Introduction to Types of Sound Change

This reference resource lists and explains the types of sound change that take place in languages over time. I give the terms for each change, define that term, then give examples of that change in English & other languages. The menu above divides this page into five lessons, but here are the exact sound changes you're going to learn about:

Languages change over time - you already know that much if you've gone through my intro to language history. You've seen how an individual word changes as it passes from a proto-language into related daughter languages, where it transforms into distinct words called cognates. The differences between cognates manifest themselves most noticeably in the differences between the way the words sound. Because of this, you were able to spot sound correspondences that mark cognates in related languages, but also to recognize the sound changes that separate the daughter languages from their common parent. In the list below, we'll take a closer look at the types of sound change that shape the history of languages and language families all around the world, foreign and familiar alike.

Historical linguists often split sound change into two halves: regular changes versus irregular changes. A regular sound change takes place systematically, and in every instance of an environment. For example, if a language undergoes a sound change whereby /r/ becomes pronounced as /l/, every /r/ will change to /l/. If in that same language /n/ turns into /m/ between a vowel and a /b/, then every /n/ found after a vowel and before a /b/ will undergo that sound change. Early in the twentieth century, the regularity hypothesis proposed that sound change necessarily acts this way.

Irregular sound changes are less common and quirky in nature. They're actually known as sporadic sound changes because the sound change does not occur routinely in a given environment. Let's take an example from English. Some older dialects of English pronounced the word for a male fox fox and the word for a female of the species fixen. (We owe the i in fixen to an entirely regular sound change called "umlaut" that we'll discuss below). In other dialects, the two words were pronounced vox and vixen. Thanks to dialect mixing, modern English speakers ended up opposing fox to vixen. The change from Old Germanic /f/ to Modern English /v/ in /viksen/ is sporadic, as it doesn't match up with any regular, across-the-board sound change in English.

To understand sporadic sound change, we'd really need to look into the features of local dialects and regional language patterns. That's at least clear in the fox/vixen example, where the two words were inherited from two different dialects. We aren't going to attempt to uncover the background behind that kind of sound change here.

The topic of regular sound change will be the order of the day in this webpage. Specifically, we'll look at the various types of regular pronunciation change that are common to languages around the world, be they foreign languages like Japanese and Arabic or native like English. Notice that pronunciation changes very often make combinations of sounds easier to pronounce. In other words, one of the key motivations for speakers to change a sound is ease of articulation.

On to the various processes of sound change. Either before or as you learn the pronunciation changes below, please educate yourself on the hows and whys of language change and language families to make better sense of the crucial role sound change plays in language history.

Assimilation & Dissimilation

Assimilation

Assimilation. Two sounds are involved, and one becomes more like the other. The assimilating phoneme picks up one or more of the features of another nearby phoneme. The Latin root simil- is key here: two sounds become more similar. (Follow the "features" link above to refresh yourself on the features of IPA consonants.)

Example. The English phoneme /n/ has the features [alveolar, nasal, voiced]. The phoneme /p/ has the features [bilabial, plosive, voiceless]. The word input is sometimes pronounced ['ɪmpʊt]. In this word, /n/ has assimilated to /p/ - it comes to have the features [bilabial, nasal, voiced] instead of [alveolar, nasal, voiced]. (/n/ -> [m] before /p/).

Example. The Latin phoneme /d/ has the features [dental, plosive, voiced]. The phoneme /s/ has the features [alveolar, fricative, voiced]. The Latin word ad prefixed to the word similatio results in the pronunciation assimilatio. In other words, /d/ comes to share all features of the nearby /s/, and is completely assimilated to /s/.

Speakers of languages the world over use assimilation to make words easier to pronounce ("ease of articulation"). Try pronouncing the two examples above without assimilating the phonemes.

Types of Assimilation

Partial versus complete assimilation. Does the assimilating sound take on every feature of the phoneme that triggered assimilation (complete) or only some of or even just one of the features (partial)?

Example. The assimilation in the Latin assimilatio above is complete. On the other hand, the /n/ in ['ɪmpʊt] undergoes a partial assimilation.

Contiguous versus non-contiguous assimilation. Are the two sounds side by side (contiguous) or are they more distant (non-contiguous)?

Example. The two examples of assimilation I have given you both involved contiguous phonemes. The phonemes are non-contiguous in the accidental mispronunciation /'rʌʃən 'ʃʌbməri:n/ for /'rʌʃən 'sʌbməri:n/.

Progressive versus regressive assimilation. Does the changed sound come after the sound it's assimilating to (a regressive assimilation) or before the sound it's assimilating to (a progressive assimilation)?

Example. The mispronunciation /'rʌʃən 'ʃʌbməri:n/ is an instance of regressive assimilation. The assimilation of /n/ to /p/ in ['ɪmpʊt] is a fine example of progressive assimilation.

We can identify the precise type of assimilation by putting these concepts together:

Spelling IPA Type of Assimilation
input ['ɪmpʊt] Progressive partial contiguous
sells [sɛɫz] Regressive partial contiguous
Susie sells seassells... [su:zi: sɛɫz si:sɛɫz] Regressive complete non-contiguous

Assimilation by voicing, place & manner of articulation

Voicing. If the assimilating phoneme is originally voiceless, does it pick up the feature +voice from the other sound?

Example. Nearly all vowels in human language are voiced. Some Italian speakers pronounce the phoneme /s/ between two vowels as /z/, like in /la 'mia 'kaza/ 'my house'. The change from /s/ to /z/ is an instance of intervocalic voicing.

Devoicing. If the assimilating phoneme is voiced, does it pick up the feature -voice from the other sound?

Example. The word have ends in the voiced phoneme /v/. The word to begins with voiceless /t/. Some speakers devoice /v/ to /f/ when they pronounce the expression have to as hafta.

Palatalization. Does the assimilating sound move its place of articulation closer to the palate? This happens when the sound it's assimilating to already has a palatal or near-palatal place of articulation. Significant triggers include the consonant /j/ ("y sound") and the vowel /i/ ("ee sound").

Example. Some English speakers pronounce the word student as /'stʃju:dənt/. In rapid speech, the alveolar /t/ assimilates to the palatal /j/ and becomes palato-alveolar /tʃ/, resulting in the pronunciation /'sju:dənt/.

Fricativization. Does the assimilating sound change its manner of articulation so that the airflow is closer to a fricative? Less restricted sounds trigger this kind of change including fricatives and vowels. Stops (plosives) often undergo this assimilation.

Example. The mispronunciation /ʔʌʔfθru:/ for up through fricativizes the plosive /p/ as it assimilates to the fricative /θ/.

Of course, a phoneme can assimilate to any place or manner of articulation. I won't list all of them here. Further examples of assimilation to specific places of articulation include labialization (when the assimilating sound becomes more labial) and velarization (when the assimilating sound becomes more velar). When it comes to manners of articulation, the assimilating sound can become a plosive, an affricate and so on.

Dissimilation

Dissimilation. Two sounds are involved, and one becomes less like the other. The dissimilating phoneme loses one or more of the features it shares with another nearby phoneme. The two phonemes become more dissimilar as a result.

Example. The English phoneme /p/ is a voiceless bilabial stop. If you say Peter Piper picked a peck of pickled peppers as fast as you can, chances are you dissimilate one or more /p/'s by pronouncing them as an /f/.

Example. The expected Ancient Greek verb /thethne:ke/ '(s)he has died' was actually pronounced /tethne:ke/. The first aspirated /th/ dissimilated from the following /th/.

Example. The Latin root peregrin- comes into English as pilgrim rather than pirgrim. The first /r/ dissimilates to an /l/.

Languages use dissimilation to make words easier to pronounce ("ease of articulation"). Try pronouncing the examples I gave without dissimilating the phonemes.

Types of dissimilation

Dissimilation has the same characteristics found in assimilation above. The one consistent difference is that this process involves a phoneme that dissimilates from another sound.

Spelling Characteristics of Dissimilation
pilgrim Progressive non-contiguous
Peter Piper picked a feck of fickled... Regressive non-contiguous

Epenthesis & Elision

Epenthesis

Epenthesis. A sound is added to a word. The new phoneme is called an epenthetic sound.

Example. Some speakers add an epenthetic /r/ to the word sherbet, pronouncing it sherbert.

Types of epenthesis

Prothesis (or prosthesis). Is the phoneme added to the beginning of a word?

Example. Sardinian speakers changed /roza/ 'rose' to /aroza/ by adding a prothetic /a/.

Paragoge. Is the phoneme added to the end of a word?

Example. Ancient Greek speakers sometimes pronounced the word estí as estín, adding a paragogic /n/.

Anaptyxis (anaptyctic vowel). Is the additional sound a vowel? Consider the example of /aroza/ above.

Excrescence (excrescent consonant). Is the added sound a consonant? Look at the example with estín above.

Elision or Deletion

Elision. A sound is removed from to a word. The lost phoneme is said to be elided or deleted.

Example. The final /b/ sound has been elided from the word thumb. (Compare /b/ in thimble, literally 'little thumb'.)

Types of elision

Apheresis (especially if vowel). Is the phoneme removed from the beginning of a word?

Example. Speakers may shorten about time to 'bout time.

Syncope. Is the sound removed from the middle of a word?

Example. The Luiseño word hunwutumi 'the bears' is pronounced /hunwutmi/. Speakers syncopate the third /u/.

Apocope. Is the phoneme removed from the end of a word?

Example. The word thumb lost its final /b/ over time.

Various Changes

Lenition

Lenition. A sound is weakened. This weakening often happens adjacent to less restricted sounds like vowels or fricatives. This is a multi-step process - a lenited sound may simply go from voiceless to voiced (or plosive to fricative), while a fully lenited sound will disappear altogether.

Example. The Spanish stop /d/ is lenited to a fricative [ð] between vowels: dados 'dice' is pronounced /daðos/. Most speakers glide past this middle [ð] with even less restricted airflow: [dað̞os]. Some speakers take the lenition a step further by omitting [ð] entirely: [daos]. The path of this lenition was d > ð > ð̞ > 0.

Types of lenition

Opening. Does the sound go from a stop to a fricative to an approximant?

Example. Tuscan Italian speakers soften /k/ to a fricative /h/: una Coca Cola comes out as /una hoha hola/.

Sonorization. Does the sound go from voiceless to voiced, and then become progressively less restricted (from a stop to a fricative to an approximant)?

Example. The Spanish word /dados/ above was inherited from Latin /datos/. You can see a clear example of sonorization in this progression: /datos/ > /dados/ > /daðos/ > /dað̞os/ > /daos/.

Fortition

Fortition. A sound is strengthened. Strengthening turns a less restricted sound (like an approximant or a fricative) into a more restricted one (like an affricate or a stop).

Example. The "y-sound" /j/ in you is fortified to an affricate /dʒ/ in the colloquial pronunciation did ya? ['dɪʔdʒə].

Compensatory lengthening & metathesis

Compensatory lengthening. A consonant is deleted, but the preceding vowel is lengthened to make up for the loss.

Example. The Proto-Greek word *pods 'foot' eventually lost its /d/, but Ancient Greek speakers lengthened the vowel /o/ to compensate for the loss: /po:s/.

Metathesis. Two or more sounds swap places.

Example. Vowel and consonant are switched in the pronunciation of nuclear as nucular. Similarly, Sardinian speakers say trau 'bull' instead of the expected *taru.

Vowel changes

Vowel harmony. One or more vowels become like another vowel within the same word. i-mutation or umlaut.

Example. The vowel /a/ in Proto-Germanic *manniz 'people' was raised & fronted to /e/ in men under the influence of the high front vowel /i/. This is both an example of umlaut and of vowel harmony more generally.

Accent shift. The framework for assigning stress or pitch accent to vowels changes.

Example. Early Indo-European stressed different syllables in different words. However, Germanic languages like English have fixed the stress accent to the first syllable. Sanskrit and Greek accent the last syllable in pitá / patér, but the first in máta / méter; the English equivalents are fáther and móther.

Apophony (ablaut, gradation). Vowels shift to indicate different grammatical forms of the same word. This is not a historical sound change, it plays a part in a grammatical pattern.

Example. The vowel in the verb /spi:k/ ablauts to form the past tense /spo:k/.

Chain shifts

Chain shift. Multiple sounds move in an orderly fashion across one or more sound features. Initiated by a sound that either moves too close to another (pushing sounds away) or distances itself (pulling sounds closer).

Example. Early Modern English long vowels underwent the chain shift a: > e: > i: > ai, so that the word bathed came to be pronounced /be:ðd/, me as /mi:/, and so on.

Sound Laws, Phonological Rules & Ordering

Sound laws formalize and make explicit a regular sound change pattern in a language, or from parent to daughter languages across history. Key sound laws are named after their discoverers or popularizers (like Grimm's Law or Lachmann's Law). Laws may involve lenition, assimilation, or any of the above sound change types. The principles for formulating sound rules apply.

Example. Grimm's Law defines a systematic sound change in Proto-Germanic. Part of Grimm's law involves devoicing of voiced stops, so that the Proto-Indo-European root *gwen- turned into *kwen- 'woman', and the root *doru- became *trew- 'tree'. We can write the rule as: C [+plosive +voice] > C [+plosive -voice].

Sound laws

Sound laws are built out of rules - the same type of rules you learned about in intro to phonology. The order and relationship of rules is crucial for understanding their significance in the development of a language.

For instance, you can use your understanding of rules to establish a relative chronology by determining the sequence of sound shifts.

Example. Imagine that Language A underwent two minor sound shifts: one in which /n/ became /m/ before a labial consonant (this language has labials /p/ and /f/) (assimilation), and another in which a short unstressed vowel /i/ was deleted, provided it occured in the middle of a word (syncope). Language A has a word /'emfi/. Comparing data from closely related languages, we've been able to reconstruct the proto-form of that word as *yénifi. Which sound change happened first: /i/ > 0 or /n/ > /m/?

Feeding & Counterfeeding

When an earlier rule creates an environment in which a later rule can operate, the two rules have a relationship described as a feeding order.

Example. In the previous example, rule #1, which deleted /i/, allowed rule #2 (changing /m/ > /n/) to apply.

When the rules are reversed - the earlier rule abandoning an environment which is filled by the later rule - the result is a counterfeeding order.

Example. Imagine that, early in its history, our Language A had a rule that changed preconsonantal /m/ > 0. Language A inherited a proto-form *lámfi. Rule #1 ( m > 0 / V_C = /m/ deleted between a vowel and a consonant) applied to the proto-word, giving the expected output *láfi. Rule #2, which changed /n/ to /m/ before a labial consonant, produced the very result that was operated on by the first rule: /m/.

Bleeding & Counterbleeding

When an earlier rule creates an environment in which a later rule cannot operate, the two rules are in a bleeding order. Result: first rule applies, second does not.

Example. Language A has rule #1 that deletes /m/ after a vowel, before a consonant ( m > 0 / V_C ). It was this rule that changed the proto-word *lámfi to *láfi. Young speakers of Language A are applying a new rule, rule #2, which inserts a /p/ in the consonant cluster /mf/. For example, they pronounce the word émfi as ['e:mpfi]. However, rule #2 fails to apply to *láfi, which has already undergone rule #1.

When the bleeding rules are reversed - the earlier rule operates, followed by the later rule which would have prevented the earlier rule from applying - the result is a counterbleeding order. Result: both rules apply.

Example. Imagine that rule #1 in Language A deletes an unstressed /i/ in the middle of a word (syncope). Rule #2 moves the stress to the penultimate (second-to-last) syllable of a polyllabic word. Rule #1 caused *afisá to change to *afsá. Rule #2 moves the stress to the first syllable, resulting in the modern word áfsa. Notice that, had rule #2 applied first (*afisá > Xafísa), it would have prevented rule #1.