Same letter, different codes
Some letters that look identical in Arabic-script text have different codes. The yeh (ی) and kaf (ک) typed on a Persian keyboard look the same as ي and ك on an Arabic keyboard but are different characters for search. If they are not matched, readers cannot find the entry they are looking for.
Hamza, alef and diacritics
- Reducing alef variants such as أ إ آ ا to one form for search
- Ignoring diacritics (fatha, kasra, damma, shadda) in search
- Removing the elongation mark (tatweel, ـ)
- Matching ending variants such as teh marbuta (ة) and heh (ه)
Transliteration differences
Ottoman Turkish and Persian words are written in Latin script under different transliteration systems: long vowels such as â, î and û, ayn and hamza marks, and dotted letters such as ḥ, ṣ and ẓ. Readers often leave these marks out, so search should match the unmarked spelling to the marked form.
The consonant-skeleton approach
Because short vowels are usually not written in Arabic script, a key made of a word's consonants (its consonant skeleton) helps link Arabic- and Latin-script spellings. The manuscript search on MÖKKEM matches spelling variants this way.
Right-to-left and mixed text
When Arabic-script and Latin-script text appear on the same line, punctuation and numbers end up in the wrong place unless text direction is marked correctly. Each field on an entry page should have its own direction.
The solution in dictionary software
- A separate, normalised search field
- Links between Arabic- and Latin-script spellings
- From inflected and suffixed forms to the entry
- Correct text direction and fonts