Delicious Berber apples

While most Berber varieties use an Arabic loanword for "apple", several are reported to preserve a non-Arabic word: Jerbi a-ḏəffu (Brugnatelli), Nefusi dəffu (Motylinski), Zuara a-dəffu (Baghni). This word was derived by Vycichl (1952) from Punic *tappūḥ, a derivation generally accepted in subsequent work; Kossmann (2013:146) explains various forms along the lines of ta-dəffaḥ-t as blends between this and the Arabic form. Such an etymology makes sense on extra-linguistic as well as linguistic grounds: domestic apples originated much further east, in Central Asia, so a loanword is expected a priori, and given the important role of Carthage in early North African history, Punic appears the obvious source.

Talking to a speaker from near Batna yesterday, however, I realised that the Chaoui word for "apple" is really aḍfu, with an emphatic d. This cannot be explained in terms of regular sound change from the Punic form: the distinction between d and ḍ is in general very stable in Berber, particularly in the absence of any adjacent emphatic or laryngeal, and the apparent loss of gemination is also irregular.

The solution is Berber-internal. In more westerly varieties (cf. Nait-Zerrad, p. 451), we find a root ḍf-t for "taste, savour": Ait Atta t-aṭfi (verb iṭfi-t), Tashelhiyr tiḍfi (verb aḍfu-t), Zenaga taṭfih - also borrowed into Korandje təṭfi. While its geographical distribution seems relatively limited, nothing about this root suggests a foreign origin, and its attestation in Zenaga suggests a priori that it goes back to proto-Berber. We may therefore plausibly assume that at some point it was familiar to Chaoui speakers, if it isn't still. An otherwise unanalysable term for "apple" would therefore have been reinterpreted as, essentially "the tasty one".