Which language code set (WALS, Linguasphere, ISO 639‑3 or Glottolog) should be the basis for a comparative linguistic project?

Question

To compile a word-list database which includes the available scattered word lists (mainly Swadesh lists), which language code should one use?

• WALS with 2679 entries, lots of linguistic data included. (Maintained Max Planck Institute)

• Glottolog with 7943 entries and also lots of data included. (Maintained Max Planck Institute)

• Linguasphere with about 21000 entries and some info included from 2000 or 2011. Theoretically limit of 18.9 million entries, thus theoretically flexible to include dialects and historic languages. (as far as I can tell dormant, otherwise Linguasphere Observatory)

• ISO 639-3 see also with currently about 7865 entries (theoretical limit of 17576 entries), used by several other sources. (Maintained by SIL)

• a different language code set all together?

I realise this is a bit of a subjective question but does anyone feel confident enough to say "this language code set should be used for the following reasons"?

Answer 1

If you can afford the extra work, use as many of the lists as possible, of course using explicit tags declaring from which set the code is taken, e.g.,

<languagecode type="iso639-3">eng</languagecode>

If you want to use only one of them, use ISO 639-3 if possible. It is an International Standard, well-maintained, frequently used and easily accessible. It is well received in the community and used, e.g. by OLAC: Open Language Archives Community and by the CLARIN Project.