Difference between revisions of "Scottish Gaelic and Irish"

From Apertium
Jump to navigation Jump to search
m (Irish doesn't have synthetic adjectives)
 
(23 intermediate revisions by 2 users not shown)
Line 3: Line 3:
 
==Todo==
 
==Todo==
   
  +
===Irish dictionary===
* Add ability to analyse initial mutations to the monolingual dictionary.
 
  +
::-- I have most of the work done for this -- [[User:Jimregan|Jimregan]]
 
  +
# <s>put the paradigms in 1 entry-per-line format</s>
* Add all closed categories to the monolingual dictionaries.
 
  +
# noun paradigms
::--
 
  +
## some have only one entry -- these are defective? -- e.g. <code>bá__n_m</code>
* Improve the tagger -- write restrictions/constraints, and then retrain.
 
  +
## some have three entries -- defective also? -- e.g. <code>band/ia__n_m</code>
* Perform an intersection on the monolingual dictionaries.
 
  +
# verb paradigms
** We only want stuff in the Irish analyser that we can translate into Scottish Gaelic -- so, in order for a word to be included, it should be in both the Irish monolingual, bilingual and the translation in the Scottish Gaelic monolingual. With the words for which we don't have translations we can just comment them out for now.
 
  +
## sort the entries so that the order makes sense
::-- Count me out on this one; I will suggest using <e i="yes"> etc. instead of xml comments -- [[User:Jimregan|Jimregan]]
 
  +
## <s>is there an imperative p1.sg ???</s>
  +
# adjective paradigms
  +
## some paradigms have more entries than others, e.g. <code>ca/s__adj</code> has 3, and <code>bré/an__adj</code> has 4
  +
# are some proper nouns marked with common noun paradigms instead of proper noun paradigms ?
  +
## find out with <code>cat apertium-ga-gd.ga.dix | grep '<e lm="[A-Z]'</code>
  +
# sort the entries in the <section id="main"> by a) part-of-speech, b) alphabetical order
  +
# i think we're missing possessives and demonstratives, quantifiers and perhaps some definite/indefinite pronouns
  +
  +
==Old todo==
  +
  +
* '''Perform an intersection on the monolingual dictionaries. (Making them consistent)'''
  +
** We only want stuff in the Irish analyser that we can translate into Scottish Gaelic -- so, in order for a word to be included, it should be in both the Irish monolingual, bilingual and the translation in the Scottish Gaelic monolingual. With the words for which we don't have translations we can just comment them out -- or move them to a separate file in <code>dev/</code>
  +
* Add all missing closed categories to the monolingual dictionaries.
 
* Do some fixing of the bilingual dictionary
 
* Do some fixing of the bilingual dictionary
** <s>There are some entries with unknown gender on the Scottish Gaelic side.</s>
 
 
** Some restrictions probably need adding.
 
** Some restrictions probably need adding.
 
** Some conjunctions are marked "cnj" and not subdivided for "cnjcoo", "cnjsub" etc.
 
** Some conjunctions are marked "cnj" and not subdivided for "cnjcoo", "cnjsub" etc.
  +
* Making constraint grammar rules more CG-like
::-- I'll take this one too -- [[User:Jimregan|Jimregan]]
 
 
* Write rules to do initial mutations for generation.
 
* Write rules to do initial mutations for generation.
 
* Write some transfer rules.
 
* Write some transfer rules.
 
** For example to do tenses, number agreement, etc.
 
** For example to do tenses, number agreement, etc.
::-- We can probably take most of this stuff from another language pair and add the consonant etc. stuff later; for the most part, adjective chunks etc. should be the same as those in at least one other pair (I'll scout around for which) -- [[User:Jimregan|Jimregan]]
 
   
==Tagger==
+
==Testing==
   
  +
* [[/Regression tests]]
==Initial mutations==
 
  +
* [[/Pending tests]]
 
As members of the group of Celtic languages, both Scottish Gaelic and Irish exhibit initial consonant mutation. There follows a brief description of how the analysis, disambiguation and generation of this phenomenon is dealt with in the <code>apertium-ga-gd</code> package.
 
 
===Analysis and disambiguation===
 
 
===Generation===
 
 
;Overview
 
 
Generation of initial mutations takes place in two files, where <math>x</math> is the code of the language that is being generated (<code>ga</code> for Irish, <code>gd</code> for Scottish Gaelic).
 
 
* <code>apertium-ga-gd.pre-<math>x</math>.t1x</code> &mdash; Transfer rules which add tags defining the mutation to the beginning of words which should be mutated.
 
* <code>apertium-ga-gd.muta-<math>x</math>.dix</code> &mdash; A post-generation dictionary which takes the tag and the initial letter of the word and outputs the mutated form.
 
 
For example, when translating the phrase "do theach" (your house) from Irish to Scottish Gaelic, the result will be do <u>th</u>aigh (where the initial mutation is marked by an underscore). The output of <code>apertium-transfer</code> will be:
 
 
<pre>
 
^do<det><pos><p2><mf><sg>$ ^taigh<n><m><sg><nom>$
 
</pre>
 
 
This is then passed through <code>apertium-ga-gd.pre-gd</code>, which adds a tag, <code><l1></code> for lenition.
 
 
<pre>
 
^do<det><pos><p2><mf><sg>$ ^<l1>$^taigh<n><m><sg><nom>$
 
</pre>
 
 
The morphological generator then outputs the surface forms of the words, and appends an "alarm" signal to the mutation tag.
 
 
<pre>
 
do ~<l1>taigh
 
</pre>
 
 
Finally, the mutation dictionary, <code>apertium-ga-gd.muta-gd.dix</code>, replaces the string <code>~<l1>t</code> with 'th', which is the lenited form of 't'.
 
 
<pre>
 
do thaigh
 
</pre>
 
 
;apertium-ga-gd.pre-<math>x</math>.t1x
 
 
As mentioned above, the input to this stage is:
 
 
<pre>
 
^do<det><pos><p2><mf><sg>$ ^taigh<n><m><sg><nom>$
 
</pre>
 
 
A simplified (although functioning) rule in this file might look something like:
 
 
<pre>
 
<pattern>
 
<pattern-item n="det"/>
 
<pattern-item n="nom"/>
 
</pattern>
 
<choose>
 
<when> <!-- When the lemma of the determiner is "do",
 
apply lenition to the following noun -->
 
<test>
 
<equal>
 
<clip pos="1" side="tl" part="lem"/>
 
<lit v="do"/>
 
</equal>
 
</test>
 
<out>
 
<lu><clip pos="1" side="tl" part="whole"/></lu>
 
<b/>
 
<lu><lit-tag v="l1"/></lu> <!-- Lenition -->
 
<lu><clip pos="2" side="tl" part="whole"/></lu>
 
</out>
 
</when>
 
</choose>
 
 
</pre>
 
 
And the output will be,
 
 
<pre>
 
^do<det><pos><p2><mf><sg>$ ^<l1>$^taigh<n><m><sg><nom>$
 
</pre>
 
 
;apertium-ga-gd.muta-<math>x</math>.dix
 
 
The input to this stage is:
 
 
<pre>
 
do ~<l1>taigh
 
</pre>
 
 
The "rule", or rather "entry" in the mutation dictionary will look like:
 
 
<pre>
 
<e>
 
<p>
 
<l><a/><s n="l1"/>t</l>
 
<r>th</r>
 
</p>
 
<par n="alphabet"/>
 
</e>
 
</pre>
 
 
Here, the alphabet is defined as a paradigm which for any given input letter, just outputs the letter unchanged. So this basically says,
 
 
:"When we have the alarm symbol '~', followed by a tag indicating lenition followed by a 't' and then any alphabetic character, output 'th' followed by the next character"
 
 
The output of this stage is a correctly mutated phrase,
 
 
<pre>
 
do thaigh
 
</pre>
 
   
 
==See also==
 
==See also==
   
 
* [[Scottish Gaelic]]
 
* [[Scottish Gaelic]]
  +
  +
==Notes==
  +
<references/>
   
 
==External links==
 
==External links==
   
  +
* E. Uí Dhonnchadha and J. Van Genabith (2006) "[http://www.sdjt.si/bib/lrec06/pdf/193_pdf.pdf A Part-of-speech tagger for Irish using Finite-State Morphology and Constraint Grammar Disambiguation]".
 
* [http://en.wikipedia.org/wiki/Differences_between_Scottish_Gaelic_and_Irish Wikipedia: Differences between Scottish Gaelic and Irish]
 
* [http://en.wikipedia.org/wiki/Differences_between_Scottish_Gaelic_and_Irish Wikipedia: Differences between Scottish Gaelic and Irish]
  +
* [http://www.smo.uhi.ac.uk/gaidhlig/ga-ge/coimeas.html Comparison of Irish and Scottish Gaelic]
  +
* [http://www.smo.uhi.ac.uk/gaidhlig/ga-ge/faclair.html Faclair Gàidhlig-Gaeilge]
   
[[Category:Language pairs]]
+
[[Category:Scottish Gaelic and Irish|*]]

Latest revision as of 09:59, 8 May 2011

Todo[edit]

Irish dictionary[edit]

  1. put the paradigms in 1 entry-per-line format
  2. noun paradigms
    1. some have only one entry -- these are defective? -- e.g. bá__n_m
    2. some have three entries -- defective also? -- e.g. band/ia__n_m
  3. verb paradigms
    1. sort the entries so that the order makes sense
    2. is there an imperative p1.sg ???
  4. adjective paradigms
    1. some paradigms have more entries than others, e.g. ca/s__adj has 3, and bré/an__adj has 4
  5. are some proper nouns marked with common noun paradigms instead of proper noun paradigms ?
    1. find out with cat apertium-ga-gd.ga.dix | grep '<e lm="[A-Z]'
  6. sort the entries in the <section id="main"> by a) part-of-speech, b) alphabetical order
  7. i think we're missing possessives and demonstratives, quantifiers and perhaps some definite/indefinite pronouns

Old todo[edit]

  • Perform an intersection on the monolingual dictionaries. (Making them consistent)
    • We only want stuff in the Irish analyser that we can translate into Scottish Gaelic -- so, in order for a word to be included, it should be in both the Irish monolingual, bilingual and the translation in the Scottish Gaelic monolingual. With the words for which we don't have translations we can just comment them out -- or move them to a separate file in dev/
  • Add all missing closed categories to the monolingual dictionaries.
  • Do some fixing of the bilingual dictionary
    • Some restrictions probably need adding.
    • Some conjunctions are marked "cnj" and not subdivided for "cnjcoo", "cnjsub" etc.
  • Making constraint grammar rules more CG-like
  • Write rules to do initial mutations for generation.
  • Write some transfer rules.
    • For example to do tenses, number agreement, etc.

Testing[edit]

See also[edit]

Notes[edit]


External links[edit]