Due to the influence and domination of the Inca Empire, Yanesha' has many loanwords from Quechua, including some core vocabulary. Yanesha' may also have been influenced by Quechua's vowel system so that, today, it has a three-vowel system rather than a four-vowel one that is typical of related Arawakan languages.
- The affricates and /tʲ/ are phonetically aspirated
- [ŋ] is an allophone of /n/ before /k/
- /aˈnaxpʲ/ ('he answered him') vs /aˈnaxp/ ('he answered')
- /eˈʃota netsoˈʐamʲo/ ('entered my saw') vs /eˈʃota neˈnamo/ ('entered my mouth')
- /nʲa/ ('he') vs /na/ ('I')
The remaining two palatalized consonants, /lʲ/ and /tʲ/, don’t offer a one-to-one contrast with plain consonants; the former because it is the only lateral consonant and so contrasts with no other phoneme on the basis of just palatalization; /tʲ/, while contrasting with /t/, also contrasts with /ts/, /tʃ/, and /tʂ/. The bilabial palatalized consonants have a more perceptible palatal offglide than the alveolar ones. Word-finally, this offglide is voiceless for /pʲ/ and /lʲ/ while being absent for /mʲ/.
Another general feature of Yanesha' is devoicing in certain contexts. In addition to the devoicing of palatal offglides above, the retroflex fricative /ʐ/ is voiceless when word final (final devoicing) or before a voiceless consonant (regressive assimilation): /ˈaʐpa/ ('here it is') → [ˈaʂpa]. The approximants /w/ and /j/ are voiceless before voiceless stops, as in /wawˈteːna/ ('barks') and /nejˈtaʐ/ ('my door'); /j/ is also voiceless before affricates and word-finally: /aˈwej/ ('let's go').
Similarly, the stops /p/, t/, and /k/ are aspirated word-finally /eˈlʲap/ ('shotgun') → [eˈlʲapʰ]; preceding another stop or an affricate, a stop may be aspirated or unreleased so that /eːtˈkelʲ/ ('a fish') is realized as [eetʰkelʲ] or [eetkelʲ]. The velar fricative /x/ is debuccalized to [h] before another consonant.
Yanesha' has three basic vowel qualities, /a/, /e/, and /o/. Each contrasts phonemically between short, long, and "laryngeal" or glottalized forms.
Laryngealization generally consists of glottalization of the vowel in question, creating a kind of creaky voice. In pre-final contexts, a variation occurs—especially before voiced consonants—ranging from creaky phonation throughout the vowel to a sequence of a vowel, glottal stop, and a slightly rearticulated vowel: /maˀˈnʲoʐ/ ('deer') → [maʔa̯ˈnʲoʂ]. Before a word-final nasal, this rearticulated vowel may be realized as a syllabic quality of said nasal. Also, although not as long as a phonemically long vowel, laryngeal vowels are generally longer than short ones. When absolutely word-final, laryngealized vowels differ from short ones only by the presence of a following glottal stop.
/e/ is the short phoneme consisting of phones that are front and close to close-mid. Generally, it is realized as close [i] when following bilabial consonants. Otherwise, the phones [e] and [ɪ] are in free variation with each other so that /nexˈse/ ('my brother') may be realized as either [nehˈse] or [nehˈsɪ].
/eː/ is the long counterpart to /e/. It differs almost solely in its length, although when it follows /k/ it becomes a sort of diphthong with the first element being identical in vowel height while being more retracted so that /keː/ ('large kind of parrot') is realized as [ke̠e].
Laryngeal /eˀ/ consists of the same variation and allophony of the short phoneme with the minor exception that it is more likely to be realized as close following /p/ as in /peˀˈʃeːʐ/ ('parakeet') → [piˀˈʃeeʂ] 'parakeet'
/a/ is the short phoneme consisting of phones that are central. Its most frequent realization is that of an open central unrounded vowel [ä] (represented hereafter without the centralizing diacritic). Before /k/, there is free variation between this and [ə] so that /naˈnakʰ/ ('exceedingly') may be realized as [naˈnakʰ] or [nanˈəkʰ]. While the laryngeal counterpart is qualitatively identical to the short, the long counterpart, /aː/, differs only in that [ə] is not a potential realization.
/o/ is the short phoneme consisting of phones that are back as well as rounded. Generally, [o] and [u] are in free variation so that /ojoˈwa/ ('wild pig') may be realized as [ojoˈwa] or [ujuˈwa]. The phone [ʊ] is another potential realization, although it most frequently occurs before stops so that /not/ ('my hand') may be realized as [nʊtʰ]. [ʊ] is not a potential realization of long /oː/ but both the long and laryngeal counterparts are otherwise qualitatively identical to short /o/.
All consonants appear initially, medially, and finally with the exception that /ɣ/ and /w/ do not occur word-finally. With two exceptions (/tsʐ/ and /mw/), initial clusters include at least one stop. The other possible initial clusters are:
- /pw/, /pr/, /tr/, /kj/
- /tʃp/, /ʐp/, /tʃt/, /ʃt/
Word final clusters consist of either a nasal or /x/ followed by a plosive or affricate:
- /mp/, /nt/, /nk/, /ntʲ/, /ntʃ/, /ntʂ/
Medial clusters may be of two or three consonants.
Although apparently phonemic, stress tends to occur on the penultimate syllable but also in the ultimate. Less frequently, it is antepenultimate. Some words, like /ˈotʂen/~/oˈtʂen/ ('comb'), have stress in free variation.