LINGUIST List 5.514

Wed 04 May 1994

Qs: Metaphor, Functionalist problems, Translation, Developmental

Editor for this issue: <>

Directory

1. Lynn Lesueur, metaphor
2. Johanna Rubba, Functionalist problems
3. Hr. Reinke, evaluation of integrated translation systems
4. , developmental linguistics

hello out there.
has anyone been keeping up with the literature on cognitive research on
metaphor comprehension? if so, what have you seen of interest in the
past four years? please send along references.
thanks.
L. Lynn LeSueur
MGH, Harvard Med School, Boston

I would like to start compiling a collection of homework/exam
problems in functionalist/cognitive syntax/semantics. I am wondering if
anyone out there has formulated some such problems and would
be willing to share them with me. I have a few that I have made up and
would be willing to pass them along to anyone interested.

Please direct your answers to:

jrubbaselway.umt.edu

I will post a summary to the list.

Jo Rubba
The University of Montana

There is a recent tendency in machine-aided translation to develop
systems that enable the translator to "recycle" former translation
units. Besides a terminological database these memory-based
translation systems - or integrated translation systems, as I
prefer calling them - contain a database, the so-called
translation memory, that stores translation units and compares
source-language units currently to be translated to the units in
the memory in order to find exact matches or "similar" sequences
and make the former translation available again. This leads to the
question whether and how translation units that are "similar" to a
section currently to be translated are found in the memory. An
even more interesting and fundamental question is "what does
'similarity' of translation sequences mean to the machine and what
does it mean to a human being? And what kinds of 'similarities'
are there in different types of texts?" These are some of the
questions I would like to deal with in my PhD-thesis in order to
find ways for the evaluation of integrated translation
systems.

So far, I tried to analyse the linguistic performance of two
commercial translation memory systems (IBM's "Translation
Manager/2" and Trados' "Translator's Workbench II"). As a first
start, I tried to name some broad types of syntactic/semantic
similarities, such as paradigmatic alterations,
expansion of phrases by further attributes, altered position of
phrases in a sentence, changed position of clauses, changes from
passive voice to active voice etc. I used these types as
categories for the analysis. The major results are contained in a
paper that will be published by Langenscheidt in one of the next
issues of "Lebende Sprachen".

One major topic I would now like to focus on, is the aspect of
"likeness" of (sentence) patterns in technical documentation. This
would enable me to find a more detailed classification of semantic
and syntactic "likeness" and to compare human understanding of
this phenomenon to the performance of the different systems.

At the moment, my major problem is to collect a suitable amount of
machine-readable texts in English, German and French. This is, why
I would like to find out, whether there might be some list-members
who could perhaps help solving this problem.

The kind of texts I am looking for could be roughly described as
follows:

- "follow-up versions" of texts (i. e. different texts of the same
 text type belonging to the documentation of a newer and an older
 version of the same product)
- texts with different structures and/or functions belonging to
 one and the same product (e. g. online-help and user manual of a
 software)
- texts with similar functions belonging to different products of
 the same kind (e. g. a manual for a car produced by company A and
 a manual for a car produced by company B.)

I am basically interested in source-language texts. If I could
also get hold of translations this would help testing the
alignment tools now on the market, which are used to create
translation memory databases from machine-readable source- and
target-language texts.

I know that asking for text material - particularly for translations - might
be quite a problematic thing. But as the material will be used for
scientific purposes only, I think there may be some people around who might
be able to contribute to my
text corpus or might have some ideas where to get further assistance.

In case you also know something about research that has been or is
being done on the alignment of sentences in parallel corpora, I
would be grateful for any kind of information.

I would really very much appreciate your support.

Yours

Uwe Reinke


Please use my private e-mail address for replies !


 ***********************************************
 * *
 * Uwe Reinke *
 * Universitaet des Saarlandes *
 * Fachrichtung 8.6 *
 * D-66041 Saarbruecken *
 * Tel.:++49/681/302-2929 *
 * Fax: ++49/681/302-4440 *
 * E-Mail: reinkerz.uni-sb.de *
 * *
 ***********************************************

Hi,

Does anyone know of any networks, societies or associations for developmental
linguistics?

Chris Li