Quasicode for transfer from sme to smj

Overall plan
Goal: Create sme-smj lexical pairs, with sme as a starting point:
A. Cognates
1. Divide the sme lexicon in homogenous groups
2. Translate each group to quasi-smj
3. Spell-check or analyse the result
4. Group entries into analysable and non-analysable
5. Go manually through the result

B. Non-cognates
1. Always record non-cognate pairs when going through the cognate list
2. go through frequency lists and mark non-cognates
3. go through parallel text and look for non-cognates

C. The dictionary
1. Make sme<tab>smj lists
2. Make real dict5ionary entries
3. Convert the sme<tab>smj lists into xml entries

D. machine translation


Detailed code for transfer:

According to etymology:
Recent loans transfer ok, now for checking
Incorporated Sámi phon

According to stem type:
Even-syllable stems
Odd-syllable stems
Contracted stems
Closed classes


Even-syllable stems
 
Odd-syllable stems
 
Contracted stems
 
Closed classes


Report, Oct 4th

Tested out odd-syll verbs:

  466  cat ../sme/src/verb-sme-lex.txt | egrep '(ALIST|ARVIL|ARVVASJ|BALAT|BEAGASJ|BONJAT|BORGGIST|BOTNJAS|BUOVVAL|COASKKIT|CUOHCIT|CUORPMAST|GEAGAT|HALIID|JORGGIID|LASSAN|MUITAL|MUITTASJ|SUOTNJAL|VUORDIL)' | grep -v '^\!' | cut -d":" -f1 > smeoddverbs
  468  cat smeoddverbs | lookup -flags mbTT -utf8 ej.fst | grep -v '^$' | l

Report, Oct 8th

Tested out contract verbs:

  688  cat ~/gt/sme/src/verb-sme-lex.txt | grep LOAN | egrep '(DOHPPE|GILLE|CIRRO|BASSO|BORGE|MUITA|FERTE) ' | grep -v '^\!' | tr -d "^0'" | cut -d":" -f1 | cut -d" " -f1 > smecontrloanverbs
  689  cat ~/gt/sme/src/verb-sme-lex.txt | grep -v LOAN | egrep '(DOHPPE|GILLE|CIRRO|BASSO|BORGE|MUITA|FERTE) ' | grep -v '^\!' | tr -d "^0'" | cut -d":" -f1 | cut -d" " -f1 > smecontrverbs

Made a working version of the converted smeoddverbs file:

cat smeoddverbs7 | tr -d "0\'" | lookup -flags mbTT -utf8 ej.fst | grep -v '^$' > src/inc-smesmj-alloddverbs.txt

Next steps:
1. split the inc-smesmj-alloddverbs.txt file in two: one that we can analyse and one that we cannot
2. add them to the lexicon, manually. The recognised ones fast, the other ones with care
3. rewrite the conversion script
4. convert and add the contracted verbs
3. then even-syll verbs
4. then Nouns
5. then djectives


How to get the words:

ejodd.fst: sme-to-smj
jok.fst: smj-to-analysis
jerr.fst: smj