Is there a sound change from [ɡ] to [i] or [j]?

Also, is it possible for [i] to become [ɡ] or only vice versa (as what I'm looking for).

I looked for information about it on Google and it was difficult to me to find it.


5 Answers 5


There certainly is.

For example, final -y in English often comes from an older -ig (and there is often a current German cognate that ends in -ig)

It's a widespread phenomenon, one form of palatalization.

I'm not aware of any instances of the reverse change, but I hesitate to say any sound change isn't "possible"

  • French had "y" /j/ > /ʒ/. It probably isn't too hard to find /ʒ/ > g
    – Tristan
    Jul 15, 2020 at 12:27
  • 8
    You need look no further than Germanic for the reverse development either: hardening of Proto-Germanic *jj and *ww yielded *ggj and *ggw outside West-Germanic (known as Verschärfung or Holtzmann’s Law). E.g., PG *ajjiz > ON eggr and Crimean Gothic ada (pl., from *addja- < *aggja-), but German Ei and OE ǣğ (English egg is from Norse). It even happened again later on in Faroese: cf. Icelandic ey ‘island’ and róa ‘to row’ vs Faroese oyggj and rógva. Jul 15, 2020 at 14:02

Yes, there are, and this kind of sound change is quite common.

Famous examples of this sound change are the German dialects of Berlin and Cologne, from the Berlin dialect we have the phrasal word jotwede (standard spelling jwd) that is explained as an abbreviation j.w.d. janz weit draußen (High German "ganz weit draußen" = "far away"), from Cologne the catch phrase es hätt' noch emmer joot jejange (High German "Es wäre noch immer gut gegangen", approximate meaning "What worked in the past will also work in the future").

Another example is the English language: In Anglo-Saxon, all /g/'s before front vowels were shifted to /j/ (spelled y), see also this answer for examples and counter-examples.

EDIT: Examples for the other direction are more rare and sparse, and often involve hypercorrection as a mechanism. Saxonian dialect Gung (High German: Junge "boy" comes to my mind, in this case a borrowing with hypercorrection from the Berlin-Brandenburg dialect). It is also present in proper names like Tigges from Matthias (Matthew).

  • In the Scandinavian languanges, written g is generally pronounced as /j/ when it comes before i or ei (e.g. geitost /j'aɪt:ʊst/), as well as a bunch of other situations. This also applies to gi /ji/, unlike the corresponding English give /giv/. Also, relatedly, k is often softened to /ç/. Jul 16, 2020 at 11:10
  • 3
    @leftaroundabout Only in Norwegian and Swedish (and k > [ɕ] rather than [ç] in Swedish). Danish doesn’t palatalise at all (except on the isle of Bornholm), Icelandic palatalises to [ɡʲ], and Faroese palatalises to [c ~ dʒ]. The Swedish/Norwegian version with full palatalisation of /g/ to /j/ before front vowels is, of course, also the same as the one that happened in Old English before influence from non-palatalising Old Norse confused things (e.g., give would have been yive if it hadn’t been for Norse influence). Jul 16, 2020 at 17:13

I do not believe that it is possible for [i] to become [g], plain and simple as you said. It is possible that some instances of [i] can in some context become [g], so a chain of events can turn [Cia] into [Cga] (intermediate steps being [Cja, Cɟa, Cga]), but that would not affect plain [Ci] – bia > bga but not bi > *bg. It is unclear whether you mean for these changes to be in all contexts (thus every instance of g > i), or simply in some context. For instance, in Norwegian, earlier g became [j] before front vowels (and k became ç), but g did not generally become j.

  • But see Janus Bahs Jacquet's comment to my answer.
    – Colin Fine
    Jul 15, 2020 at 16:21
  • Indeed, that is an example of what I'm talking about w.r.t. context.
    – user6726
    Jul 15, 2020 at 16:57

[g] > [i] is a common phenomenon in Germanic/Romance and might be elsewhere. Examples are:

(Unshifted words are to the left, shifted ones are to the right):

Proto-Germanic *sagjaną > English say (cognate with German sagen)

Vulgar Latin rege > Spanish rey, Portuguese rei

Modern Swedish pronunciation of torg > /tɔrj/

I cannot think of examples in the opposite direction. The only thing that comes to mind is a different change in Egyptian Arabic, where "j" is pronounced as "g" in words such as "hijab". If you can find a previous Semitic [i] that turned into Arabic j, you would then have the desired change.

  • 3
    Thank you. Just a note about Egyptian Arabic, the letter ج was pronounced as G and then it became to J, according to the most researchers, rather than vice versa. See Watson, Janet CE. The phonology and morphology of Arabic. Oxford University Press on Demand, 2002, p.15-16. Jul 16, 2020 at 22:58

Not sure if it fits the question, but Turkish ğ sounds (to me) quite close to j. Example: büyük (big), büyüğüm (I am big).

Also Mongolian seems to just have lost a lot of g sounds. Basically whenever people would have written vowel + g + vowel some centuries ago, they just write a double vowel nowadays. And the double vowel isn't even relevant to pronunciation, only to stress.

One might suspect that the old orthography was (in the case of g between vowels) not representative of old pronunciation, but one word that has cognates in other languages (baatar, which means hero) has a consonant between the first two a's in other languages. E.g. bahadur in India, bogatyr in Russian. It is written baghatur when written in traditional Mongolian script.

  • Turkish ğ is said to represent a phoneme /ɰ/ which is mostly lost (often with compensatory lengthening of a preceding vowel), unless between front vowels where it gets fronted to [ɰ̟] or [j] – so yes, in some contexts, it is pronounced as [j]. I’m not sure why you’d suspect the old Mongolian orthography was not representative of pronunciation, though, when loss of intervocalic [ɣ] or [ɰ] is so exceedingly common typologically. Apart from Turkish, it also happens in Finnish, Danish, Romance, Insular Celtic, and probably hundreds of other languages. Jul 18, 2020 at 9:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.