italian mor

Brian MacWhinney macw at mac.com
Wed Jul 21 23:10:04 UTC 2004


Dear Anna,

   Sorry about the delay in replying.  Looks like you are getting a good 
handle on some of the
details here.  The treatment of [gen] is a special case, based on a 
request that we received in 2002 from
several Spanish researchers who wanted to make sure that this feature 
was available for analysis in the output.  In general features such as 
[gen masc] are just used internally.  If they do not associate to a
particular separate morpheme, they will not surface in the output, so 
we made an exception here and forced
&MASC and &FEM into the output.  Since Spanish has a separate analytic 
suffix for number, we don't have
to worry about it there.  However, Italian is not so analytic, so we 
may need to do the same for [num sq] and
[num pl] in Italian, right?  I can ask Leonid to do this, if you wish, 
but let's be sure which affixes and pronouns should be involved and how 
they should end up.
    Regarding your second question, the general answer regarding 
features like [c b] is that they are added as
"blocking" features to filter out incomplete forms.  The blocking is 
implemented in the "all-end" rule that
comes at the end of the c rules.  If the [c b] feature is still in the 
form, it means the the required additional suffixes are not present and 
that the particular allomorph generated by the a rules has to be 
filtered out.

Hope this helps.

Buona fortuna,

Brian MacW


On Jul 20, 2004, at 6:50 AM, Anna Gudmundson wrote:

>
> Dear Brian and others,
>
> I have been trying to make changes in the italian mor rules, adding
> or changing some features for nouns, articles, verbs and pronouns.
> But there are some output, concerning for example the articles,
> that I can't understand.
>
> I've split the articles in two groups - det:def and det:indef, but I
> don't think that has anything to do with my problem, since I've also
> changed the names in the c-rules as follow:
>
> lexicon
> gli {[scat det:def] [gen masc] [num sg]}
> l' {[scat det:def] [num sg]}
>
> c-rules
> RULENAME: bare-start
> CTYPE: START
>  if
>  NEXTCAT = [scat OR det:def det:indef prepart conj part prep pro:
> rel pro:int pro:indef \
>    pro:pers pro:dem pro:refl vpart vimp vinf vpres vsub vpret vpas]
>  then
>  RESULTCAT = NEXTCAT
>  RULEPACKAGE = {}
>
> Before I changed the names of the articles the category  "art" was
> declared  in the c-rule, shouldn't that have been "det"?
>
> This is the result of the parsing of "gli" and "l'":
>
> mor (:h help)> gli
> Debugging file: Superbob:Applications:CLANX:lib:mor:italian:
> debug.cdc
> parse 1:
> 	lex info: {[scat pro:pers] [pers 3] [cas dat] [gen masc]}
> 	morphemes: gli
> 	translation:
> parse 2:
> 	lex info: {[scat det:def] [gen masc] [num sg]}
> 	morphemes: gli
> 	translation:
>
> Result: pro:pers|gli&MASC^det:def|gli&MASC
>
> mor (:h help)> l'
> Debugging file: Superbob:Applications:CLANX:lib:mor:italian:
> debug.cdc
> parse 1:
> 	lex info: {[scat pro:pers] [num sg] [pers 3] [cas acc]}
> 	morphemes: l'
> 	translation:
> parse 2:
> 	lex info: {[scat det:def] [num sg]}
> 	morphemes: l'
> 	translation:
>
> Result: pro:pers|l'^det:def|l'
>
> My question is, why is "gen" the only feature shown in the output,
> and not, for example, "num"? I cannot find any place, neither in the
> a-rules nor in the c-rules that handles this phenomena.
>
> Then I have another question concerning the feature [c b], what
> does it do? It is added in the a-rules, never matched but deleted in
> the c-rules as follow:
>
> RULENAME:	noun-stems
> LEX-ENTRY:  masc-o
> LEXSURF = $Xo
> LEXCAT = [scat n]  	
> ALLO:
> ALLOSURF = $X		% masc sg
> ALLOCAT = LEXCAT, ADD [gen masc], ADD [num sg], ADD [c b]
>
> RULENAME: n-gennum
> CTYPE: -
>  if
>  STARTCAT = [scat OR n ndim adj adv pro:dem pro:poss vpart]
>  NEXTCAT = [scat nsfx]
>  MATCHCAT [gen]
>  MATCHCAT [num]
>  then
>  RESULTCAT = STARTCAT, NEXTCAT [gen], NEXTCAT [num], DEL [c]
>  RULEPACKAGE = {}
>
>
> Maybe my questions are a bit unclear, but I would really appreciate
> some help in this matter.
>
> Many thanks in advance
>
> Anna :-)
>
> --
> Anna Gudmundson
> Institutionen för franska och italienska
> Stockholms universitet
> +46(8)163588
> --
>
>



More information about the Chibolts mailing list