfective aspect by
means of prepositions. The parade example in this regard is Welsh:
progressive "in V" and perfect "after V" (it is more typical to have
"at" vs "in").

What is unexpected is the use of (a)pe in French creoles from apres
"after" for the progressive (perfect with some variation of "fin").
Does anyone have a historical explanation for apres as progressive
rather than perfect?

To whom it may concern:

We would like to get some information about the recent literature in
lingusticis and psycholinguistics on the topic of proper names and
quantifiers. Could you please refer us to any?

Lourdes Valdivia
Cog.Sci. Program,
Universidad Nacional Autonoma de Mexico

nov 8,

In isolating languages such as Burmese or Mandarin, tense-aspect is
perfectively read, while the progressive must be marked by other
means: Burmese auxiliary verb nei "stay/live" or Mandarin (zheng) zai
'(just) at'. These constructions are difficult to find in a grammar,
let alone finding a description.
I can't get an indication about the progressive construction in these
In Vietnamese and Cambodian, we find much the same state of affairs as
in Burmese, Chinese, etc. But the grammars are difficult to use, and

languages. If anyone can help explain the tense-aspect systems of
Mon-Khmer languages, I would greatly appreciate it.

Most parsers seem to parse left to right. That makes a good deal of
sense, especially if the language is read left to right. It also
allows for grammar rules to be expressed left to right as well, as
shown below:

s --> np, vp.
np --> d, n.
vp --> v, np.

'the man saw the boy' can be parsed left to right with this grammar.

Suppose that I want to parse 'the man saw the boy' starting from the
right and going left. I could reverse the grammar rules and reverse
the sentence and use the same parser. But what I'd rather do is leave
the grammar rules and the sentence alone and have the parser
automatically proceed right to left.

My reason for wanting to do this is an attempt to parse sentences with
words that are not in the lexicon. It seems that if you could approach
the parsing of a sentence with an unknown word from both left to right
and right to left it would be possible to converge upon a syntactic
category for an unknown word faster than if you just parse left to
right.

I'd be very interested in hearing suggestions about how a right to
left parser could be implemented (esp in Prolog) or seeing an example
if you've done that yourself.

I'd also be curious if anyone has approached the unknown word problem
in parsing by using the sort of "bi-directional" approach I sketch
above.

Regards
Ted

PS. This idea started after working with Link Grammar in the context
of dealing with unknown words. The advantage of using the Link Grammar
is that it is very easy to parse left to right and right to left. In
fact, the Link Parser is designed to work that way. I'd like to be
able to use some of the same ideas in more standard linguistic

* Ted Pedersen pedersenseas.smu.edu *
* Department of Computer Science and Engineering, *