Debuccalization

/q/ is debuccalized to /ʔ/ in several Arabic varieties, such as northern Egyptian, Lebanese, western Syrian, and urban Palestinian dialects, partially also in Jordanian Arabic (especially by female speakers).[4] The Maltese language, which was originally an Arabic dialect, also shows this feature.

Most English-speakers in England and many speakers of American English debuccalize /t/ to a glottal stop [ʔ] in two environments: in word-final position before another consonant (American English IPA)

• get ready [ˈɡɛʔˈɹɛɾi]
• not much [ˈnɑʔˈmʌtʃ]
• not good [ˈnɑʔˈɡʊd̚]
• it says [ɪʔˈsɛz]

and before a syllabic [n̩] following /l/, /r/, /n/, or a vowel. Here the /t/ may also be nasally released. (American English IPA)

• Milton [ˈmɪlʔn̩]
• Martin [ˈmɑɹʔn̩]
• mountain [ˈmæʊnʔn̩]
• cotton [ˈkʰɑʔn̩]
• Latin [ˈlæʔn̩]

In Cockney English, /t/ is often realized as a glottal stop [ʔ] between vowels, liquids and nasals (notably in the word bottle), a process called t-glottalization.

In German, voiceless stops are commonly debuccalized before syllabic nasals in the following clusters: /mpm̩, ltn̩, ntn̩, ŋkŋ̍/, which thus become [mʔm̩, lʔn̩, nʔn̩, ŋʔŋ̍]. For example, Lumpen (“rags, tatters”) is pronounced [lʊmʔm̩]. In some dialects, debuccalization may also occur before syllabic [l̩], though this is less common.

Voiced stops are not usually debuccalized. However, many Upper German and East Central German dialects merge voiced and unvoiced stops at least word-internally and these merged consonants will be debuccalized. Thus in Bavarian, Anten (“ducks”) and Anden (“Andes”) are both pronounced [ˈɑnʔn̩]. Speakers are often unaware of this.

All coda consonants in Slavey must be glottal. When a non-glottal consonant would otherwise be positioned in a syllable coda, it debuccalizes to [h]:[5]

• /ts’ad/ → [ts’ah] ('hat')
• /xaz/ → [xah] ('scar')
• /tl’uɮ/ → [tl’uh] ('rope')

In some varieties of Scots (and Scottish English), particularly on the West coast, a non word-final /θ/ th shifted to [h], a process called th-debuccalization. For example, /θɪn/ is realised as [hɪn].

In Proto-Greek, /s/ shifted to [h] initially and between sonorants (vowels, liquids and nasals).

• Proto-Indo-European *septḿ̥ → Proto-Greek *heptə́ → Ancient Greek heptá (ἑπτά) "seven" (vs. Latin septem)

Intervocalic /h/ was lost by the time of Ancient Greek, and vowels in hiatus were contracted in the Attic dialect.

• post-PIE *ǵénesos → Proto-Greek *génehos → Ionic géneos (γένεος) : Attic génous (γένους) "of a race"

Before a liquid or nasal, an /h/ was assimilated to the preceding vowel in Attic-Ionic and Doric and to the following nasal in Aeolic. The process is also described as loss of /h/ and subsequent lengthening of a vowel or consonant to keep the syllable the same length (compensatory lengthening).

• PIE *h₁ésmi → Proto-Greek *ehmi → Attic-Ionic ēmí (εἰμί) : Aeolic émmi (ἔμμι) "I am"

In Sanskrit, /s/ becomes [h] (written ḥ in transliteration) when utterance-final, e.g., kā́mas "erotic love" becomes kā́maḥ.

Additionally, the Indo-European aspirated voiced palato-velar *ǵʰ- became [ɦ], e.g. *bʰeh₂ǵʰús "arm" becomes Sanskrit bāhúḥ.

A number of Spanish dialects debuccalize /s/ at the end of a syllable to [h] or [ɦ].

In many varieties of Galician as well as in Galician-influenced Spanish, the phoneme /g/ may debuccalize (gheada) to [ħ] in most or all instances, though [x] and [h] are also possible realizations. There is also an inverse hypercorrection process of older or less educated Galician speakers replacing the phoneme /x/ of the Spanish language by [g], what is called gueada.

Portuguese is much less affected by debuccalization, but it is especially notorious in its Brazilian variety.

All over Brazil, the phoneme /ʁ/ (historically an alveolar trill /r/ that moved to an uvular position) has a rather long inventory of allophones: [r ɻ̝̊ ç x ɣ χ ʁ ʀ ħ h ɦ]. Only [ɣ] is uncommon. Few dialects, such as sulista and fluminense, give preference to voiced allophones; elsewhere, they are common only as coda, before voiced consonants.

In such dialects, especially among people speaking an educated variety of Portuguese, it is usual for the rhotic coda in the syllable rhyme to be an alveolar tap, as in European Portuguese and many registers of Spanish, or to be realized as [χ] or [x]. In the rest of the country, it is generally realized as [h], even among speakers not normally using this allophone or deleting it entirely, as is common in the vernacular.

However, in some mineiro- and mineiro-influenced fluminense rural registers, [h] is used but as an allophone of /l/ (rhotic consonants are most often deleted), a mar-mal merger, instead of the much more common and less stigmatized mau-mal merger characteristic of all Brazilian urban centers except for those bordering Mercosur countries, where coda [ɫ] was preserved, and the entire North and Northeast regions. Its origin is the replacement of indigenous languages and línguas gerais by Portuguese, which created [ɹ], [ɻ] and r-colored vowel as allophones of both /ɾ/ (now mostly /ʁ/) and /l/ (now mostly [u̯ ~ ʊ̯]) phonemes in the coda since Natives could not easily pronounce them (caipira dialect). The later Portuguese influence from other regions made those allophones become rarer in some areas, but the mar-mal merger did not disappear in a few isolated villages and towns.

Finally, many fluminense registers, especially those of the poor and of the youth, most northern and northeastern dialects, and, to a much minor degree, all other Brazilian dialects, debuccalize /s/ (that is, [ɕ ~ ʑ]) but less so than in Spanish. However, a mar-mas merger or even a mar-mais merger occurs: mas mesmo assim "but even so" or mas mesma, sim "though, right, the same (f) one" [mɐɦ ˈmeɦmə ˈsĩ]; mais light "lighter, more slim", or also "less caloric/fatty" [ˈmaɦ ˈlajtɕ]; mas de mim, não "but from me, no" or mais de mim, não "not more from me" [ˈmaɦ dʑi ˈmĩ ˈnɜ̃w]. A coda rhotic in the Brazilian dialects in the Centro-Sul area is hardly ever glottal, and the debuccalized /s/ is unlikely to be confused with it.

In the Moldavian dialect of Romanian, /f/ is debuccalised to [h] and so, for example, să fie becomes să hie. The same occurred in Old Spanish, Old Gascon and Old Japanese and occurs today in Sylheti.

In Scottish and Irish Gaelic, s and t changed by lenition to [h], spelled sh and th.