What is phonological shift

1. Phonetic vs. phonological change

1.1. Phonetic change

Purely phonetic change only changes the phonetic characteristics of segments. If the paradigmatic relationships of the changed sounds to the others in the system do not change as a result, phonetic change at most introduces allophony, but does not change the sound system.

The change between [ç] and [χ] in New High German did not yet exist in Old High German. The phoneme / x / had a single (allo) phon there, [χ]. Then the change from B1.a took place. After that there were two allophones for / x /, but the relations of the phoneme to the rest of the system remained unaffected. The same can be said of the change in B1.b.

B1.a.ahd.[χ] → [ç]/ X __
Condition: X ≠ [+ behind]
b.engl. & kölsch[l] → [ɫ]/ __K0(i.e. in syllable rhyme)

In Asturian and other Spanish dialects, the velar variant occurs for / n / at the end of the word:

 c.astur./ n / → [ŋ] / __ ##

Ancient Greek had three series of occlusions:

  • [- sth, - asp] / p t k /;
  • [+ sth, - asp] / b d g /;
  • [- sth, + asp] / pʰ tʰ kʰ /.

On the way to Middle Greek, the last row was converted into fricatives (traditionally “spirantized”). The result was the sounds [f θ χ] (as in the words philosophía, theología, kharaktér). Only the feature [+ asp] was replaced by the feature [+ kont]. Since there were previously no fricatives at these points of articulation, the phonemes of the language remained the same; they just sounded different.

1.2. Phonological change

Phonological change results in a change in the phonological system.

Ancient Greek has had front spread vowels and back rounded vowels since Urindo-Ceramic times. In the Ionic dialect, the / u / (written as <υ> “Ypsilon”) was shifted to the front in historical times, i.e. it was spoken as [y] (e.g. in words like hupónumos “Hyponym”). This, too, is initially just a phonetic change. However, it changes the balance of the vocal system:

  • The / u / is overcharacterized as [y], because ceteris paribus the default [+ at the end] is sufficient for a closed round vowel. Since now the position of [u] (which is assumed in the same system according to the theory of markedness of [y]) is not occupied, it would be in the sense of the system to fill it. This also happens through a further change in which the diphthong / ou / is monophthongized to [u]. This now takes the position of the phoneme / u /, and at the same moment [y] (so far the only (allo) phon of the phoneme / u /) becomes the new phoneme / y /.
  • The / y / is now a front vowel, so it is in minimal opposition (no longer to / o /, but) to / i /. If this opposition is given up, / y / as a marked link loses the feature [+ round] and merges with the / i /. This is exactly what happens on the way to Middle Greek.

The two changes ‘/ u / → / y /’ and ‘/ ou / → / u /’ have just been shown as if the former triggered the latter. It could also have been the other way around. Or, even more likely, they took place in a systematic context and thus simultaneously.

The phonetic vs. phonological change was formerly also called subphonemic vs. phonemic change.

Sound change (as well as synchronous phonological processes) is mostly due to the context. His name is then combinatorial change. If it is not, a unit of the phonemic system is changed wherever it appears in texts. This is called more spontaneous (not triggered by anything) sound change.

2.1. Spontaneous change

Spontaneous change affects a segment or a class of segments regardless of conditions, i.e. in all contexts.

B2.idg./O/germ./ a /
idg.* ok̑towahd.ahtau "eight"

As a result, the former / o / coincides with the former / a / to form the new / a / (see below on coincidence and splitting).

The Germanic sound shift is a spontaneous sound change.1 The result is - with some simplification - a non-collapse change.

The loss of / h / in Latin is a spontaneous change in sound. As a result, the former / h / coincides with Ø, which means, for example, also hortus “Garden” with ortus Becomes homophonic.

However, change is only “spontaneous” if the segment is not analyzed in terms of characteristics. When sound change affects characteristics, segmental (“spontaneous”) change is due to other characteristics of the same segment. In any case, the underlying idea is correct that spontaneous change has no immediate phonetic plausibility (“takes place without a comprehensible cause”).

2.2. Combinatorial change

Combinatorial change affects a segment or a class of segments under certain conditions, i.e. in some of the contexts in which they occur. The conditions can, as in Phonetics & Phonology, Chap. 10 be phonological or morphological in nature.

2.2.1. Phonological conditioning

Assimilation is the most important type of phonologically conditioned sound change (or phonetic-phonological process). Total assimilation leads to the collapse of phonemes. The partial assimilation present in the Spanish intervowel detuning of occlusions leads, as can be seen in their representation, to splitting and collapse; the partial assimilation present in Italian palatalization leads to the splitting of a phoneme.

2.2.2. Morphological conditioning

The context conditions of a phonological change can also be morphological (see elsewhere on conditioning). One example is Latin rhotazism. The morphological conditioning leads here in the first step to the splitting of a phoneme. The second step, on the other hand, is a spontaneous change that transfers all [z] (there are only those that arose from / s /) into [r] and consequently subsumes under the phoneme / r /.

As seen in the description of the phonological processes, the umlaut begins historically as a phonologically conditioned process. In the further course of the development of the German language, however, it is limited by morphological conditions. E.g. the stem of cloud although in Clouds um, but not in cloudyalthough the suffix -ig in grainy but triggers the umlaut. Ditto it is said of Office though Offices, but official, of stupid but definitely stupid. Some suffixes with a closed vowel trigger the umlaut only in certain stems, and some stems are not accessible to the umlaut or only with certain suffixes. So morphological conditions apply here.

There are also from father and mother the umlaut forms Fathers and Motherswhere phonological conditions no longer exist. In such cases the umlaut is a morphological process, namely one of the allomorphs expressing certain grammatical categories - here the nominal plural.

3.1. Collapse and splitting

3.1.1. Collapse

In Uriberoromanic there was an opposition between / ʧ / and / s /. It has been preserved in Castilian in minimal pairs such as cerrar "shut down" - serrar “Saw” or cesión “Assignment” - sesión “Meeting”, where however the former / ʧ / has meanwhile shifted to / θ /. In other Ibero-Romance languages, however, the / ʧ / became / s / (via intermediate stages). In Andalusian and Portuguese are therefore cerrar and serrar homophon (port. [səˈʀaʁ]), and also port. cessão “Assignment” and sessão “Session” ([səˈsɐ̃w̃]). Viewed synchronously, the opposition is neutralized. As a result, firstly the phoneme system contains one less member and associated oppositions, and secondly the lexicon contains a number of homonyms more.

The Latin monophthongings also lead to collapse.

The resulting loss of a phoneme must be distinguished from its deletion in the syntagm:

Coincidence with another phonemeDeletion in the Syntagma
phonological mechanismNeutralization of the opposition with another phoneme in certain / all contextsDeletion of the phoneme without replacement from certain / all contexts
Consistency in the systemPhoneme disappearsPhoneme disappears
Consistency in textsit remains the counterpart of the phonemenothing remains of the phoneme

3.1.2. Splitting up

In Urromanischen there were the half-open vowels / e / and / o /, which as such e.g. in ital. probo 1) “trying”, 2) “proficient” and in port. neto 1) “grandchildren”, 2) “clean” are preserved. In Castilian, under certain conditions, they are diphthongized with an accent to / ue / or / ie /. This creates e.g. pruebo “Try” and nieto "Grandson". The diphthongs [ie] and [uo] are initially just conditioned allophones of the phonemes / e / and / o /. In a second step, however, further sound changes occur, which lead to these diphthongs appearing in the same contexts as the monophthongs, their former allophones. Now in Castilian there is an opposition between probo “Efficient” and pruebo “Try” between neto “Clean” and nieto "Grandson". The phonemes / e / and / o / have split into two phonemes or phoneme complexes.

Spontaneous changes in sound can only lead to collapse, not to splitting. Splitting is always contextual, i.e. allophonic. New phonemic units arise in a language in such a way that initially there is an allophony conditioned by the context, which is phonetically new. Secondarily, the conditioning factors disappear and the previous allophony becomes the opposition. The reanalysis of such an alternation as an opposition is its phonologization. We have already seen an English example of such a reanalysis. No new phonemes were introduced there. However, this happens in the following two examples:

3.2. French nasal vowels

Old French only had oral vowels (phase 0 in the diagram below). Vowels before nasal in the same syllable were nasalized allophonically (phase 1). The contrast between oral and nasal vowels is phonologized (phase 2) only when final consonants - including the one that formerly followed the vowel - are apocopied in a further, independent change, and the context that conditions the allophony no longer applies.

3.3. Italian palatalization

In the first stage, the palatalization seen in the presentation of the phonological processes takes place. It leads to allophony; the variants [k] ~ [ʧ] are distributed complementarily.

In a second stage, [k] and [g] are newly introduced before the front vowel, mainly by simplifying [kw] to [k], as in Latin. qui [kwi] "who"> Italian. chi [ki]. Now you have the two sounds in the same context, hence in opposition.

stepProcessVelarPalatal
SpellingAccording toimportanceSpellingAccording toimportance
old italy.Allophonyamicoa'mikofriendamicia'miʧiFriends
leggo'Legoreadinglegge'leʤ: eread
new italy.Phonological
ization
chikiwhociʧithere
casco'comprehensive insurancehelmetciascunoʧas'kunoeveryone
ghiro'giroDormousegiro'ʤirorotation
gallo'gal: oRoostergiallo'ʤal: oyellow

This sound change and its consequences for the orthography are shown on another website in a generally understandable way.

3.4. Phonologization and morphologization

A phonetic alternation (see Phonetics & Phonology, Chapter 10.4) changes into a phonological one if the former variants also occur without the conditions of the alternation being met. The paradigmatic relationship then becomes phonologized. This happens e.g. with the French nasalization and the Italian palatalization.

A morphologically conditioned phonological alternation goes into one morphological over if it occurs without the conditions of alternation being met. What the morphological condition was is then reanalyzed as the importance of the alternation. The alternation will morphologized. This happens e.g. with umlaut in German.

3.5. Dephonologization

A phonological opposition is dephonologized when it becomes allophony. The former phonemes come into free variation or complementary distribution, or the opposition is neutralized in favor of one of the terms. The examples of coincidence given above are at the same time cases of the dephonologization of an opposition.


1 The Germanic and High German sound shifts traditionally serve as prime examples of spontaneous sound changes. Strictly speaking, both are combinatorial in that they do not take place in certain contexts. E.g. the voiceless unaspirated plosives are not shifted to / s /. But that means nothing else than that these sound shifts are also subject to certain contextual conditions.