| In phonetics, when 'nasal' is used as a noun it means a nasal consonant. As an adjective it can refer to nasal vowels, nasal release, nasal resonance: see nasalization for a full discussion. The rest of the present write-up is about nasal consonants.
Nasals are stops and sonorants, that is there is a complete blockage of the mouth, and the voice resonates freely without friction. They are almost always voiced. Burmese has voiceless nasals hm hn hng, as do some languages near it such as Hmong. In Old English there were a few words like hnutu, modern 'nut', which might have been pronounced as a voiceless nasal, or might have been a sequence h + n with a stronger h.
Most languages have two nasals, m n. It is very rare for a language to lack either or both of these, though some Salish languages of North-Western America do. In Yoruba the consonant n occurs only before i, as an allophone of l; though Yoruba also has nasal vowels, which are written with a following letter N.
Less common than the near-universal bilabial m and dental or alveolar n are nasals at other places of articulation. The palatal nasal is written ñ in Spanish, gn in French and Italian, nh in Portuguese, and ny in Hungarian and Catalan. Similar NY-like sounds occur in Polish and Japanese.
The velar nasal is the NG sound of sing. In almost all dialects of English this word ends in a nasal, with no consonant G after it, but historically it came from a cluster N + G, with the G lost somewhere in the Middle Ages. The NG is a single consonant sound: sing matches sin, sill, sick, not sink, sinned, silt. It also occurs before K and the K-like C, as in sink, uncle. Sometimes it occurs before an audible G, as in finger, single, and English, which (for most English-speakers) have NGG as opposed to the simple NG of singer. German has a similar kind of distribution, and a similar phonetic history of this sound. The phonetic symbol is a tailed n, which you might or might not be able to see: [ŋ].
The NG sound also occurs in Italian, Spanish, Greek, Japanese, and Hungarian, but not independently: it only occurs before the K and G sounds, as in cinque, cinco. This assimilation is obligatory: San Carlo and San Carlos are pronounced with sang, not san.
Italian and Spanish also have a labiodental nasal by assimilation when M or N precedes F or (in Italian) V. In English we might also get this sound in triumph, envy, though the assimilation is not automatic. Again, it's not an independent sound, and in fact I don't know of a language that has an independent labiodental nasal phoneme.
The uvular nasal, phonetic symbol [N], is extremely rare. It's reported in (some forms of) Inuktitut as a separate phoneme, and can occur by assimilation of N to a following Q in Persian.
Because the passage to the nasal cavity has to be open, nasals can't be made anywhere in the throat because the stoppage would prevent them being sonorant. However, clicks bypass the throat air-stream so can be nasal.
Retroflex nasals occur in languages that have other retroflex stops: Indian languages and Australian Aboriginal languages. Transcriptions from Indian usually use a dot underneath, while Australian languages may represent it as RN or as N.
Nasals can be syllabic. In English this often happens with N at the end of a word: button, burden when pronounced butt-n, burd-n without a vowel. We have interjections mmm and hm. A syllabic M can occur in a name like Clapham when pronounced Clap-m. In Cantonese the velar nasal ng can also be syllabic and occurs on its own as a surname. In Bantu languages an initial M or N is usually not syllabic, Mbeki being only two syllables (see Bantu, and the section 'prenasalization' under nasalization).
In Proto-Indo-European syllabic M and N are believed to have occurred, but have given way to other sounds. The negative prefix appears as un- in English, in- in Latin, and a- in Greek. The word for 'hundred', kmtom, has also changed its syllabic nasal to -un- in English and -a- in Greek hekaton, and became centum in Latin. | 
