This article contains IPA phonetic symbols. Without proper rendering support, you may see question marks, boxes, or other symbols instead of Unicode characters. For an introductory guide on IPA symbols, see Help:IPA.

Consonant mutation is change in a consonant in a word according to its morphological or syntactic environment.

Mutation occurs in languages around the world. A prototypical example of consonant mutation is the initial consonant mutation of all modern Celtic languages. Initial consonant mutation is also found in Indonesian or Malay, in Nivkh, in Southern Paiute and in several West African languages such as Fula. The Nilotic language Dholuo, spoken in Kenya, shows mutation of stem-final consonants, as does English to a small extent. Mutation of initial, medial and final consonants is found in Modern Hebrew. Also, Japanese exhibits word medial consonant mutation involving voicing, rendaku, in many compounds. Uralic languages like Finnish show consonant gradation, a type of consonant mutation.

Similar sound changes

Initial consonant mutation must not be confused with sandhi, which can refer to word-initial alternations triggered by their phonological environment, unlike mutations, which are triggered by their morphosyntactic environment. Some examples of word-initial sandhi are listed below.

• Spanish: [b, d, ?], occurring after nasals and pause, alternate with [?, ð, ?], occurring after vowels and liquid consonants. Example: un [b]arco 'a boat', mi [?]arco 'my boat'. This also occurs in Hebrew (as begedkefet, an acronym for the consonants this affects) Aramaic and Tamil.
• Scottish Gaelic: in some dialects, stops in stressed syllables are voiced after nasals, e.g. [k?aht] 'a cat', [ ?aht] 'the cat'.

Sandhi effects like these (or other phonological processes) are usually the historical origin of morphosyntactically triggered mutation. For example, the English fricative mutation described above originates in an allophonic alternation of Old English, where a voiced fricative occurred between vowels (or other voiced consonants), and a voiceless one occurred initially or finally, and also when adjacent to voiceless consonants. Old English infinitives ended in -(i)an and plural nouns (of one very common declension class) ended in -as. Thus, h?s 'a house' had [s], while h?sas 'houses' and h?sian 'to house' had [z]. After most endings were lost in English, and the contrast between voiced and voiceless fricatives phonemicized (largely due to the influx of French loanwords), the alternation was morphologized.

Examples

English

In Old English, velar stops were palatalized in certain cases and not in others. This resulted in some alternations. Many of these have been levelled, but traces occur in some word doublets such as ditch /d?t?/ and dike /da?k/.

In the past tense of certain verbs, English also retains traces of several ancient sound developments such as *kt > *xt and *?x > *x; many of these have been further complicated following the loss of /x/ in the Middle English period.

• seek /si:k/ : sought /s?:t/
• think /?k/ : thought /:t/

The pair teach /ti:t/ : taught /t?:t/ has a combination of both this and palatalization.

A second palatalization, called yod-coalescence, occurs in loanwords from Latin. One subtype affects the sibilant consonants: earlier /sj/ and /zj/ were palatalized, leading to an alternation between alveolar /s z/ and postalveolar /? ?/.

• confess /k?n'f?s/ : confession /k?n'f???n/
• fuse /fju:z/ : fusion /'fju:??n/

Another, unproductive layer results from the Vulgar Latin palatalization of velar stops before front vowels, and is thus imported from the Romance languages. Here /k ?/ alternate with /s d?/.

• induce /?n'dju:s/ : induction /in'd?kn/
• magic /'mæd??k/ : magus /me???s/

A combination of inherited and loaned alternation also occurs: an alternation pattern *t : *sj was brought over in Latinate loanwords, which in English was then turned into an alternation between /t/ and /?/.

• act /ækt/ : action /'æk??n/

Celtic languages

The Celtic languages are well-known for their initial consonant mutations.^[1][2] The individual languages vary on the number of mutations available: Scottish Gaelic has one, Irish and Manx have two, Welsh, Cornish and Breton have four (counting mixed mutations). Cornish and Breton have so-called mixed mutations, where a trigger causes one mutation to some sounds and another to other sounds. Welsh also has a mixed mutation (triggered by na, ni, and oni). The languages vary on the environments for the mutations, though some generalizations can be made. In all the languages, feminine singular nouns are mutated after the definite article, and adjectives are mutated after feminine singular nouns. In most of the languages, the possessive determiners trigger various mutations. Following are some examples from Breton, Irish, Scottish Gaelic, and Welsh:

Older textbooks on Gaelic sometimes refer to the c -> ch mutation as "aspiration", but it is not aspiration in the sense of the word used by modern phoneticians^[], and linguists prefer to speak of lenition here.

Historically, the Celtic initial mutations originated from progressive assimilation and sandhi phenomena between adjacent words. For example, the mutating effect of the conjunction a 'and' is due to the fact that it used to have the form *ak, and the final consonant influenced the following sounds.^[3]

Welsh

Welsh has three main classes of initial consonant mutation: soft mutation (Welsh: treiglad meddal), nasal mutation (Welsh: treiglad trwynol), and aspirate mutation which is sometimes called spirant mutation (Welsh: treiglad llaes). The fourth category is mixed mutation which calls for the aspirate mutation where possible, otherwise soft mutation. The following tables show the range of Welsh mutations, with examples. A blank cell indicates no change occurs.

The mutation ts -> j reflects a change heard in modern words borrowed from English. Borrowed words like tsips/jips (chips) can often be heard in Wales. Dw i'n mynd i gael tsips 'I'm going to get (some) chips'; Mae gen i jips 'I have chips'. Despite this the ts -> j mutation is not usually included the classic list of Welsh mutations and is rarely taught in formal classes. Nevertheless, it is a part of the colloquial language and is used by native, first-language speakers.

h-prothesis

h-prothesis is a phenomenon in Welsh where a word which is vowel-initial becomes h-initial. This occurs after the possessive pronouns ei 'her', ein 'our' and eu 'their', e.g. oedran 'age', ei hoedran 'her age' (c.f. ei oedran 'his age'). It also occurs with ugain 'twenty' after ar 'on' in the traditional counting system, e.g. un ar hugain 'twenty-one', literally "one on twenty".

Irish

Irish, like Manx and colloquial Scottish Gaelic, uses two consonant mutations: lenition (Irish: séimhiú ['?e:.v?u:]) and eclipsis (Irish: urú ['?.u:]).

Lenition

Lenition (séimhiú) is indicated by an h following the consonant in question or, in some older typefaces and texts, by a dot () above the letter that has undergone lenition. The effects of lenition are as follows:

1. A stop becomes a fricative. Voicing is retained, as is place of articulation except with the coronals.
   • /p?/ -> /f?/
   • /p?/ -> /f?/
   • /t/ -> /h?/
   • /t?/ -> /h/
   • /k/ -> /x/
   • /c/ -> /ç/
   • /b?/ -> /w/, /v/
   • /b?/ -> /v?/
   • /d/ -> /?/
   • /d?/ -> /j/
   • /?/ -> /?/
   • /?/ -> /j/
2. /m?/ becomes /w/ or /v/; /m?/ becomes /v?/.
3. /s?/ and /?/ become /h/; but /s?p(?)/, /s?m(?)/, /s?t/, /?t?/, /s?k/, and /?c/ do not mutate.
4. /f?/ and /f?/ are deleted.

Eclipsis

The following tables show how eclipsis affects the start of words. Eclipsis is symbolised in the orthography by adding a letter--or occasionally two letters--to the start of the word. If the word is to be capitalised, the original first letter is capitalised, not the letter or letters added for eclipsis. An example is the "F" in Ireland's national anthem, Amhrán na bhFiann.

Russian

In Russian, consonant mutation and alternations are a very common phenomenon during word formation, conjugation and in comparative adjectives.

The most common classes of mutations are the alternation between velar and postalveolar consonants:

• ? /k/ -> ? /t?/
• ? /?/ -> ? /?/
• ? /x/ -> ? /?/, as in "quiet" and ? "quieter"
• Gain or loss of palatalization: ? "tsar" and ? "of the tsar" (adjective)

Other common mutations are:

• ? /t/ -> ? /t?/, ? /d/ -> ? /?/
• ? /z/ -> ? /?/, ? /s/ -> ? /?/, ? /ts/ -> ? /t?/
• /sk/ -> ? /?:/? -> "splash" / "(he) splashes", /st/ -> ? /?:/? -> "to whistle" / "I whistle"

Hebrew

Modern Hebrew shows a limited set of mutation alternations, involving spirantization only.^[4] The consonants affected may be stem-initial, stem-medial, or stem-final.

however, in Modern Hebrew, stop and fricative variants of ?‎‏, ?‎ and ?‎ are sometimes distinct phonemes, compare e.g.:

For a more in depth discussion of this phenomenon, see Begadkefat.

Japanese

Rendaku (meaning sequential voicing) is a mutation of the initial consonant of a non-initial component in a Japanese compound word.

Some compounds exhibiting rendaku:

• nigiri + sushi -> nigirizushi ("grip (with the hand)" + "sushi" -> "hand-shaped sushi")
• nigori + sake -> nigorizake ("muddy" + "rice wine" -> "unfiltered sake")

Uralic languages

Word-medial consonant mutation is found in several Uralic languages, where it goes by the traditional name of consonant gradation. Gradation is pervasive especially in the Samic and Finnic branches.

Finnish

Consonant gradation involves an alternation in consonants, between a strong grade in some forms of a word and a weak grade in others. The strong grade usually appears in the nominative singular of nominals and the infinitive of verbs.

The consonants subject to this change are the plosives (p, t, k) when preceded by a vowel, sonorant (m, n, l, r) or h. Plosives that are preceded by any other obstruent, or followed by any consonant, do not display gradation.

The gradation of loanwords may include new gradation patterns that are not native to Finnish:

Burmese

Burmese exhibits consonant mutation, involving voicing in many compound words.

The primary type of consonant mutation is when two syllables are joined to form a compound word, the initial consonant of the second syllable becomes voiced. This shift occurs in the following phones:

• /k?, k/ -> /?/
• /t, t?/ -> /d?/
• /s?, s/ -> /z/
• /t?, t/ -> /d/
• /p?, p/ -> /b/
• /?/ -> /ð/

Examples of this type include:

s?é () + k?á? (?) > s?é ?á? ("medicine" + "room" -> "clinic")

The second type of consonant mutation occurs when the phoneme /d?/, following the nasalized final /?/, can become a /j/ sound in compound words.

Examples of this type include:

"blouse" (? angkyi) can be pronounced /èd?í/ or /èjí/.

The third type of consonant mutation occurs when phonemes /p, p?, b, t, t?, d/, following the nasalized final /?/, can become /m/ in compound words. Examples include:

tà () + p () > tà m () ("to consult")

tá () + pà? () > tá mà? ("to apologize")

lè j () + pjà? () > lè mjà? ("airplane")

Central Vanuatu languages

Mutation of the initial consonant of verbs is a characteristic feature of many Austronesian languages spoken in central Vanuatu.

For example, in Raga language:

nan vano "I went"

nam bano "I go"

These patterns of mutations probably arose when a nasal prefix, used to indicate realis mood, became combined with the initial consonant of the verb.^[5] The possible ancestral pattern of mutation, and its descendants in some modern Central Vanuatu languages, are shown below:

Dholuo

The Dholuo language (one of the Luo languages) shows alternations between voiced and voiceless states of the final consonant of a noun stem.^[6] In the construct state (the form that means 'hill of', 'stick of', etc.) the voicing of the final consonant is switched from the absolute state. (There are also often vowel alternations that are independent of consonant mutation.)

• t 'hill' (abs.), god (const.)
• l?? 'stick' (abs.), luð (const.)
• k?do 'appearance' (abs.), kit (const.)
• t?o?o 'bone' (abs.), t?ok (const.)
• buk 'book' (abs.), bug (const.)
• k?tabu 'book' (abs.), k?tap (const.)

Fula

Consonant mutation is a prominent feature of the Fula language. The Gombe dialect spoken in Nigeria, for example, shows mutation triggered by declension class.^[7] The mutation grades are fortition and prenasalization:

For example, the stems rim- 'free man' and [?im-] 'person' have the following forms:

• [rim?e] (class 2), dimo (class 1), ndimon (class 6)
• [?im?e] (class 2), gim?o (class 1), ?gimkon (class 6)

Tagalog

Verb affixes ending in a velar nasal [?] (e.g. pang-, nang-) transform into a bilabial nasal [m] if the root begins with a bilabial ([b], [p]).

• bili (to buy) + pang- -> pambili (= for purchasing, buying)
• bata (child) + pang- -> pambata (= for children)
• bato (stone, rock, to throw) + nang- -> nambato (= threw [something] at someone)
• pili (to choose, select) + nang- -> namili (= chose [something])

Indonesian and Malay

The active form of a multisyllabic verb with an initial stop consonant or fricative consonant is formed by prefixing the verb stem with meN-, in which N stands for a nasal sharing the same place of articulation as the initial consonant.

• garuk -> menggaruk (= to scratch), hitung -> menghitung (= to count),
• beri -> memberi (= to give), fitnah -> memfitnah (= to accuse falsely),
• cari -> mencari (= to search), dapat -> mendapat (= to obtain), *jangkau -> menjangkau (= to reach)

If the initial consonant is an unvoiced stop or s, it disappears, leaving only the nasal in its place.

• kandung -> mengandung (= to contain or to be pregnant),
• putih -> memutih (= to turn white),
• satu -> menyatu (= to become one / to unite),
• tulis -> menulis (= to write).

Applied to verbs starting with a vowel, the nasal is realized as ng ([?]).

Monosyllabic verbs add an epenthetic vowel before prefixing, producing the prefix menge-.

• bor (= boring tool / drill) -> mengebor (= to make a hole with drill).

Verbs starting with a nasal or approximant consonant do not add the mutant nasal at all, just me-.^[8]

The colloquial version lose me- prefix and instead tends to use nasalization process.^[]

• tanya -> menanya -> nanya
• pikir -> memikir -> mikir
• merepotkan -> ng(e)repotin

Latvian

Additional info in Latvian

Also two consonants can mutate as a group.

Ute

In Ute, also called Southern Paiute, there are three consonant mutations, which are triggered by different word-stems.^[9] The mutations are Spirantization, Gemination, and Prenasalization:

For example, the absolutive suffix -pi appears in different forms, according to which noun stem it is suffixed to:

• movi-ppi 'nose'
• sappI-vi 'belly'
• a?o-mpi 'tongue'

Constructed languages

Sindarin

The Sindarin language created by J. R. R. Tolkien has mutation patterns inspired by those of Welsh. The first letter of a noun usually undergoes mutation when the noun follows a closely associated word such as an article or preposition. Thus, we get certh 'rune' and i gerth 'the rune'. Also, second elements of compounds and direct objects of verbs undergo mutation, as in Welsh. As of 2020 the Sindarin mutation patterns are not fully understood as Tolkien's notes on the language are not readily available. However some Sindarin enthusiasts and linguists have extrapolated patterns from published Sindarin texts. The linguist David Salo, who worked on the Peter Jackson trilogies The Lord of the Rings (2001 - 2003) and The Hobbit (2012 - 2014), proposes the following mutations in his book A Gateway to Sindarin (2004)^[10]:

A blank cell indicates no change.

The nasal mutation however does not affect 'd' and 'g' when found in the clusters 'dr', 'gr', 'gl' or 'gw'.

See also

• Lenition
• Fortition
• Consonant gradation
• Rendaku
• Sonority hierarchy
• Apophony
• Elision
• Historical linguistics

References

Further reading

• Grijzenhout, Janet. 2011. 'Consonant Mutation' in Marc van Oostendorp, Colin J. Ewen, Elizabeth Hume and Keren Rice (eds.) The Blackwell Companion to Phonology (Oxford: Blackwell) III: 1537-1558.
• Zimmer, Stefan. The Celtic Mutations: some typological comparisons. A Companion in Linguistics, a Festschrift for Anders Ahlqvist, ed. B. Smelik, R. Hofman, C. Hamans, D. Cram. Nijmegen: de Keltische Draak / Münster: Nodus 2004, 127-140.