Comments on your Old Irish subpages

Fragment of a discussion from User talk:Catsidhe
Jump to navigation Jump to search

Re: lenition of s: thank you. That was exactly the detail I knew I was missing, but could neither remember nor find. I suppose the {{sga-lenite|w}}ord invocation will need to be strongly deprecated, then.

Re: naming: I would be replacing the tables with their *2 version, of which only the masc exists so far. It's complicated to write, and complicated squared to do anything clever.

Re: page titles: I'm ahead of you there. If you look at the *2 template, it shows the lenited/nasalised version, but links to the unmutated form.

Part of the joy is figuring out how to do the calculations of which form to infer if it's not specified for a given declension. This, of course, changes per stem as well as per gender. -o- and -io- currently get a correct table with only ns, as, gs and ds supplied. I'm trying to extract these patterns from Strachan and to a lesser extent Thurneysen, but it's almost as brain-stretching as is writing the templates in the first place.

--Catsidhe (talk) 20:52, 24 May 2012 (UTC)

Catsidhe (talk)20:52, 24 May 2012

But as I've mentioned somewhere else, I'm not sure we should be listing inferred forms. Old Irish is a very unpredictable language, and it may be best if we list only forms that are actually attested, not forms that are inferred or reconstructed.

Also, I see I was mistaken: User talk:Catsidhe/sga-lenite is considered the talk page of User:Catsidhe/sga-lenite, so I could have commented there, but this way the conversation is all in one place.

Angr22:28, 24 May 2012

Unpredictable, at times, and often opaque, but hardly random.

How about, then: if no stem is given ({{sga-decl-m||..., then no inferral is done. Otherwise, inferral per standard tables. -o- and -io- stems are pretty well attested, and Strachan does point out that, eg., Vocative forms aren't provided for any consonantal stem cases because vs=ns and vp=ap for all of them. In any case, an override is fairly simple to provide.

It's also possible (although tricky) to set up an override such that if, say, as is not provided then it is replaced with vs, but if it is provided but blank (ie., ...|as=|..., then it is left blank in the table. (ie., overridden with nothing at all.)

Catsidhe (talk)00:31, 25 May 2012

Those both sound look good ideas.

Angr12:00, 25 May 2012