In phonetics and historical linguistics, fusion, or coalescence, is a sound change where two or more segments with distinctive features merge into a single segment. This can occur both on consonants and in vowels. A word like educate is one that may exhibit fusion, e.g. pronounced as //ˈɛdjʊkeɪt// or pronounced as //ˈɛdʒʊkeɪt//. A merger between two segments can also occur between word boundaries, an example being the phrase got ya pronounced as //ˈɡɒt jə// being pronounced like gotcha pronounced as //ˈɡɒtʃə//. Most cases of fusion lead to allophonic variation, though some sequences of segments may lead to wholly distinct phonemes.
A common form of fusion is found in the development of nasal vowels, which frequently become phonemic when final nasal consonants are lost from a language. This occurred in French and Portuguese. Compare the French words French: un vin blanc in French pronounced as /œ̃ vɛ̃ blɑ̃/ "a white wine" with their English cognates, one, wine, blank, which retain the n's.
Often the resulting sound has the place of articulation of one of the source sounds and the manner of articulation of the other, as in Malay.
Vowel coalescence is extremely common. The resulting vowel is often long, and either between the two original vowels in vowel space, as in pronounced as /[ai]/ → pronounced as /[eː]/ → pronounced as /[e]/ and pronounced as /[au]/ → pronounced as /[oː]/ → pronounced as /[o]/ in French (compare English day pronounced as /[deɪ]/ and law pronounced as /[lɔː]/), in Hindi (with pronounced as /[ɛː], [ɔː]/), and in some varieties of Arabic; or combines features of the vowels, as in pronounced as /[ui]/ → pronounced as /[yː]/ → pronounced as /[y]/ and pronounced as /[oi]/ → pronounced as /[øː]/ → pronounced as /[ø]/.
Compensatory lengthening may be considered an extreme form of fusion.
Historically, the alveolar plosives and fricatives have fused with pronounced as //j//, in a process referred to as yod coalescence. Words like nature and omission have had such consonant clusters, being pronounced like pronounced as //naːˈtiu̯r// and pronounced as //ɔˈmisjən//. Words ending in the Latin-derived suffixes -tion and -sion, such as fiction and mission, are examples that exhibit yod coalescence.
This sound change was not, however, distributed evenly. Words like module may be realised as either pronounced as //ˈmɒdjuːl// or pronounced as //ˈmɒdʒuːl//. Words that did not experience universal yod coalescence, are always realised as two segments in accents like Received Pronunciation. Most other dialects do pronounce them as one segment, however, like American English.
Words with primary stress on a syllable with such a cluster did not experience coalescence either. Examples include tune pronounced as //tjuːn// and assume pronounced as //əˈsjuːm//. Some dialects exhibit coalescence in these cases, where some coalesce only pronounced as //tj// and pronounced as //dj//, while others also coalesce pronounced as //sj// and pronounced as //zj//. In General American, pronounced as //j// elides entirely when following alveolar consonants, in a process called yod dropping. The previous examples end up as pronounced as //tuːn// and pronounced as //əˈsuːm//. Words that have already coalesced are not affected by this.
Australian English exhibits yod coalescence to an extreme degree, even when the cluster is in a stressed syllable, though there is some sociolectal variation. In an accent with full yod coalescence, tune and assume are pronounced like pronounced as //tʃuːn// and pronounced as //əˈʃuːm//. This can result in homophony between previously distinct words, as between dune and June, which are both pronounced pronounced as //d͡ʒuːn//.
Most Romance languages have coalesced sequences of consonants followed by pronounced as //j//. Sequences of plosives followed by pronounced as //j// most often became affricates, often being intermediary stages to other manners of articulation. Sonorants in such a sequence (except bilabial consonants) mostly became palatalized.
During the development of Ancient Greek from Proto-Greek, the labiovelar pronounced as /[kʷ]/, pronounced as /[kʷʰ]/, and pronounced as /[ɡʷ]/ became pronounced as /[p]/, pronounced as /[pʰ]/, and pronounced as /[b]/. Although the labiovelars were already a single consonant, they had two places of articulation, a velar articulation and labial secondary articulation (pronounced as /[ʷ]/). However, the development of labiovelars varies from dialect to dialect, and some may have become dental instead. An example is the word "cow" from Proto-Greek .
A vowel coalescence from Ancient Greek to Koine Greek fused many diphthongs, especially those including pronounced as //i̯//. E.g. pronounced as //ai̯// > pronounced as //e//; pronounced as //aːi̯// > pronounced as //a//; pronounced as //ɛːi̯// and pronounced as //oi̯// > pronounced as //i// and pronounced as //ɔːi̯// > pronounced as //o//.
Several consonant clusters in Proto-Celtic underwent fusion, most prominently /*ɡ/ to the following consonant in coda position. Examples include to and to in Old Irish.
In Norwegian and Swedish, this process occurs whenever the phoneme pronounced as //ɾ// is followed by an alveolar consonant. The articulation of the resulting fusion becomes retroflex. Examples include the Norwegian pronounced as /[bɑʈ]/ and Swedish pronounced as /[nuːɖ]/. This even occurs across word boundaries, as in the sentence "går det bra?" becoming pronounced as //ɡoː‿ɖə brɑː//.
This process will continue for as long as there are more alveolar consonants, though when this amount exceeds four, people usually try to break it up or shorten it, usually by replacing pronounced as //ʂ// with pronounced as //s//, or eliding pronounced as //d//. An extreme example of this would be the word ordensstraff pronounced as //ɔ.ɖɳ̩ʂ.ʂʈɽɑfː//, having six retroflex consonants in a row.
In colloquial Norwegian, the sequence /rt/ may even coalesce over non-alveolar phonemes, changing their place of articulation to retroflex, even if /r/ normally wouldn't trigger it. Examples include pronounced as //stæɾkt// pronounced as /[stæʈː]/, pronounced as //skɑɾpt// pronounced as /[skɑʈː]/, pronounced as //ʋæɾk.tœʏ̯// pronounced as /[ʋæʈ.ʈœʏ̯]/ and pronounced as //ʋɑɾmt// pronounced as /[ʋɑɳʈ]/. This process does not occur across word boundaries, e.g. sterk tann is pronounced pronounced as //stæɾk tɑnː// and not pronounced as /
In dialects where pronounced as //r// is articulated uvularly, this process invariably takes place on idiolectal level. For example, pronounced as //rɑːrt// may be realised as pronounced as /[ʁɑːʁt]/ or pronounced as /[ʁɑːʈ]/. This may appear in regions where /r/ has recently become uvular.[2]
In Malay, the final consonant of the prefix pronounced as //məN-// (where N stands for a "placeless nasal", i.e. a nasal with no specified place of articulation) coalesces with a voiceless stop at the beginning of the root to which the prefix is attached. The resulting sound is a nasal that has the place of articulation of the root-initial consonant.[3] For example:
Vowel coalescence occurs in Owari Japanese. The Diphthongs pronounced as //ai// and pronounced as //ae// change to pronounced as /[æː]/, pronounced as //oi// and pronounced as //oe// change to pronounced as /[øː]/ and pronounced as //ui// changes to pronounced as /[yː]/. E.g. pronounced as //raineN// > pronounced as /[ræ:nen]/, pronounced as //koi// > pronounced as /[køː]/, pronounced as //atsui// > pronounced as /[atsyː~atɕːyː]/. Younger speakers may vary between Standard Japanese diphthongs and dialectal monophthongs.[4]