Re: How many forms for orthographic variants?

John, I ran into the same problem in a different system.

You will need a construction that correlates orthographic variants with 
syntactic roles, plurality, tense... It is important here not to linger 
too much in English and rather look at German, Latin .. languages with 
multiple genders, variations in grammatical cases, to maybe build a 
superset of possible variations with a overwrite to 'common', for 
simpler languages.

For efficiency purposes we ended up with an orthographic script variant 
that is language neutral but not alphabet neutral, linked to the 
syntactical variants (rather than duplicating it) and their regional 
transliteration, IPA, .... The syntactical variants then tie into a 
lexeme-like form: noun, verb, 2 forms of adjective, adverb that have a 
pretty stable pattern between each other, if you allow that not all 
concepts have all forms.

These lexeme forms then link to the semantic information.

Have not found anything yet, in any language, that couldn't be 
represented like that.

Hope this helps.

Tom

On 07/11/2014 05:26 AM, John P. McCrae wrote:
> Hi Philipp, all,
>
> I notice in the spec and in the examples in the repository we still 
> have the following:
> :lex_colour a ontolex:LexicalEntry;
>       ontolex:form :form_colour;
>       ontolex:form :form_color.
>
> :form_colour ontolex:writtenRep "colour"@en-GB.
> :form_color ontolex:writtenRep "color"@en-US.
> That is: we use a different form URL for "color"@en-US to 
> "colour"@en-GB. This is IMHO not correct for two reasons
>
>   * It involves duplicating any syntactic annotations on the form
>     (e.g., saying "colours" is a third-person present singular form of
>     the verb and so is "colors")
>   * It makes it impossible to decide which form is canonical, without
>     putting the US form above the rest-of-the-world's form or vica versa.
>
> Is this just an oversight, or does this still need to be discussed?
>
> Regards,
> John
>