English Loans in Tuvaluan and Chichewa
Tuvaluan is a Polynesian language, so it has a fairly limited set of phonemes. Unlike Chichewa or English, it only allows voiceless stops (so while English and Chichewa has /b d g p t k/, Tuvaluan has only /p t k/). It also disallows any consonant clusters, but allows vowels to come in sequence (though not within the same syllable - the sequence /ai/ is two syllables, unlike in English). Vowel length is phonemic. Words and syllables must end in a vowel.
Chichewa is a language I know much less about, but it is a Bantu language and a few things are immediately observable from John's brief word list (and a quick look at the sound system via Omniglot). It does not appear that vowel length is phonemic, and it does not appear that vowel sequences are allowed without an interceding glide. Consonant clusters seem to be limited to nasal + stop or glide. Like Tuvaluan, words may not be consonant-final.
Interestingly, in Chichewa voiceless stops can be aspirated (like in English) or unaspirated, but only some English voiceless stops correspond to aspirated voiceless stops in loans:
book [bʊkʰ]* > bukhu [bukʰu]
cup [kʰʌpʰ] > kapu [kapu]
Another interesting point in the loans is the treatment of English /ɹ/. Now, in Tuvaluan - as in many Polynesian languages - English /ɹ/ and /l/ merge to a single sound - /l/ in this case. Chichewa (at least that of the Shire, Malawi) has, by current reports to me, allophonic lateral and central taps (/ɺ/ and /ɾ/) like in Japanese, and as was reported in early documentations of many Polynesian languages (Hawaiian in particular). Now, let's look at how this turns up in English loans:
labor [leɪbə]* > leipa [lɛipa]
retire [ɹɪ'taɪə] > lītaea [liːtaɛa]
cholera [kʰɒləɹʌ] > kolera [koɺeɾa]
newspaper [njuzpʰeɪpʰəɹ] > nyuzipepala [njuzipepaɺa]
computer [kʰʌ̃mpjuɾəɹ] > kompyuta [kõmpjuta]
Now the sort of inconsistency seen in Chichewa loans is of course par for the course with borrowing - different sources, different times, an imperfect understanding of the source language's phonological rules, and reliance on spelling for pronunciation information can do all sorts of things to how a language interprets words it borrows, especially when English is involved.
What really interests me comparing these two is the treatment of palatalization before the vowel /u/ in English. Compare:
ENG cucumber [kʰjukʌ̃mbəɹ] > TUV kukama [kukama]
ENG computer [kʰʌ̃mpjuɾəɹ] > CHI kompyuta [kõmpjuta]
Tuvaluan does not have a palatal glide as Chichewa and English do, nor does it have any other palatal consonants, so English palatalization is completely dropped. In Chichewa, on the other hand, this glide is used not only in this case but in this other totally cool way:
key [kʰi:] > kiyi [kiji]
The glide gets placed essentially between iterations of a lengthened /i/ as a way of avoiding long vowels.
Conclusion: languages do cool things when they get together and hang out.
*A non-rhotic variety is the assumed source for Tuvaluan borrowings