Re: Intermediate forms

Dear Max, thanks for your reply.

On 25/11/24 16:36, Max Ionov wrote:
> Dear Cristiano,
>
> You raise a very good point, this kind of intermediate forms are 
> important to model and it is not restricted to historical linguistics. 
> And it is indeed something that was discussed in the context of the 
> upcoming Ontolex Morph module, with regard to representing 
> intermediate stages for morphophonological transformations (which is 
> what you were referring to previously). We ultimately decided that it 
> was out of scope for the Morph module but maybe later this will be 
> dealt with in a separate module devoted to automata and transformations.
Yes, concerning this is feel that the morph:Rule and morph:Replacement 
are suitable to be generalized to encompass generic phonological rules, 
like the ours. However, we must keep in mind that these phenomena are 
not necessarily "regular".
> For now I think the solution you suggested originally — to use a 
> user-defined superclass of an ontolex:Form is perfectly valid. 
> Alternatively, maybe an ontolex:Observable class from the upcoming 
> FrAC module 
> <https://ontolex.github.io/frequency-attestation-corpus-information/#observations> could 
> be used? It is an abstraction over any ontolex entity that can be 
> observed (including forms).
> It might sound counter intuitive, since these forms are hypothetical, 
> but if they are observed in any source (e.g. a reconstruction), I 
> think they /are/ Observables.

Honestly, I can't say anything about that. However, being Observable or 
not it is orthogonal to the problem of how to represent these forms.

CL

>
> Best,
> Max
>
> On Mon, 25 Nov 2024 at 11:46, Cristiano Longo 
> <cristianolongo@opendatahacklab.org> wrote:
>
>     Thank you Gilles, I'm very convinced that this discussion is not
>     restricted to historical lingustics. I identified some
>     peculiarities of these intermediate forms which distinguish them
>     from ontolex forms:
>
>     - they are not bounded to any lexical entry (or, at least, to one
>     belonging to an hypothetical language)
>
>     - they must have almost one written representation and
>
>     - almost one associated phonetic representation.
>
>     CL
>
>     On 25/11/24 10:16, Gilles Sérasset wrote:
>>     Hi all,
>>
>>     Very interesting discussion.
>>
>>     I am not a specialist in any way, hence maybe a naive question,
>>     but in which way are such hypothetical forms related to lexical
>>     entries in reconstructed language ?
>>
>>     They may be of different nature, but share the “hypothetical”
>>     feature, hence are there any similarities that could be used to
>>     treat them similarly.
>>
>>     It also remind me of non lexicalized forms in derivational
>>     morphology when 2 derivations are used to model the derivation
>>     process, eg. Verb -> adj -> adv where the adjective form is not
>>     lexicalized and never attested. In the process modelling it is
>>     often “computed” and represented (with a star prefix).
>>
>>     Maybe these could inspire a similar modelling.
>>
>>     Regards,
>>
>>     Gilles,
>>
>>>     On 24 Nov 2024, at 11:46, Cristiano Longo
>>>     <cristianolongo@opendatahacklab.org>
>>>     <mailto:cristianolongo@opendatahacklab.org> wrote:
>>>
>>>     Dear Fahad Khan, thanks for your observations which deserve
>>>     careful considerations. In the meanwhile,
>>>
>>>     at first glance, I observe that of course etymologies (in the
>>>     sense of lemonEty) are just hypotheses,
>>>
>>>     but stating that a lexical expression is a ontolex:Form is an
>>>     assertion with a precise meaning. In other words, etymologies
>>>     are hypothetical derivations grounded on well attested lexical
>>>     expression in some language. Instead, our case is quite
>>>     different as our intermediate forms are properly hypotetical.
>>>     This is clarified by observing that a source expression (which
>>>     of course is a form) can be turned into the corresponding one in
>>>     the recipient language through more than one derivation.
>>>
>>>     In the example we have two derivations from patrem to padre:
>>>
>>>     patrem -> padrem -> padre, and
>>>
>>>     patrem -> patre -> padre.
>>>
>>>     For these reason, I think that asserting that "padrem" or
>>>     "patre" was lexical expression of some intermediate language is
>>>     quite hazardous.
>>>
>>>     CL
>>>
>>>     On 22/11/24 17:22, Fahad Khan wrote:
>>>>     Dear Cristiano,
>>>>     As far as I'm aware an intermediate form is an unattested form
>>>>     that is hypothesized by linguists on the basis of (usually
>>>>     well-attested) linguistic rules; as such it is usually prefixed
>>>>     with an asterisk (e.g., /*patrem/). But the hypothesis *is*
>>>>     that it was used by speakers at a certain point in the
>>>>     evolution of a word, and therefore did belong to a certain
>>>>     historical stage of a language. In which case, I don't
>>>>     understand why you couldn't use Form, or at least create a
>>>>     subclass of Form for asterisked forms?
>>>>     Cheers
>>>>     Fahad
>>>>
>>>>     Il giorno mer 20 nov 2024 alle ore 12:49 Cristiano Longo
>>>>     <cristianolongo@opendatahacklab.org> ha scritto:
>>>>
>>>>         Good morning all. In my last work I faced with strings
>>>>         that, in my
>>>>         opinion, cannot be modelled using ontolex:Form, as they are
>>>>         just
>>>>         "intermediate forms" which does not belong to any language.
>>>>
>>>>         An example is reported in Figure 2 at
>>>>         https://ceur-ws.org/Vol-3809/paper2.pdf. Here the latin
>>>>         word "patrem"
>>>>         changes to an intermediate form "padrem" through lenition,
>>>>         and finally
>>>>         becomes the italian word "padre".
>>>>
>>>>         However, the notion of intermediate forms was previously
>>>>         introduced in
>>>>         the areas concerning phonology and morfology, as reported
>>>>         in [1].
>>>>
>>>>         To deal with such intermediate forms I introduced a new
>>>>         superclass of
>>>>         ontolex:Form (i.e., LanguageObject). However, I'm not
>>>>         really sure that
>>>>         this design choice is correct. Of course, intermediate
>>>>         forms are not morphs.
>>>>
>>>>         I wonder if there are other works where these kind of
>>>>         strings have been
>>>>         modelled in OWL.
>>>>
>>>>         Any suggestion and hint is wellcome,
>>>>
>>>>         thanks in advance,
>>>>
>>>>         CL
>>>>
>>>>         [1] A. Hurskainen, K. Koskenniemi, T. Pirinen, L. Antonsen,
>>>>         E. Axelson,
>>>>         E. Bick, B. Gaup, S. Hardwick,
>>>>         K. Hiovain, F. Karlsson, K. Lindén, I. Listenmaa, I.
>>>>         Mikkelsen, S.
>>>>         Moshagen, A. Ranta, J. Rueter,
>>>>         D. Swanson, T. Trosterud, L. Wiechetek, Rule-Based Language
>>>>         Technology,
>>>>         2023.
>>>>
>>>>
>>

Received on Wednesday, 11 December 2024 07:16:40 UTC