lexical relations
Martin Hosken
Looking at the use of relations, I notice that a relation may be a child of an entry or of a sense and even of a variant (which I assume would be tantamount to an entry level relation). It seems to me that the set of sense relations is a different set from the set of entry relations and should probably be modelled using a different range set. What do people think? Should I add an entry-relations range and say all entry and variant level relations are taken from that range and lexical-relations are only used for sense relations?
TIA,
Martin
Ronald Moe
English)?
Ron Moe
Checked by AVG - www.avg.com
Version: 8.0.238 / Virus Database: 270.12.8/2086 - Release Date: 04/29/09
06:37:00
Martin Hosken
> Can you give an example of an entry level lexical relation (preferably from
> English)?
GB,
Martin
John Hatton
from
>> English)?
>sub entry is the obvious one that springs to mind. Notice that a relation
is anything that relates one entry to another >for any particular reason
however implementation specific.
To me (and to WeSay and FLEx), "sub entry" is a publication issue, not a
lexical relation itself. It is very much one of those things that different
publications may want handled different. So we want to model, say, a
stem-to-compound relationship, leaving as a separate issue, outside of LIFT,
how that affects the printed or electronic publications. Same for "minor
entry".
John Hatton
Ronald Moe
entry level lexical relation. (There was method in my madness.) Entry level
stuff is all related to the form of the lexeme. (There's some craziness in
that statement, but I'll let it go for now.) "Subentries" are a type of
entry in a published dictionary normally used to handle complex forms
(compounds, derivatives, and multi-word expressions (MWE)). Complex forms
are related to their roots by a form:form relation, not a sense:sense
relation. Since lexical relations are sense relations, it doesn't make any
sense to use the same list of lexical relations in the entry level Cross
Reference field and the sense level Lexical Relations field. In FLEx we
don't use lexical relations to capture complex forms. We use a different
system because we are trying to capture the relationship between one form
and another. I personally don't know what the Cross Reference field would be
used for. I believe Chuck Grimes gave some examples in an email, but I can't
find them. The MDF manual has examples for how to use the \cf field, but
they are all examples of complex forms, variants, or sense level lexical
relations.
06:01:00
Martin Hosken
> entry level lexical relation. (There was method in my madness.) Entry level
> stuff is all related to the form of the lexeme. (There's some craziness in
> that statement, but I'll let it go for now.) "Subentries" are a type of
> entry in a published dictionary normally used to handle complex forms
> (compounds, derivatives, and multi-word expressions (MWE)). Complex forms
> are related to their roots by a form:form relation, not a sense:sense
> relation. Since lexical relations are sense relations, it doesn't make any
> sense to use the same list of lexical relations in the entry level Cross
> Reference field and the sense level Lexical Relations field.
That's what I was trying to saying (but probably very badly :)) And while subentries are a publication issue, people are going to want to store them in their lexical databases, at some point, whether we like it or not.
>In FLEx we
> don't use lexical relations to capture complex forms. We use a different
> system because we are trying to capture the relationship between one form
> and another.
But it's still a relation. That it's not a 'lexical' relation, is what you are telling me. Which helps me in getting my nomenclature more correct.
> find them. The MDF manual has examples for how to use the \cf field, but
> they are all examples of complex forms, variants, or sense level lexical
> relations.
So, given I define an entry-relation range, which should be in the default standardised set? All contributions welcomed and considered.
Yours,
Martin
John Hatton
standardised set? All contributions welcomed and considered.
A reasonable set could be had by looking at all the entry-to-entry
relationships offered by FLEx 6 (now in alpha).
JH
Ronald Moe
Here is a list of standard entry-entry relations (not to be confused with
sense level lexical relations). There are two problems. (1) Some of these
relations are combined into more generalized relations. FLEx does this. (2)
Some of the relations are one-to-many relations. For instance
'blood-and-guts' is a single compound composed of three roots and an
inflectional suffix. So I'll give the list as a hierarchy.
1 component-complex form
1.1 component-derivative
1.1.1 root-derivative (e.g. nomin:nominalization)
1.1.2 stem-derivative (e.g. nominalize:nominalization, both of which are
from the root 'nomin' and the simpler stem 'nominal')
1.1.3 derivational affix-derivative (e.g. -ation:nominalization, the inverse
is derivative-derivational affixes, note the plural of affix, and note that
the derivative is nomin-al-iz-ation, so this is a one-to-many relation)
1.2 component-compound
1.2.1 root-compound (e.g. lap:laptop, the inverse of this is
'compound-roots', note the plural of root, so it is a one-to-many relation)
1.2.2 root-derived compound (e.g. break:broken-hearted; derived compounds
can be any combination of roots and affixes, e.g. root + root + derivational
affix, or root + derivational affix + root + derivational affix)
1.2.3 stem-derived compound (e.g. broken:broken-hearted)
1.2.4 inflected form-compound (e.g. has:has-been; warning--this is difficult
to handle; it is usually just handled as a root-compound, have:has-been)
1.3 component-MWE (multi-word expression, e.g. idiom, phrasal verb, saying)
1.3.1 root-MWE (e.g. bright:brighten up)
1.3.2 word-MWE (e.g. brighten:brighten up)
2 basic form-variant
2.1 basic form-dialect variant (e.g. toward (Am):towards (Brit))
2.2 basic form-spelling variant (e.g. minuscule/miniscule)
2.3 basic form-dialect and spelling variant (e.g. center:centre)
2.4 basic form-register variant (e.g. cannot:can't)
2.5 basic form-inflectional variant (e.g. leaped:lept; warning--this is
difficult to handle)
2.6 basic form-free variant (a catch-all term for variants that are not
easily classified)
3 basic form-irregularly inflected form
3.1 root-irregularly inflected form (e.g. have:had)
3.2 stem-irregularly inflected form (e.g. understand:understood)
There are also two other kinds of relations that are not usually handled by
entry:entry relations: (1) allomorphs (e.g. feel:fel(t), in-:im-), (2)
inflection (e.g. stop:stopped). However in rare situations a dictionary
might handle them by setting up separate entries for them.
Ron Moe
-----Original Message-----
From: LexiconInter...@googlegroups.com
[mailto:LexiconInter...@googlegroups.com] On Behalf Of Martin
Hosken
Sent: Thursday, April 30, 2009 3:01 PM
To: LexiconInter...@googlegroups.com
Subject: [LIFT] Re: lexical relations
Martin Hosken
> I was wondering if I was misunderstanding your nomenclature. :)
GB,
Martin