Search these works ...
Exclude these works ...
Search these languages ...
In an exact search:
B following a vowel requires breathiness.
C following a vowel requires creakiness.
D following a vowel requires a diphthong.
L following a vowel requires a long vowel.
In a normal search:- diacritics marking tone and phonation are ignored.
- long vowels and diphtongs are matched automatically.
- script and regular 'g' (ɡ / g) are equivalent.
- ordinary 'h' matches IPA ʰ for aspirated.
- IPA m̥ matches traditional hm for unvoiced.
- ':' matches IPA 'ː' for long vowels.
|User interface||Underlying data|
Browser It is difficult to make pages display in
identical fashion on all browsers (and still have time for
useful work as well). This site will look best on Firefox 3.5+.
Latitude / longitude Lat and long figures are approximate guides to the locations where dictionary data was gathered. Points are often intentionally offset (e.g. Nicobaric) to make the colored pins visible.
Map rendering For unknown reasons, maps sometimes render slowly or only partially. We're working on it.
Item counts Counts of source items (in both the Database and the Dictionary) are sometimes reported inconsistently; this will be fixed.
Phonemic / phonetic For consistency in searching a phonemic
representation is most desirable. However, source data often
comes in a close phonetic rendering, or in native orthography.
Conversion is ongoing.
Transcription We make every effort to transcribe with 100% accuracy, but errors do occur. In most cases the original source page images are available through the Database; please report any errors.
Primary / secondary citations We try to identify, indicate, and rebuild secondary sources. As a result, single source items may appear more than once. Such references will be unified in the long run.
Large datasets In a very few cases (esp. Khmer data) the underlying data set was so large and comprehensive that individually redacting glosses would have greatly delayed getting any data on line. As a temporary solution, we have automatically extracted the leading glosses of all headwords not marked as Indic.
Re language values, note that "unfold" means "all languages in the prior branch."