This paper illustrates and discusses recurrent sound correspondences between Pumé (also known as Yaruro) and Chocoan languages. Pumé is a language of the Apure state of Venezuela and has so far been considered an isolate. Chocoan is a small language family of western Colombia and eastern Panama. Until now, these language groups have never been considered together and compared systematically. It is argued here that the recurrent sound correspondences attested in Chocoan and Pumé basic vocabulary are difficult to explain by coincidence or language contact. It is therefore concluded that there should be enough evidence to postulate a genealogical link between both language groups.
I follow Constenla and Margery (1991) in indicating Proto-Chocoan forms with two asterisks and Proto-Emberá forms with one asterisk. Transcriptions of indigenous language data are phonemic according to the sources and use IPA symbols.
Dyck and Dyck (2007) do not seem to consider the existence of this phoneme. In the examples given by Mosonyi et al. (2000: 552) its occurrence seems to be restricted to specific environments e.g. in compound expressions preceding an onset vowel of the second element.
In the transcription of Dyck and Dyck (2007) ŋ and ɲ do not have phonemic status but are allophones of /ɡ/ and /ɟ͡ʝ/ respectively in a nasal environment. Cf. <ɡõa> ‘have’ (Dyck and Dyck 2007) ‘have’ (Mosonyi et al. 2000: 551); ‘say’ (Dyck and Dyck 2007) <ñõ> ‘say’ (Mosonyi et al. 2000: 551). Although both Dyck and Dyck (2007) and Mosonyi et al. (2000) interpret /m/ and /n/ as phonemes of Pumé their phonemic status may also need to be thoroughly assessed for similar reasons (see also the discussions in Subsections 5.2 and 5.5). In the transcriptions used in this paper <ŋ> and <ɲ> are only used in the context of data from Mosonyi et al. (2000) who interpret these elements as distinct phonemes in the Pumé variety they refer to.
Dyck and Dyck (2007) apparently do not interpret the opposition between the nasal and oral tap as phonemic: cf. <ãr̃õr̃ẽ́> ‘we’ and ‘to drink’ given by Mosonyi et al. (2000: 549; 551) and <ãrõrẽ> ‘we’ and ‘to drink’ given by Dyck and Dyck (2007). In the transcriptions used in this paper nasality of a tap is only indicated in the context of data from Mosonyi et al. (2000) who interpret the nasal tap as a distinct phoneme. In general terms a phonemic contrast between /ɾ̃/ and /n/ is rather unusual (Maddieson et al. 2014–2015; I am grateful to Natacha Chevrier for pointing out this information).