In
linguistics, the imparting of a
nasal quality to a sound, by opening up the nasal cavity as an extra
resonator. All speech sounds are made with some configuration of the throat and oral cavity. The
velum or soft palate acts as a drawbridge: it is normally retracted so as to close off the nasal cavity, but in nasal sounds it is lowered to allow air to resonate simultaneously through the nose.
Nasal consonants, like M and N and the Ng in sing and the Ñ in España, have a complete closure somewhere in the mouth, and air escapes only through the nose. They may also be called nasal stops.
Nasal vowels have both the oral and nasal cavities open. Nasalization is thus an extra articulation on an oral vowel, so nasal vowels may also be called nasalized vowels. This is nitpicking, but there are no solely nasal sounds. Also, configuration of the nasal cavity can't be altered (say by flaring the nostrils), or at least if this is physically possible it's never been reported as used in any language. The only parameter of nasality is whether the velum is open or shut.
Familiar languages with nasal vowels are French, Portuguese, Polish, and Hindi, and they are quite common world-wide, especially in West African languages. The four French vowels are illustrated by the phrase un bon vin blanc 'a good white wine'. All languages without exception have oral vowels, and usually have more oral than nasal.
The IPA phonetic symbol for nasalization is a tilde, thus [œ˜ bõ vE˜ blã]. (Only the Portuguese letters ã õ will show up correctly in HTML, so I've had to use a separate tilde: it should be over the vowel in all cases.) In Polish Lech Wałęsa = [lex va'we˜sa]
This is called primary nasalization when the language systematically uses nasal vowels as distinct phonemes: French [bõ] 'good' contrasts with [bo] 'beautiful'.
Most consonants can't be nasalized, or rather a nasalized B just is an M. They are the same orally, and instead of exploding abruptly as B does, the air is released continuously through the nose in M. But some consonants, those called sonorants, allow their normal oral articulation to have simultaneous nasality imposed over them. This set includes the laterals or L-like sounds, the rhotics or R-like sounds, and the approximants including W and Y. However, it is extremely rare to have nasalized sonorants as phonemes: usually they are the result of secondary nasalization.
Secondary nasalization
In moving between oral vowels and nasal consonants, the
velum must be raised or lowered, and this takes time. It seems to be very general among the world's languages that some degree of secondary nasalization takes place, with the stretch of
vowel nearest the nasal
consonant being nasalized. The extent of this non-distinctive nasalization varies among languages.
It may extend only across the nearest fraction of the vowel, or the whole vowel, or the whole syllable. By the whole syllable, I mean any other consonants that can be nasalized: the sonorants but not the stops. This occurs in the Nigerian language Igbo, for example.
It can be progressive, with a nasal influencing a following vowel, so na pronounced [nã]; or regressive, with an pronounced [ãn]. The latter seems to be more common. Historically, this is a common source of nasal vowels: the vowel gets nasalized, then the consonant that caused it is deleted. So French [bõ] from older [bon], which is preserved in its spelling.
American English is said to use regressive nasalization more thoroughly than some other dialects. In French, which has to maintain a functional difference between oral and nasal vowels, only the final stretch of an oral vowel is nasalized, as in bonne 'good'. Experiments with deleting the consonant from recordings and getting native speakers to judge what followed show that in English the nasalization is more easily detectable than in French.
In Arabic a long nasalized approximant [jj~] arises from the coalescence of n with y ([j]) in e.g. Zaydun yaktubu 'Zayd is writing'.
Degrees of nasalization
Onset and offset of
velic closure are the main way nasalization varies in degree. While
Brazilian Portuguese has fully nasal vowels,
diphthongs, and even
triphthongs, in European
Portuguese only the last part of a nasal vowel
phoneme is nasalized. So
cão 'dog' is
[kau˜], not
[kãu˜]. Also, a short
homorganic consonantal segment is inserted before another consonant: so
vento 'wind',
[ve˜tu] is more narrowly
[ve˜ntu].
Although no variation of the shape within the nasal cavity is possible, some degree of opening of the velum is used in a few rare cases. Australian and New Zealand accents are sometimes described as slightly nasalized throughout.
Cases have been described where primary and secondary nasalization are of different strengths: primary stronger in Breton, but secondary stronger in Bengali. One Central American language is reported as having three phonemic degrees of vowel, oral, weak nasal, and strong nasal. Although other phonetic features might also be varying here, these seem to be cases where degree of opening of the velum makes a difference in nasality.
Prenasalization
Although a
stop (such as B, D, G) can't be nasalized and keep its abrupt stop quality, a similar effect can be had with a brief
homorganic nasal consonant before the stop. These can be notated phonetically as e.g.
[mb nd]. These sounds are very common in
Bantu languages, and also in
Melanesian languages such as
Fijian.
Of course many languages have consecutive sequences mb and nd, but mostly these are true sequences, transparently made from two consonants, and having a longer duration. But prenasalized stops are single consonants, functioning as such in the language's phoneme inventory and syllabification: so the Congolese town Mbandaka is three syllables, Mba-nda-ka.
Perhaps because the transition between the two segments is so quick, and there is no time to switch voice on and off, they are usually voiced pairs MB ND NgG, not voiceless MP NT NgK.
Postnasalization
Likewise there can be single consonants consisting of a stop with a short nasal segment following. These are quite rare. The
Arrernte language group of
Central Australia has PM, TN, KNg, and KNgw, as in the painter
Emily Kame Kngwarreye.
Such sounds may occur marginally in English, but across syllable boundaries: upmost may be pronounced with a single nasally-exploded P. But just try saying the syllable pmost and keeping the nasal release. It's much easier to insert a slight vowel or to explode the P separately, phmost.
The TN of button can also be released nasally, without anything intervening, but in this case the N is syllabic: the division up-most, butt-n makes them more pronounceable.
Arrernte is the only language I know offhand to have these sounds as phonemes, and use them in initial position. Russian permits sequences as in the River Dniestr, but this is probably not a prenasalized D: any native Russian-speakers out there? So voiceless prenasalized stops are possible: there seems to be a different energy constraint from the postnasalized case.
Nasal clicks
In general, stops can't be nasalized and keep the short
explosive quality of stops. A
stop is made by briefly blocking the mouth at some point, and allowing pressure of the
air stream to build up, so that there is an audible explosion on release. If the velum was open for nasality, the air from the throat would escape up the nose without forming a pressure difference.
The exception is where the stop closure is wholly within the oral cavity: the southern African sounds known as clicks. These form one closure at the velum using the tongue, and another further forward in the mouth, then deform the space above the tongue. It is possible to use this technique whether the velum drawbridge is lowered or raised: so the Zulu and Xhosa sounds written C Q X in their alphabet have nasalized versions NC NQ NX.
Information on degrees of nasalization, Igbo, and Brazilian Portuguese is from J. Laver, Principles of Phonetics, Cambridge, 1994.