FastSpell (yet another) next round

[mbanon]

Some candidates here (WIP):

• Dzongkha vs Tibetan
• Guarani vs Spanish
• Minangkabau (no hunspell dicts) vs Indonesian
• Occitan vs Catalan vs French
• Shan (no hunspell dicts) vs Burmese
• Turkmen vs Turkish
• Maithili vs Hindi
• Sardinian vs Italian vs Spanish
• Latgalian vs Latvian
• Chhattisgarhi vs Hindi
• Sicilian vs Italian
• Silesian vs Polish
• Papiamento
• Venetian vs Italian
• Quechua vs Spanish
• Manipuri vs Bangla
• Magahi vs Hindi
• Friulian vs Italian
• Acehnese (Arab) vs Arabic
• Limburgian vs Dutch vs Western Frisian
• Scottish Gaelic vs English

[mbanon]

• Dzongkha vs Tibetan: not happening because dictionaries are actually syllabaries

[mbanon]

• Guarani vs Spanish: added hunspell for Guarani,but still a lot of false es results .

Then:

    643 gn
    206 es
     27 pt
     ...

Now:

    656 gn
    193 es
     27 pt
     ...

(out of 1000 sentences from Flores-plus)

[mbanon]

• Occitan vs Spanish vs Catalan: Very nice results. I used the Occitan dictionary that we already had for Catalan, not the linked above one.

Then:

    628 oc
    308 ca
     40 es
     16 fr
      3 ro
      2 pt

Now:

    976 oc
     16 fr
      3 ro
      2 pt

[mbanon]

• Turkmen vs Turkish: Very nice results too. Used TK and TR dictionaries already in the dictionary pack.

Then:

    871 tk
    103 tr
      9 en
      5 et
      2 uz
      ...

Now;

    974 tk
      9 en
      5 et
      2 uz
      ...

[mbanon]

• Maithili vs Hindi: also added.

Then:

    755 hi
    148 mai
     93 bh
      1 mr

Now:

    893 mai
     93 bh
     10 hi
      1 mr

[mbanon]

• Sardinian: Added with Spanish, Catalan, Italian and Romanian. Might add French at some point if needed.

Then:

    252 es
    233 ca
    188 it
    127 ro
     54 fr
     32 oc
     26 pt
     11 tl
      9 sc

Now:

    809 sc
     54 fr
     32 oc
     26 pt
     11 tl

[mbanon]

• Latgalian: Added LT and LV (also in the reversed direction, LV with LTG). Not perfect, but at least it's supported now.

Then:

    801 lv
    134 lt
     20 fi
     17 sl

Now:

    465 ltg
    405 lv
     65 lt
     20 fi

[mbanon]

• Silesian vs Polish: Very cool!

Then:

    996 pl
      1 hr

Now:

    994 szl
      2 pl
      1 hr

[mbanon]

• Papiamento: I added dictionaries from Aruba's Papiamento (Curaçao was also available but seemed derived from Aruba's one), result not perfect but at least Papiamento is supported now.

Then:

    549 es
    194 it
     36 pt
     26 ca
     ...

Now:

    633 pap
    103 es
     36 pt
     26 ca
     24 io
     ...

[mbanon]

• Quechua: Tried with Quechua dictionaries from Ecuador. Not a great improvement. Not adding.

Then:

    779 qu
     97 es
     49 en
     29 pt
    ...

Now:

    781 qu
     95 es
     49 en
     29 pt

[mbanon]

• Magahi: Not perfect, adding it anyway for Magahi support.

Then:

    994 hi
      2 bh
      1 ne

Now:

    542 hi
    452 mag
      2 bh
      1 ne

[mbanon]

• Friulian: Added with Italian and French.

Then:

    609 it
    223 fr
     89 ro
     27 es
     11 pl
     10 eml
...

Now:

    832 fur
     89 ro
     27 es
     11 pl
     10 eml
...