Re: Issues with backwards compatibility

OK, I hadn't realised this was quite so controversial. Especially,
considering the common definition of phrase "A short written or spoken
expression.", and the use of the terms "phrasebook" and "phraseological
dictionary" in lexicography to indicate a dictionary with many mulit-word
units, including full sentences (e.g., Brewer's dictionary or any Berlitz
phrasebook).

At any rate, we should not focus on definitions from pertinent fields of
linguistic like syntax (or we may as well use the definition from
statistical machine translation (Koehn et al. 2003)), instead we should
look to relevant fields, i.e., lexicography.

Hartmann's "Dictionary of Lexicography" gives the following definitions:

*multi-word expression: *A phrase consisting of two or more words
functioning as a single LEXEME. The constituents are relatively stable
(FIXED EXPRESSION) and, if used idiomatically, their combined meaning is
more or other than the sum of the parts, e.g. fly-by-night, face the music.

*multi-word lexical unit*: see MULTI-WORD EXPRESSION.

*phrase*: Two or more words combined into a unit which performs a syntactic
function.

I would read "phrase" as the more general definition and in fact closer to
what we wish to represent. That said long discussions about naming are
something I wish to avoid and if the majority think that MWE is more
appropriate I am OK with this.

Regards,
John





On Fri, Jun 13, 2014 at 2:12 PM, Francis Bond <bond@ieee.org> wrote:

> I think phrase is wider than the normal use of MWU.  "a very
> interesting book I picked up last Thursday" is a phrase, as is "dog
> ate two cats with relish", but they would not normally be called
> multi-word units.  Of course, we can define our own meanings, but it
> is good not to strain the standard usage too much.
>
>
>
> On Fri, Jun 13, 2014 at 7:09 PM, John P. McCrae
> <jmccrae@cit-ec.uni-bielefeld.de> wrote:
> > Hi,
> >
> > The term "Phrase" is for me preferable to MultiWordUnit as it is more
> > linguistic, less technical, shorter and the same as the lemon model. I
> would
> > also introduce a disjoint class "Word" as this is useful for saying an
> entry
> > isn't a multi-word unit. If we do I don't think it hurts to include
> "Affix"
> > as well to cover all our bases (that is Phrase for >1 words, Word for =1
> > word and Affix for <1 words).
> >
> > I have no objection to extending the use of confidence to senses (other
> than
> > my existing objections to confidence being too poorly defined at the
> moment
> > ;).
> >
> > I was discussing some use cases that required incompatibility in the
> case of
> > diachronic changes in meaning, but thinking more about, it is quite
> narrow
> > and perhaps should be pushed to LexInfo 3.0 (or whatever we are going to
> do
> > as a more complete but non-standard model).
> >
> > Regards,
> > John
> >
> >
> > On Thu, Jun 12, 2014 at 11:49 PM, Philipp Cimiano
> > <cimiano@cit-ec.uni-bielefeld.de> wrote:
> >>
> >> John, all,
> >>
> >>    a few things. I am in favour of introducing the class "MultiWordUnit"
> >> as a subclass of LexicalEntry, fair enough.
> >>
> >> Concerning the properties "context", "condition" and "incompatibility".
> >>
> >> "context" and "condition" are useful, clearly. But then the property
> >> "confidence" of a Translation should also be there. I see the three
> equally
> >> useful and equaly vague semantically as they could have anyhting as a
> range.
> >>
> >> Concerning "incompatibility": not sure, this seems like one of many
> >> possible properties that could be defined between senses, so it seems
> quite
> >> arbitraty to pick this one out.
> >>
> >> Just my two cents,
> >>
> >> Philipp.
> >>
> >> Am 06.06.14 17:25, schrieb John P. McCrae:
> >>
> >> Hi all,
> >>
> >> Due to the large number of resources using the previous Monnet lemon
> >> vocabulary it seems natural that we should support users who wish to
> >> transition to the W3C OntoLex lemon vocabulary. As such I was looking
> into
> >> the conversion.
> >>
> >> https://www.w3.org/community/ontolex/wiki/Monnet_OntoLex_Compatibility
> >>
> >> There are some areas where the previous model has significant
> differences
> >> that we should consider whether to adopt. (Of course I do not assume
> that
> >> everything in Monnet Lemon should be transferred across but we should
> >> attempt to be able to represent relevant use cases already addressed by
> >> Monnet Lemon).
> >>
> >> From my analysis, there are two main issues that we should still address
> >>
> >> Monnet lemon has more sophisticated description of senses, in
> particular,
> >> mechanisms such as contexts, conditions, definitions, examples and
> >> incompatibility
> >> Monnet lemon allows us to say if a lexical entry is a multi-word
> >> expression, affix or word.
> >>
> >> Any comments on whether we should allow this modelling
> >>
> >> Regards,
> >> John
> >>
> >>
> >>
> >> --
> >>
> >> Prof. Dr. Philipp Cimiano
> >>
> >> Phone: +49 521 106 12249
> >> Fax: +49 521 106 12412
> >> Mail: cimiano@cit-ec.uni-bielefeld.de
> >>
> >> Forschungsbau Intelligente Systeme (FBIIS)
> >> Raum 2.307
> >> Universität Bielefeld
> >> Inspiration 1
> >> 33619 Bielefeld
> >
> >
>
>
>
> --
> Francis Bond <http://www3.ntu.edu.sg/home/fcbond/>
> Division of Linguistics and Multilingual Studies
> Nanyang Technological University
>