Malayalam phonology

• As in other Dravidian languages, the retroflex series are true subapical consonants, in which the underside of the tongue contacts the roof.^[7]
• All of the alveolars except /s/ are apical.^[2]
• /ɕ ~ ʃ, t͡ɕ ~ t͡ʃ, t͡ɕʰ ~ t͡ʃʰ, d͡ʑ ~ d͡ʒ, d͡ʑʱ ~ d͡ʒʱ/ can either be postalveolar or alveolo-palatal depending upon the speaker and dialect; the postalveolar and alveolo-palatal realizations are in free variation.^[1]
• The alveolar nasal once had a separate character ⟨ഩ⟩ that is now obsolete and the sound is now almost always represented by the symbol that was originally used only for the dental nasal. However, both sounds are extensively used in current colloquial and official Malayalam, and although they were allophones in Old Malayalam, they now occasionally contrast in gemination – for example, eṉṉāl ('by me', first person singular pronoun in the instrumental case) and ennāl ('if that is so'), which are both written ennāl (എന്നാൽ) and tiṉṉŭ "eat!", tinnŭ "ate".
• The unaspirated alveolar stop also had a separate character ⟨ഺ⟩ but it has become obsolete, as the sound only occurs in geminate form (when geminated it is written with a റ below another റ ⟨റ്റ⟩) or immediately following other consonants (in these cases, റ or ററ are usually written in small size underneath the first consonant).
• The proto Dravidian alveolar stop *ṯ developed into an alveolar trill /r/ in many of the Dravidian languages and *ṉṯ became nn in Malayalam while *ṯṯ remained.^[5] Currently Malayalam only has [nd] in the genitive case ending -ṉṟe and a word formed with it taṉṟēṭam; Malayalam regained it from the older genitive case ending -ṉuṭaiya > -ṉuṭe > -ṉṭe > -ṉṟe, Malayalam still retains both forms in words like eṉṉuṭe and eṉṟe though the former is dated, a similar process happened in some Sri Lankan Tamil dialects.^[8]
• ന്റ is pronounced as [nd] but ൻറ can be pronounced as [nd] or [nr]; [nr] doesn't occur natively but it occurs in loans like എൻറോൾ (en̠r̠ōḷ) 'enroll' or ഹെൻറി (hen̠r̠i) 'Henry'.^[9]
• All non-geminated voiceless stops and affricate become voiced intervocalically and after a nasal as in Tamil.^[2][5]
• The geminated velars /kk/ and /ŋŋ/ are sometimes palatalized word medially after /j, i(:), e(:)/ like in the words കിടക്കുക [kiɖɐk:ugɐ] vs ഇരിക്കുക [iɾikʲ:ugɐ] and മങ്ങൽ [mɐŋ:ɐl] vs. മത്തങ്ങ [mɐt̪:ɐŋʲ:ɐ], their distribution is unpredictable e.g. it doesn't palatalize in vikkŭ but does in irikkŭ. If the palatalization is from /j/ it is sometimes deleted e.g. poykko can be [pojkʲːo] or [pokʲːo], aḍaykka as [ɐɖɐjkʲːɐ] or [ɐɖɐkʲːɐ]. Some of the northern dialects might pronounce them without palatalization.^[1][2]
• The letter ഫ represents both /pʰ/, a phoneme occurring in Sanskrit loanwords, and /f/, which is mostly found in comparatively recent borrowings from European languages. Though nowadays most people (especially youngsters) pronounce /pʰ/ as /f/ like in the word ഫലം /falam/.^[2] In the Jesari dialect the native word-initial /p/ is also spirantized to [f].^{[10][page needed]}
• /m, n, ɳ, l, ɭ/ are unreleased word finally.^[2] Words will never begin or end with a geminated consonant. All consonants appear word medially.^[2]
• The plain stops, affricates, nasals, laterals, the fricatives other than /h/ and /ʂ/ and approximants other than /ɻ/ can be geminated and gemination can sometimes change the meaning of the word, e.g. kaḷam 'cell', kaḷḷam 'lie'.^[2] /ɲ, ŋ, t/ only occur in geminated form intervocalically.^[1]
• The approximant /ɻ/ has both rhotic and lateral qualities, and is indeterminate between an approximant and a fricative. The articulation of /ɻ/ changes part-way through, perhaps explaining why it behaves as both a rhotic and a lateral, both an approximant and a fricative, but the nature of the change is not understood.^[11]
• /n, t, ɾ, l, ɻ/ are palatalized and have an advanced tongue root while /n̪, r, ɭ/ are clear or velarized and have a retracted tongue root, particularly noticeable in geminates. /ɾ, r/ can vary between ⟦ɹ̟ʲ ~ ɾ̟ʲ ~ r̟ʲ, ɾ̠ˠ ~ r̠ˠ⟧.^[1][12][13]
• /ʋ/ may be realized as [ʋ], [v], [w] or [β̞].^[1]
• Around 75% of nk and 50% of ñc and nt from Old Malayalam got assimilated to ṅṅ, ññ and nn, almost all of the ṉṯ merged with nn suggesting an earlier merger of some of the n̠t̠ and nt (for e.g. the cognate of Tamil nan̠r̠i is spelt as nandi (influenced by Sanskrit nandi) and pronounced nanni); mp and ṇṭ were unchanged, e.g. Tamil mūṉṟu, maruntu, kañci, teṅku, but Malayalam mūnnŭ, marunnŭ, kaññi, teṅṅŭ.^[14] Word final ai, āy and ey became a unless the word is monosyllabic, e.g. Tamil avai, māṅgāy, veṇṇey but Malayalam ava, māṅṅa, veṇṇa. Final āy in monosyllabic words became āya e.g. Tamil kāy, but Malayalam kāya.
• Loanwords with /z/ are replaced with /s/ and not with /d͡ʒ/ like in Hindi or Telugu e.g. /brasi:l/ English "Brazil" unless it was loaned through Indo-Aryan then the Indo-Aryan pronunciation is taken e.g. /d͡ʒilla/ Hi. /d͡ʒilaː/ Per. /zilʔ/, other Perso-Arabic phonemes like /q, x, ɣ, ħ, Cˤ, ʕ, ʔ/ are replaced with /k, kʰ, g, h, C, ∅, ∅/, sometimes /q, x/ are replaced with /kʰ, k/ e.g. Arabic قطر qaṭar, خَطّ‎ xaṭṭ became ഖത്തർ khattaṟ, കത്ത് kattŭ. English loans with /θ, ð, ʒ/ are replaced with /t̪, d̪, ʃ/; the dentals do not clash with English loans with /t, d/, which are replaced with [t, d] or [ʈ, ɖ], though [d] is rare because of the limited distribution natively e.g. "taxi" as ṯāksi or more commonly ṭāksi. The English /ɹ/ is loaned as either /ɾ/ or /r/ unpredictably, for e.g. 'current' got loaned as karaṇḍŭ but 'maroon' got loaned as mar̠ūṇ or mer̠ūṇ but the cluster /ɹs/ is loaned as /ɻs/ other clusters are loaned as /rC/ or /ɾC/, /ɻ/ only occurs in words with /ɹs/ e.g. 'force' as fōḻsŭ. Speakers with non rhotic English accents don't have /ɹC/ clusters in English loans and pronounce it as fōs(ŭ). In Sanskrit, loans with /t̪C/ and /d̪C/ (unless C is a sonorant or a dental stop) sometimes the /t̪, d̪/ becomes /l/ especially in /t̪s/ e.g. utsava > ulsavam, utpādana > ulpādaṉam, udghāṭana > ulghāḍaṉam.^[15] There are some native words with /s/ (urasŭ), /ʃ/ (vīśŭ), /ʂ/ (muṣiyuka), /pʰ/ (apphan), /bʰ/ (ēbhyan), some interjections with /h/ (ōhō, āhā) but rest of the fricatives (except /f/ in native words of Jesari) and aspirates are only found in loans.
• Rarely, some speakers pronounce the voiced aspirated consonants as voiceless so words like dhaṉam as thaṉam, it is more commonly deaspirated so dhaṉam as daṉam and kharam as karam, intervocalically the voiceless aspirate also becomes voiced so mukham as mogam.^[16]

The following text is Article 1 of the Universal Declaration of Human Rights.

All human beings are born free and equal in dignity and rights. They are endowed with reason and conscience and should act towards one another in a spirit of brotherhood.

മനുഷ്യരെല്ലാവരും തുല്യാവകാശങ്ങളോടും അന്തസ്സോടും സ്വാതന്ത്ര്യത്തോടുംകൂടി ജനിച്ചിട്ടുള്ളവരാണ്‌. അന്യോന്യം ഭ്രാതൃഭാവത്തോടെ പെരുമാറുവാനാണ്‌ മനുഷ്യന് വിവേകബുദ്ധിയും മനസാക്ഷിയും സിദ്ധമായിരിക്കുന്നത്‌.

manuṣyarellāvaruṁ tulyāvakāśaṅṅaḷōṭuṁ antassōṭuṁ svātantryattōṭuṅkūṭi janicciṭṭuḷḷavarāṇ‌ŭ. anyōnyaṁ bhrātr̥bhāvattōṭe perumāṟuvānāṇ‌ŭ manuṣyanŭ vivēkabuddhiyuṁ manasākṣiyuṁ siddhamāyirikkunnat‌ŭ.

/manuʂjaɾellaːʋaɾum t̪uljaːʋakaːʃaŋŋaɭoːʈum an̪t̪assoːʈum sʋaːt̪an̪t̪rjat̪t̪oːʈuŋkuːʈi d͡ʒanit͡ʃt͡ʃiʈʈuɭɭaʋaɾaːɳɨ̆ ǁ anjoːnjam bʱraːt̪rɨ̆bʱaːʋat̪t̪oːʈe peɾumaːruʋaːnaːɳɨ̆ manuʂjanɨ̆ ʋiʋeːkabud̪d̪ʱijum manasaːkʂijum sid̪d̪ʱamaːjiɾikkun̪ːat̪ɨ̆ ǁ/

^[5]

• Steever, Sanford B. (15 April 2015). The Dravidian Languages. Routledge. ISBN 978-1-136-91171-2.