It has puzzled me for a long time, why the G sound is so different in Dutch, compared to the way it is pronounced in the languages around it, like English, French and German. While others are contented to just accept it as it is, I set out to discover the evolution of the G sound in Dutch.
This essay attempts to answer the following questions:
- How is ‘G’ pronounced in Modern Dutch?
- Has the ‘G’ sound undergone any changes in the evolution of the language?
- If so, why has the ‘G’ sound changed and why have these changes affected only Dutch and not the nearby languages?
- If these changes have really occurred, when did it take happen?
I will also use the following acronyms to denote the various developmental stages of languages:
ONL: Old Dutch
EMNL: Early Middle Dutch
MNL: Middle Dutch
ModNL: Modern Dutch
OE: Old English
ME: Middle English
ModE: Modern English
OHD: Old High German
MHD: Middle High German
ModHG: Modern High German
OND: Old Low German
MND: Middle Low German
ModND: Modern Low German
I will also refer to the following sounds written in IPA:
Voiced velar stop: /g/ as in ‘girl‘
Voiceless velar stop: /k/ as in ‘car‘
Voiced velar fricative: /ɣ/ as in ‘geld‘ ModNL
Voiceless velar fricative: /x/ as in ‘loch‘ ModHD
Voiced palatal fricative: /ʝ/ as in ‘hja‘ West Frisian
Voiceless palatal fricative: /ç/ as in ‘ich‘ ModHD
Voiced glottal fricative: /ɦ/ as in ‘haat‘ ModNL
Voiceless glottal fricative: /h/ as in ‘hat‘
The different ways of pronouncing <g> in Dutch
English-speaking learners of Dutch are often confounded by how <g> is pronounced like some sort of guttural <h> sound, like how the French pronounced their <r>s. In the Netherlands, the <g> sound is pronounced as /ɣ/, while <h> is pronounced often as /h/ and rarely as /ɦ/. This is to enable speakers to make a difference between the two :
‘hier‘ /hir/ vs ‘gier‘ /ɣir/
‘heus‘ /həs/ vs ‘geus‘ /ɣəs/
But this distinction is not always that big throughout the whole Dutch-speaking region: In former Southern Netherlands, modern-day Belgium, the standard <g> sound is pronounced as /x/ :
‘hier‘ /hir/ vs ‘gier‘ /xir/
‘heus‘ /həs/ vs ‘geus‘ /xəs/
This is known as the ‘soft G’ or ‘zachte G‘ /x/ in Dutch  which is a voiceless velar fricative, as opposed to the ‘hard G’ /ɣ/ spoken in the north, which is a voiced velar fricative. The soft <g> is used on television and radio and in formal speech.
But the reality is such that most Belgian Dutch-speakers further weakens the <g> sound into a voiced glottal fricative /ɦ/, such that the combinations of words are barely indistinguishable. In the West Flanders region of the country, where the <g> sound is the weakest, it is pronounced as an /h/:
‘hier‘ /hir/ vs ‘gier‘ /ɦir/ in common Belgian speech and /hir/ in West Flemish
‘heus‘ /həs/ vs ‘geus‘ /ɦəs/ in common Belgian speech and /həs/ in West Flemish
In West Flanders, in order to avoid confusion between g and h since they are now pronounced the same, the <h> sound is dropped entirely:
‘gouden‘ /houdən/ (golden) vs ‘houden‘ /oudən/ (to hold)
Is it the ‘Big House’, the ‘Goth House’ or the ‘Red House’?
Like the <g> sound, the <r> is pronounced differently in different regions: you have the Flemish r, the Gooise r, the Limburg r, the Rotterdam r etc . But in big metropolis like Amsterdam, Brussels and former French-speaking cities like Ghent and Leuven, the “more prestigious, French” r /ʁ/ has taken root , as opposed to the “traditional” rhotic r /r/ spoken in West Flanders.
This situation makes for the pronunciation of the common consonant group <gr> difficult to perceive to non-locals. So the phrase ‘het grote huis‘ (the big house) take on other meanings when spoken by different someone from Holland or from Flanders:
Amsterdam: /hət ɣʁotə hœys/ like ‘het Gotenhuis‘ (the Goth house)
Kortrijk: /t hrotə hys/ like ‘het rode huis‘ (the red house)
The origins of the Dutch language
No matter how <g> is pronounced in Modern Dutch, it is clear that the sound is not a velar stop /g/, like in English and German. Could there have been a similar development that happened in the past, that might explain this curious feature?
As you know, Dutch used to be known as Nederduits. The modern-day name Nederlands is only 200 years old and was invented for the national identity of the new kingdom of the Netherlands. Its old name and its name in English aptly describe its place in the West Germanic branch of Indo-European languages. It is the westernmost language that is part of the dialect continuum that stretches east across the north of Germany. The continuum denotes the Low German language and its border is marked by the Benrath Line . South of the line is High German, from which Modern High German is derived.
Dutch is also the direct descendant of Low Franconian. Low Franconian was spoken by Frankish tribes that crossed the Rhine into the Roman Empire in the 4th century. They were also the ones who conquered France in 496 and gave it its name. The Franks are also the ancestors of all Dutch-speaking people in Belgium and the Netherlands.
Sound changes in the High German consonant shift (3rd-9th century)
The West Germanic languages (except English) experienced what linguists call the High German consonant shift in the first millennium AD. By the time that we have written records of the languages affected, the changes have already been complete. Thus all of the changes are derived from comparing written records and current differences between the different languages. Certain shifts concern only High German (/p/ > /ff/), while other are consistent throughout West Germanic (/θ/ > /d/) .
The consonant shift effectively divided Low and High German. These differences are more or less consistent throughout the two languages, and are fundamental differences between modern Dutch and High German:
/p/ > /ff/ : ‘slapen‘ ModNL – ‘schlafen‘ ModHD
/p/ > /pf/ : ‘appel‘ ModNL – ‘Apfel‘ ModHD
/t/ > /ts/ : ‘tijd‘ ModNL – ‘Zeit‘ ModHD
These are just a few of the shifts observed. But I would like to draw your attention to this particular shift – /k/ > /x/ :
‘ik‘ ModNL – ‘ich‘ ModHD
‘maken‘ ModNL – ‘machen‘ /x/ ModHD
‘breken‘ ModNL – ‘brechen‘ /ç/ ModHD
But the velar stop /k/ did not become a velar fricative /x/ sound directly.
It first weakened into a glottal fricative /h/ before turning into a <ch> sound. This can be attested in Old High German (OHD):
OE: ‘rīce‘ /rɪk/
ModNL: ‘rijk‘ /rɛik/
OHD: ‘rīhhi‘ /rɪhɪ/
ModHD: ‘reich‘ /ʁaɪ̯ç/
Interestingly, this mirrors the current state of <g> in Dutch, where a velar stop is pronounced either as a glottal fricative /h/ in West Flanders or velar fricatives /x/ and /ɣ/ in other places.
In some cases, the transition from to <ch> stopped halfway. In Alemannic and Austro-Bavarian, some words are pronounced as /kx/:
‘werk‘ ModNL – ‘Werkch‘
‘kind‘ ModNL – ‘Kchind‘
This softening is complete only for one of the velar stops <k>. There is still the <g> sound.
In Modern High German, g is pronounced /g/ at the beginning of a word and /k/ at the end of a word. But curiously, in the <-ig> suffix, it weakens into a /ç/, as with the historical change of <k> to <ch> like in the other suffix <-lich>:
just like in ‘ziemlich‘ /tsimlɪç/ or ‘königlich‘ /khønɪklɪç/ .
This weakening of the <g> sound gets more and more frequent as one travels northwards, so much that whenever a word ends with <g> it is pronounced as the velar fricative /x/: Weg, Berg, genug.
This weakening could be due to the fact that the <g> is part of an unaccented syllable. Another example of this weakening is in the prefix <ge->.
In the south of Germany, this prefix is pronounced /ge/. As you travel northwards, it is pronounced with a schwa /gə/. This neutral sound gives it more tendency to weaken further. In some varieties of Low German, you can see how the prefix is weakened until it disappeared :
Standard ModHD: ‘gekauft‘
ModND Eastphalia: ‘ekofft‘
ModND North Lower Saxony: ‘kööpt‘
Standard ModHD: ‘geschlafen‘
ModND Eastphalia: ‘eslapen‘
ModND North Lower Saxony: ‘slapen‘
Standard ModHD: ‘genug‘
How does this show that the <g> sound the Dutch went from /g/ to /ɣ/ and not the other way round?
Like how the <k> sound weakened into <h> and <ch> during the High German consonant shift, the <g> sound in Low Franconian and related languages also weakened in the similar manner like how it is starting to do in High German’s <-ig> suffix.
I am postulating that this weakening is so widespread that it affected not only Old Dutch ONL, but also Old English OE.
Some linguists claim that the Dutch retained the West Germanic pronunciation of <g> /ɣ/, and it is during that High German consonant shift that it shifted into /g/ . I argue the opposite: <g> was originally pronounced as /g/ but it weakened into /ɣ/ and /x/ in Dutch.
Here are the reasons why I think so:
1. Traces of the old velar stop /k/ sound of <g> in Dutch
‘sprinkhaan‘ from ‘spring haan‘
‘jonkheer‘ from ‘jong heer‘ 
Interestingly, the same word in Afrikaans is spelled ‘springhaan‘ and pronounced with a /k/.
This shows that the pronunciation of /g/ or /k/ in Low Franconian lingered on in these few words in Modern Dutch.
The retention of the original velar stop /k/ sound of <g> in <-ng> constructions is also present elsewhere in other Low German dialects, also preserved when it is preceded by an n:
ModND: ‘lang‘ /lank/
ModND: ‘Kring‘ /krink/
ModND: ‘Lewing‘ /lewink/
These examples show that the <g> in Dutch and Low German used to be pronounced as velar stop /k/ and not as velar frivatives /ɣ/ and /x/ or glottal fricatives /ɦ/ and /h/ as is in the case today.
2. Lenition of <g> and <k> sounds in Old English
The second evidence I want to provide is the closest living relative to Dutch, which is English.
There are reasons to believe, that English underwent the sort of weakening of velar stops of both <k> and <g>, and that for the case of <g>, it first went through a lenition process  when it turned into /x/, /ɣ/ or /h/ before transiting to a <y>.
<g> words in Old English all became <y> or disappeared in Modern English:
ModE: ‘to say‘
ModNL: ‘gij‘ (only in Belgium)
Old English had similar pronunciation to Low Franconian. If like what certain linguists say that all g used to be pronounced as /ɣ/: Why did the <g> in words like ‘good’ or ‘great’ not turn into <y> as well?
For a <g> sound to turn into <y>, it probably went through the transition of /h/ or /ɣ/.
One proof of this is in the word: enough
The Old English word is ‘genóg‘, cognate to Dutch ‘genoeg‘ and High German ‘genug‘.
The fact that it is now spelled with <-gh> tells us that it must have been pronounced as a /ɣ/ or /x/, like in ‘night‘ and ‘light‘, a pronunciation that has disappeared in Modern English.
So Old English ‘genóg‘ must have been pronounced as /xənox/. Taken into account the weakening of the ge- prefix in modern Low German, the ge- further contracted into <e->/jə/.
This accounts for the disappearance of ge- entirely from all past participles, which were there in Old English, evolved into /jə/, but went out of fashion in the period of Middle English:
The disappearance of the other velar stop /k/ attests to its weakening in the transition from Old to Middle English. This is why in Modern English, the < -ly> adverbial suffix lacks an ending compared to Dutch and German, like:
3. Comparison of the <g> sounds in Gothic and Danish
Another factor I want to present as evidence is the pronunciation of the <g> sound in other Germanic languages.
Although the Goths and the Franks (ancestors of modern day Dutch-speakers) were different tribes, their languages were both directly derived from Proto-Germanic.
In Gothic, <g> is pronounced as a hard /g/, in the beginning of a word, like in ‘girl‘ (ModE).
But it weakens into a /x/ or /ɣ/ after a vowel, like in ‘wenig‘ (ModNL). 
Funny enough, this feature is retained in Modern Danish.
When the <g> is in the beginning of a word, it is a hard /g/ sound.
But unlike Gothic, when it is in the middle or end of a word, it is not pronounced at all. 
So in ag, eg, ig, øg, og, all of the g is pronounced /y/ or /h/.
This is because it has weakened from /g/ to /x/ or /ɣ/ (during Gothic era) and into /h/ or /j/ in recent times.
This weakening of <g> can also be seen in Swedish and Norwegian, where the hard /g/ is now a /j/ even in the beginning of a word. 
This same evolution happened in Old English more than a thousand years ago.
But more importantly, it demonstrates that the <g> sound was hard in Proto-Germanic (still the case in High German), but it softens to /x/ or /ɣ/ (which is still the case in Dutch), and it became a /y/ or a /h/ in Old English, Modern Danish and Swedish.
When did it happen?
If you look at the following table, you will see a series of words and their evolution from Old (500AD) to Early Middle (1150AD) then Middle Dutch (1500AD). While it is true that spelling was notoriously inconsistent back in the 1200s, we can still discern from the spelling conventions clues about the pronunciation of the words back then:
|Early Middle Dutch
words ending with g pronounced as /ç/:
Words ending with g in ONL have consistently evolved to be written with ch in EMNL. This means that the g has softened and pronounced as a /ç/.
words beginning with g pronounced as /g/:
Like in Gothic, High German and English, the velar stop is still being pronounced in EMNL, because we can telling from the spelling that it is NOT written as a ch or a gh: gaen, goet, groot.
words spelled with ghe pronounced as /x/ or /ɣ/:
Past participles with the ge- prefix is written ghe- in EMNL. Some linguists think that this is a mere spelling convention that quickly ran out of fashion in MNL. But we can say for certain it is not a velar stop /g/ , otherwise it would have been spelled simply with a g. This is especially telling in the EMNL word ghegaen, which would have been pronounced /ɣə-gan/.
Because the spelling harmonised to a single g like in Modern Dutch by 1500AD, we can safely say that the weakening of the Dutch velar stop /g/ was complete by then. This implies that the velar stop /g/ in words like goed, gaan, or groot, still pronounced in the 12th century, have disappeared by the 16th century.
The way <g> is pronounced today in Dutch is part of the process of the weakening of the velar stops in Germanic languages.
Contrary to the theory proposed by certain linguists that the Dutch <g> is the remnant feature of Proto-Germanic, the above evidence show that <g> was /g/ originally and is still the case in Modern High German.
With the transcribing of languages in words, the evolution of the <g> sound is captured or frozen in time for different languages.
What has happened in Dutch, is that the velar stop <g> has weakened at the end of a word, and that weakened <g> is generalised throughout the language in the modern era, such as all g’s are now /x/ or /ɣ/ .
5. van der Wal, Marijke. “Geschiedenis van het Nederlands Utrecht, 2004, p.413-414.
8. Although written as <ch>, there are two sounds which the ancient /k/ turned into in High German: / as in ‘ich‘, ‘brechen‘, ‘Bücher‘ (after closed vowels like i and e) and /x/ as in ‘Buch‘, ‘loch‘ and ‘machen‘ (after open vowels like a and o)