- This page is work in progress used to formulate my own thoughts on the matter. If I ever reach a format I am pleased with, and this format convinces myself of the sanity of the proposition, I will post a note on IBP.
Random ramblings on specifying the forms of words
While it has been held as self-evident that the various forms a given word can take must be entered at the SynTrans level, it has at least not reached my knowledge how this should be done. It is also stated that the various forms (each of which is an Expression) of a word should all link to the same Definition, creating DefinedMeanings for each and every form. The devs may have it all figured out, in which case these random ramblings are good for nothing, but in case there still are some question marks, my thoughts might help highlight some factors, either to disregard them or implement them.
Link Expressions in Inflectional Groups
I had an idea that apart from joining with Definitions to form DefinedMeanings, Expressions could also be members of what I for lack of a better term think of as Inflectional Groups; i.e., a set of word forms each and every one of which has a specified role within the group. A group may minimally have only a single member (f ex a group for adverbs that cannot be inflected); or it may have a great many members (like f ex a French verb).
A given Expression must of course be able to participate in a given Inflectional Group several times, to handle f ex plurals identical to singulars etc.
The Inflectional Groups would provide the primary morphological information of the SynTrans; what that information would be would be language specific, but part of speech would most likely be a common feature. The language maintainer for each language would specify which Inflectional Groups are necessary for the Language and which members (and their specification) each group has. Some care should be taken; when it is feasible, to provide specifications that can be used to match forms between languages, such as deciding on f ex the preferred order of partial tags such as feminine, singular definite etc. As I see it now, the specifications of the Inflectional Groups would not bother about the actual shape of the Expressions entering them as members (word forms), but only with stating which members they have.
For each Inflectional Group one word form should also be chosen as the base form of the set, which will be used to represent the group when no equivalent form is available. Which form is this base form will of course be set by the language maintainer, and there is no need for it to be the same form (if there even was such a same form) between languages.
For Swedish there would be a Countable noun inflectional group, with the following members:
- Indefinite singular -- base form
- Indefinite singular genitive
- Definite singular
- Definite singular genitive
- Indefinite plural
- Indefinite plural genitive
- Definite plural
- Definite plural genitive
Note:I have chosen not to specify the case of the non-genitive forms, since they are in fact used in all other cases but genitive. Marking them as nominative (which is often done) is imo misleading, only in the pronominal system does Swedish have a nominitive case.
For English the corresponding Countable noun inflectional group would most likely only have the following members:
- Singular -- base form
- Singular genitive
- Plural genitive
Note:Here it could of course be discussed if using f ex Indefinite singular would better capture the essence of the English word form.
To not lose the ease with which we currently enter new SynTrans-expressions, on entering a new Expression in a SynTrans table, it would be associated with a dummy Inflectional (one-member) Group, which then also could be used to locate which SynTrans:es that has is as yet unspecified.
For the user of OmegaWiki, this would simply be presented as another heading under each DM that charts all the various forms of the SynTrans.
For editors, when editing a given Expression, apart from having available all the DM:s the Expression is linked to, there would also be a section for all Inflectional Groups. Under each DM, the section for Inflection would simply be a drop-down list to choose among the Inflectional Groups already specified for that Expression. In the section for specifying Inflectional Groups, the various types of Inflectional Groups for the language in question as specified by the language maintainer would be available, also as a drop down list. After choosing which type of Inflectional Group, the editor would also specify which member position (word form) of the inflectional group he wishes to link the Expression too. The editor can of course choose to add the Expression as another form to an existing Inflectional Group rather than creating a new Inflectional Group.
Entering Expressions into the SynTrans table
Now there would be several options on how to link the information from the Inflectional Groups into the DefinedMeanings.
The option that is closest to the current state of affairs is to create fields in the SynTrans table to link to a given Inflectional Group, and also a field specifying which member of the table that is requested. This is however imo a bit clumsy.
Another option is to have a separate table for word forms, that is all the members of the Inflectional Groups, and in fact store links to these word forms in the SynTrans table instead. But now we are tampering with the holy DefinedMeaning... *smile* A DefinedMeaning would then be seen as a a combination of a word form and a definition, where the word form would be an Expression with specified role within an Inflectional Group, and a definition would be a paraphrase of the semantic content common for all the members of that Inflectional Group.
While we are tampering, lets go all the way.... It would be possible to redefine a DefinedMeaning not as a combination of an Expression and a Definition, but as a combination of an Inflectional Group and a Definition.... However, this I think will take us too far from the current state of affairs and into consequences that at least I cannot foresee today. So lets not walk that path.
- I don't think that's a problem. In each inflectional group there is a dictionary form: infinitive for verbs in English and other languages, nominative for nouns in English and other languages, Shūshikei for Japanese verbs and so on. This dictionary form would be the representative of the inflection group and thus the "expression" of the DM. --Mkill 16:05, 6 August 2006 (CEST)