Phonological History Of English

Like many other languages, English has wide variation in pronunciation, both historically and from dialect to dialect. In general, however, the regional dialects of English share a largely similar (but not identical) phonological system. Among other things, most dialects have vowel reduction in unstressed syllables and a complex set of phonological features that distinguish fortis and lenis consonants ( stops, affricates, and fricatives).

This article describes the development of the phonology of English over time, starting from its roots in proto-Germanic to diverse changes in different dialects of modern English.

Abbreviations

In the following description, abbreviations are used as follows:

* C = any consonant
* V = any vowel
* # = end of word
* * = reconstructed
* ** = non-existent
* > = changes into
* >! = changes into, unexpectedly
* < = is derived from
* PreOE = Pre-Old English
* OE = Old English
* EME = Early Middle English
* LME = Late Middle English
* ME = Middle English
* EModE = Early Modern English
* ModE = Modern English
* GA = General American
* RP = Received Pronunciation
* PrePG = Pre-Proto-Germanic
* PG ''or'' PGmc = Proto-Germanic
* NWG = Northwest Germanic
* WG = West Germanic
* OHG = Old High German
* MHG = Middle High German
* ModG = Modern German
* PIE = Proto-Indo-European
* Goth = Gothic
* PN = Proto-Norse
* ON = Old Norse
* OEN = Old East Norse
* OWN = Old West Norse
* OS = Old Saxon

Changes by time period from Late Proto-Germanic to Old English

This section summarizes the changes occurring within distinct time periods, covering the last 2,000 years or so. Within each subsection, changes are in approximate chronological order.

The time periods for some of the early stages are quite short due to the extensive population movements occurring during the Migration Period (early AD), which resulted in rapid dialect fragmentation.

Late Proto-Germanic period

This period includes changes in late Proto-Germanic, up to about the 1st century. Only a general overview of the more important changes is given here; for a full list, see the Proto-Germanic article.

* Unstressed word-final , and were lost. Early ' > late PGmc ' "you carried (sg)".
* Word-final became .
** Word-final was then lost after unstressed syllables with nasalization of the preceding vowel. Hence *' > early PGmc ' > late PGmc > Old English "day (acc. sg.)". The nasalisation was retained at least into the earliest history of Old English.
* Word-final was lost after an unstressed syllable. This followed the loss of word-final , because it remained before : PrePGmc *' > early PGmc ' > late PGmc ' "they carried".
* was raised to in unstressed syllables.
** The original vowel remained when followed by , and was later lowered to .
* Early i-mutation: was raised to when an or followed in the next syllable.
** This occurred before deletion of word-final ; hence *' > early PGmc ' > late PGmc ' > German "over". Compare PIE *' > early PGmc ' > late PGmc ' > German "over".
** But it occurred after the raising of unstressed to : PIE *' > PGmc ' > ' "you carry (pl)".
** This also affected the diphthong , which became .
** As a consequence of this change, > . The Elder Futhark of the Proto-Norse language still contained different symbols for the two sounds.
* z-umlaut: is raised to before .
** Early PGmc ' "me, dative" > late PGmc ' > Old High German ''mir'', Old Saxon ''mi'', Old Norse ''mér'' (with general lowering and lengthening of i before r).
** This change was only sporadic at best because there were barely any words in which it could have occurred at all, since remained only in stressed syllables. The umlauting effect of remained, however, and in Old West Norse it was extended to other vowels as well. Hence OEN ''glaʀ'', ''hrauʀ'', OWN ''gler'', ''hreyrr''.
* Pre-nasal raising: > before nasal + consonant. Pre-PGmc *' > PGmc ' > ' > OE ''bindan'' > ModE ''bind'' (Latin ).
** This was later extended in Pre-Old English times to vowels before all nasals; hence Old English ''niman'' "take" but Old High German ''neman''.
* Loss of before , with nasalization and compensatory lengthening of the preceding vowel.
** The nasalization was eventually lost, but remained through the Ingvaeonic period.
** Hence Pre-PGmc *' > PGmc ' > OE ''þencan'' > ModE ''think'', but PrePG *' > PGmc ' > ' > OE > ModE ''thought''.
** This change followed the raising of before a nasal: PGmc ' > ' > ' > Gothic .
* Final-syllable short vowels were generally deleted in words of three syllables or more. PGmc ' > Goth "(he) carries" (see above), and also PGmc ', ' > ' (dative and instrumental plural ending of nouns, 1st person plural ending of verbs, as on the Stentoften Runestone).

Northwest Germanic period

This was the period after the East Germanic languages had split off. Changes during this time were shared with the North Germanic dialects, i.e. Proto-Norse. Many of the changes that occurred were areal, and took time to propagate throughout a dialect continuum that was already diversifying. Thus, the ordering of the changes is sometimes ambiguous, and can differ between dialects.

* Allophonic i-mutation/Germanic umlaut: Short back vowels were fronted when followed in the next syllable by or , by i-mutation: > , > , >
** In this initial stage, the mutated vowels were still allophonically conditioned, and were not yet distinct as phonemes. Only later, when the and were modified or lost, were the new sounds phonemicized.
** i-mutation affected all the Germanic languages except for Gothic, although with a great deal of variation. It appears to have occurred earliest, and to be most pronounced, in the Schleswig-Holstein area (the home of the Anglo-Saxons), and from there to have spread north and south. However, it is possible that this change already occurred in Proto-Germanic proper, in which case the phenomenon would have remained merely allophonic for quite some time. If that is the case, that would be the stage reflected in Gothic, where there is no orthographic evidence of i-mutation at all.
** Long vowels and diphthongs were affected only later, probably analogically, and not in all areas. Notably, they were not mutated in most (western) Dutch dialects, whereas short vowels were.
* a-mutation: is lowered to when a non-high vowel follows in the next syllable.
** This is blocked when followed by a nasal followed by a consonant, or by a cluster with in it. Hence PG ' > OE/ModE ''gold'', but PG ' > OE ''gyldan'' > ModE ''gild''.
** This produces a new phoneme , due to inconsistent application and later loss of word-final vowels.
* Final-syllable long vowels were shortened.
** Final becomes , later raised to . PG ' ("saw (tool)") > OE ''sagu'', ON ''sǫg''.
** Final becomes in ON (later raised to ), in West Germanic. PG ' ("he/she/it healed") > ON ''heilði'', but OE ''hǣlde'', OHG ''heilta''.
** The final long diphthong loses its final element and usually develops the same as from that point on. PG ' ("gift", dative singular) > NWG ' > ON ''gjǫf'', OHG ''gebu'', OE ''giefe'' (an apparent irregular development).
* "Overlong" vowels were shortened to regular long vowels.
* PG (maybe already by late PG) becomes . This preceded final shortening in West Germanic, but postdated it in North Germanic.
* Unstressed diphthongs were monophthongized. > , > . The latter merged with ''ō'' from shortened overlong ''ô''. PG ' ("son", genitive singular) > NWG ' > ON ''sonar'', OE ''suna'', OHG ''suno''; PG ' ("he/she/it take", subjunctive) > NWG ' > ON ''nemi'', OE ''nime'', OHG ''neme''; PG ' ("stone", dative singular) > NWG ' > ON ''steini'', OE , OHG ''steine''.

West Germanic period

This period occurred around the 2nd to 4th centuries. It is unclear if there was ever a distinct "Proto-West Germanic", as most changes in this period were areal, and likely spread throughout a dialect continuum that was already diversifying further. Thus, this "period" may not have been a real timespan, but may simply cover certain areal changes that did not reach into North Germanic. This period ends with the further diversification of West Germanic into several groups before and during the Migration Period: Ingvaeonic, Istvaeonic (Old Frankish) and Irminonic ( Upper German).

* Loss of word-final .
** This change occurred before rhotacization, as original word-final was not lost.
** But it must have occurred after the Northwest Germanic split, since word-final was not eliminated in Old Norse, instead merging with .
** was not lost in single-syllable words in southern and central German. Compare PG ' > OS ''mi'', OE ''me'' vs. OHG ''mir''.
** The OE nominative plural (ME ), OS nominative plural may be from original accusative plural ', due to the Ingvaeonic Nasal-Spirant law, rather than original nominative plural ', which would be expected to become *''-a'' (OHG ''-a'', compare ON ).
* Rhotacization: > .
** This change also affected Proto-Norse, but only much later. and were still distinct in the Danish and Swedish dialect of Old Norse, as is testified by distinct runes. ( is normally assumed to be a rhotic fricative in this language, but there is no actual evidence of this.)
** PG > Goth ; OE ''dēor'' > ModE ''deer''
*Intervocalic ðw > ww.
*Hardening: ð > d, β > v, and ɸ > f.
** > medially; in Old English this is extended to word-final position by analogy with inflected forms.
* West Germanic gemination: single consonants followed by except became double ( geminate). This only affected consonants preceded by a short vowel, because those preceded by a long vowel or by another consonant were never followed by due to Sievers' law.
** PG > OE ''biddan'', ''habban'' > ModE ''bid'', ''have''

Ingvaeonic and Anglo-Frisian period

This period is estimated to have lasted only a century or so, the 4th to 5th; the time during which the Franks started to spread south into Gaul (France) and the various coastal people began colonising Britain. Changes in this period affected the Ingvaeonic languages, but not the more southerly Central and Upper German languages. The Ingvaeonic group was probably never homogeneous, but was divided further into Old Saxon and Anglo-Frisian. Old Frankish (and later Old Dutch) was not in the core group, but was affected by the spread of several areal changes from the Ingvaeonic area.

The Anglo-Frisian languages shared several unique changes that were not found in the other West Germanic languages. The migration to Britain caused a further split into early Old English and early Old Frisian.

* Ingvaeonic nasal spirant law: Loss of nasals before fricatives, with nasalization and compensatory lengthening of the preceding vowel. Hence PG ' became ModG ''Mund'' but in Ingvaeonic dialects first became '. Old English then denasalised the vowels, giving OE > ModE "mouth".
** Following this > . PrePG *' > PG ' > ' > ' > OE > ModE "tooth". (ModG ''Zahn'' < OHG ''zant''.) This also applied to arising earlier in Proto-Germanic: PG ' > Late PG ' > OE > ModE "(I) thought".
* Anglo-Frisian brightening:
** Fronting of to (unless followed by a geminate, by a back vowel in the next syllable, and in certain other cases). Hence OE ''dæġ'' "day", plural ''dagas'' "days" (dialectal ModE "dawes"; compare ModE "dawn" < OE ''dagung'' ).
** This does not affect nasal . And since this is a back vowel, in a preceding syllable was prevented from being fronted as well. This created an alternation between the infinitive in *''-aną'' and strong past participle in *''-ana'' (< PG *''anaz''), where the former became -''an'' in OE but the latter became *-''ænæ'' > -''en''.
** Fronting of to (generally, unless followed).
* Final-syllable , and are lost.
** No attested West Germanic languages show any reflexes of these vowels. However, the way it affected the fronting of as described above shows that at least was retained into the separate history of Anglo-Frisian.

Old English period

This period is estimated to be c. AD 475–900. This includes changes from the split between Old English and Old Frisian (c. AD 475) up through historic early West Saxon of AD 900:

* Breaking of front vowels.
** Most generally, before , and + consonant (assumed to be velarized in these circumstances), but exact conditioning factors vary from vowel to vowel.
** Initial result was a falling diphthong ending in , but this was followed by diphthong height harmonization, producing short , , and from short , , and . Long , , and came from long , , and .
*** Written ''ea'', ''eo'', ''io'', where length is not distinguished graphically.
** Result in some dialects, for example Anglian, was back vowels rather than diphthongs. West Saxon ''ceald''; but Anglian ''cald'' > ModE ''cold''.
* Diphthong height harmonization: The height of one element of each diphthong is adjusted to match that of the other.
** > through this change, possibly through an intermediate stage . PG ' > OE > ModE ''stone''.
** was first fronted to and then harmonized to . PG ' > OE ' "joy" (cf. ModE ''dream'', ModG ''Traum''). PG ' > OE > ModE ''death'' (Goth , ModG ''Tod''). PG ' > OE ' > ModE ''eye'' (Goth , ModG ''Auge'').
** is harmonized to .
* A-restoration: Short is backed to when a back vowel follows in the next syllable.
** This produces alternations such as OE ''dæġ'' "day", pl. ''dagas'' (cf. dialectal ''dawes'' "days").
* Palatalization of velar consonants: were palatalized to in certain complex circumstances. A similar palatalization happened in Frisian, but by this point the languages had split up; the Old English palatalization must be ordered after Old-English-specific changes such as ''a''-restoration.
** Generally, the velar stops were palatalized before or ; after when not before a vowel; and was palatalized at the beginning of a word before front vowels. (At this point, there was no word-initial .)
** was palatalized in somewhat broader circumstances: By any following front vowel, as well as by a preceding front vowel when a vowel did not immediately follow the .
** later becomes , but not before the loss of older below.
** is palatalized in almost all circumstances. PG ' > ModE ''ship'' (cf. ''skipper'' < Dutch ''schipper'', where no such change happened), but West Frisian ''skip''. PG ' > OE ''scyrte'' > ModE ''shirt'', but > ON ''skyrt'' > ModE ''skirt''. An example of retained is PG ' > OE ''ascian'' > ModE ''ask''; there is evidence that OE ''ascian'' was sometimes rendered metathetized to ''acsian'', which is the presumed origin of ModE ''ask'' (and also of the modern dialectal pronunciation ''ax'').
* Palatal diphthongization: Initial palatal , , trigger spelling changes of ''a'' > ''ea'', ''e'' > ''ie''. It is disputed whether this represents an actual sound change or merely a spelling convention indicating the palatal nature of the preceding consonant (written ''g'', ''c'', ''sc'' were ambiguous in OE as to palatal , , and velar or , , , respectively).
** Similar changes of ''o'' > ''eo'', ''u'' > ''eo'' are generally recognized to be merely a spelling convention. Hence WG > OE ''geong'' > ModE "young"; if ''geong'' literally indicated an diphthong, the modern result would be *''yeng''. It is disputed whether there is Middle English evidence of the reality of this change in Old English.
* i-mutation: All back vowels were fronted before a in the next syllable, and front vowels were raised.
** > (but > before or );
** > > ;
** > ;
** , > > ; this also applied to the equivalent short diphthongs.
** Short > by an earlier pan-Germanic change under the same circumstances; often conflated with this change.
** This had dramatic effects in inflectional and derivational morphology, e.g. in noun paradigms (''fōt'' "foot", pl. ''fēt'' "feet"); verb paradigms (''bacan'' "to bake", ''bæcþ'' "he bakes"); nominal derivatives from adjectives (''strang'' "strong", ''strengþ(u)'' "strength"), from verbs (''cuman'' "to come", ''cyme'' "coming"), and from other nouns (''fox'' "fox", ''fyxenn'' "vixen"); verbal derivatives (''fōda'' "food", ''fēdan'' "to feed"); comparative adjectives (''eald'' "old", ''ieldra'' "older, elder"). Many echoes of i-mutation are still present in the modern language.
* Close-vowel loss: Loss of word-final and (also from earlier ) except when following a short syllable (i.e. one with a short vowel followed by a single consonant.) For example, PIE *' > PG ' > OE ''sunu'' "son (nom. sing.)", PIE *' > PG ' > OE ''feohu'' "cattle (nom. sing.)", PIE *' > PG ' > OE ''ƿine'' "friend (nom. sing.)", but PrePG *' > PG ' > WG ' > OE "foot (nom. pl.)".
* Loss of and following a long syllable.
** A similar change happened in the other West Germanic languages, although after the earliest records of those languages.
** This did not affect the new (< ) formed from palatalisation of PG , suggesting that it was still a palatal fricative at the time of the change. For example, PG ' > early OE *' > OE ''ƿrēġan'' ().
** Following this, PG occurred only word-initially and after (which was the only consonant that was not geminated by and hence retained a short syllable).
* ''H''-loss: Proto-Germanic is lost between vowels, and between and a vowel. The preceding vowel is lengthened.
** This leads to alternations such as ''eoh'' "horse", pl. ''ēos'', and ''ƿealh'' "foreigner", pl. ''ƿēalas''.
* Vowel assimilation: Two vowels in hiatus merge into a long vowel.
** Some examples come from ''h''-loss. Others come from loss of or between vowels, e.g. PG ''frijōndz'' > OE ''frīond'' > ''frēond'' "friend"; PG ''saiwimiz'' "sea (dat. pl.)" > *''sǣƿum'' > OE ''sǣm''.
* Back mutation: Short ''e'', ''i'' and (in Mercian only) ''a'' are sometimes broken to short ''eo'', ''io'', and ''ea'' when a back vowel follows in the next syllable.
** Hence ''seofon'' "seven" < PG ''*sebun'', ''mioluc, meoluc'' "milk" < PG ''*meluks''.
* Palatal umlaut: Short ''e, eo, io'' become ''i'' (occasionally ''ie'') before ''hs'', ''ht''.
** Hence ''riht'' "right" (cf. German ''recht''), ''siex'' "six" (cf. German ''sechs'').
* Vowel reductions in unstressed syllables:
** became in final syllables, but usually appears as ''o'' in medial syllables (although ''a'' and ''u'' both appear).
** and (if not deleted by high-vowel loss) became in final syllables.
** normally became in a final syllable except when absolutely word-final.
** In medial syllables, short are deleted; short are deleted following a long syllable but usually remain following a short syllable (except in some present-tense verb forms), merging to in the process; and long vowels are shortened.
* are unrounded to , respectively. This occurred within the literary period.
** Some Old English dialects retained the rounded vowels, however.
* Early pre-cluster shortening: Vowels were shortened when falling immediately before either three consonances or the combination of two consonants and two additional syllables in the word.
** Thus, OE ''gāst'' > ModE ''ghost'', but OE ''găstliċ'' > ModE ''ghastly'' (ā > ă) and OE ''crīst'' > ModE ''Christ'', but OE ''crĭstesmæsse'' > ModE ''Christmas'' (ī > ĭ).
** Probably occurred in the seventh century as evidenced by eighth century Anglo-Saxon missionaries' translation into Old Low German, "Gospel" as ''Gotspel,'' lit. "God news" not expected ''*Guotspel,'' "Good news" due to ''gōdspell'' > ''gŏdspell.''
* and were lowered to and between 800 and 900 AD.
* Initial became in late Old English. This occurred within the literary period, as evidenced by shifting patterns in alliterative verse.

Changes by time period from Middle English to American-British split

The Middle English Period

This period is estimated to be c. 900–1400.
* Homorganic lengthening: Vowels were lengthened before , , , , probably also , , , when not followed by a third consonant or two consonants and two syllables.
** This probably occurred around AD 1000.
** Later on, many of these vowels were shortened again; but evidence from the '' Ormulum'' shows that this lengthening was once quite general.
** Remnants persist in the Modern English pronunciations of words such as ''child'' (but not ''children'', since a third consonant follows), ''field'' (plus ''yield'', ''wield'', ''shield''), ''old'' (but not ''alderman'' as it is followed by at least two syllables), ''climb'', ''find'' (plus ''mind'', ''kind'', ''bind'', etc.), ''long'' and ''strong'' (but not ''length'' and ''strength''), ''fiend'', ''found'' (plus ''hound'', ''bound'', etc.).
* Pre-cluster shortening: Vowels were shortened when followed by two or more consonants, except when lengthened as above.
** This occurred in two stages, the first stage occurring already in late Old English and affecting only vowels followed by three or more consonants, or two or more consonants when two syllables followed (an early form of trisyllabic laxing).
* Diphthong smoothing: Inherited height-harmonic diphthongs were monophthongized by the loss of the second component, with the length remaining the same.
** and initially became and .
** and initially became and .
* Middle English stressed vowel changes:
** (from Old English ) and became and , respectively.
** (from Old English ) and merged into .
** New front-rounded and (from Old English ) were unrounded to and .
** and were unrounded to and .
* -dər > -ðər. This also occurred after the final reduction.
* became or , depending on surrounding vowels.
* New diphthongs formed from vowels followed by or (including from former ).
** Length distinctions were eliminated in these diphthongs, yielding diphthongs plus borrowed from French.
** Middle English breaking: Diphthongs also formed by the insertion of a glide or (after back and front vowels, respectively) preceding .
* Mergers of new diphthongs:
** Early on, high-mid diphthongs were raised: merged with (hence ''eye'' < OE ''ēġe'' rhymes with ''rye'' < *''riġe'' < OE ''ryġe''), merged with and merged with (hence ''rue'' < OE ''hrēoƿan'' rhymes with ''hue'' < OE ''hīƿ'' and ''new'' < OE ''nīƿe'').
** In Late Middle English, and merge as , so that ''vain'' and ''vein'' are homophones (the ''vein''–''vain'' merger).
* Trisyllabic laxing: Shortening of stressed vowels when two syllables followed.
** This results in pronunciation variants in Modern English such as ''divine'' vs ''divinity'' and ''south'' vs. ''southern'' (OE ''sūðerne'').
* Middle English open syllable lengthening: Vowels were usually lengthened in open syllables (13th century), except when trisyllabic laxing would apply.
* Reduction and loss of unstressed vowels: Remaining unstressed vowels merged into .
** Starting around 1400 AD, is lost in final syllables.
* Initial clusters , , were reduced by loss of .
* Voiced fricatives became independent phonemes through borrowing and other sound changes.
* before back vowel becomes ; becomes .
** Modern English ''sword'', ''answer'', ''lamb''.
** in ''swore'' is due to analogy with ''swear.''
* The cluster, present in words imported from Norman, is deaffricated, and merges with (which had perhaps been apical in medieval times, as in closely related Dutch and Low German), thus merging ''sell'' and ''cell''.
** But unlike French, and are fully preserved.
* In late Middle English, the extremely rare word-initial cluster ''fn-'' became '' sn-'' (EME ''fnesen'' > LME ''snezen'' > ModE ''sneeze'').
**It has been suggested that the change could be due to a misinterpretation of the uncommon initial sequence ''fn-'' as ''ſn-'' (''sn-'' written with a long s).

Up to Shakespeare's English

This period is estimated to be c. AD 1400–1600.
* H-loss completed: (written ''gh'') lost in most dialects, so that e.g. ''taught'' and ''taut'' become homophones, likewise ''bow'' (meaning "bend") and ''bough''. However, when preceded by /u/ (including in diphthongs), it sometimes got labialized to /f/, as in ''enough''.
* and when not followed by a vowel undergo mutations:
** Before , a coronal consonant or word-finally, they are diphthongized to and . (By later changes, they become and , as in modern ''salt, tall, bolt, roll''.) After this, the combinations and lose their in most accents, affecting words like ''talk'', ''caulk'', and ''folk''. Words acquired after this change (such as ''talc'') were not affected.
** Before , the becomes silent, so that ''half'' and ''calf'' are pronounced with , and ''salve'' and ''halve'' are pronounced with . is exempt, so that ''solve'' keeps its . is not wholly exempt, as the traditional pronunciation of ''golf'' was .
** Before , become , as in ''alms'', ''balm'', ''calm'', ''palm''; ''Holmes''.
** Some words have irregular pronunciations, e.g. from non-standard dialects (''salmon'') or spelling pronunciations (''falcon'' in American English).
* Short develop into lax
* Great Vowel Shift; all long vowels raised or diphthongized.
** become , respectively.
** become , respectively.
** become or , later and .
** New developed from old (see below).
*** Thus, effectively rotated in-place.
** Later, the new are shifted again to in Early Modern English, causing merger of former with ; but the two are still distinguished in spelling as ''ea'', ''ee''. the meet-meat merger (see below)
* Initial cluster reductions:
** merges into ; hence ''rap'' and ''wrap'' become homophones.
* Doubled consonants reduced to single consonants.
* Loss of most remaining diphthongs.
** became , merging with the vowel in ''broad'' and the of the lot–cloth split below.
** The long mid mergers: are raised to , eventually merging with , so that ''pane'' and ''pain'', and ''toe'' and ''tow'', become homophones in most accents.
** The above two mergers did not occur in many regional dialects as late as the 20th century (e.g. Northern England, East Anglia, South Wales, and even Newfoundland).
** merge to , so that ''dew'' ( EME < OE ''dēaƿ''), ''duke'' (EME < Old French