Voiced Stops become Voiceless Stops

IE	b	d	g
Gmc after Grimm’s Law	p	t	k

Voiced Aspirates become Voiced Fricatives*

IE	bh	dh	gh
Gmc after Grimm’s Law	ᵬ	ð	ɣ

*/ᵬ/ represents a bi-labial fricative (like /v/, but with both lips together rather than with the top teeth on the bottom lip), and /ɣ/ represents a velar fricative rather like a gurgle. Usually the voiced fricatives developed into voiced stops (/b, d, g/) later in history, so that we can ultimately say the IE voiced aspirates became voiced stops. However, depending upon the position in the word, these sounds could also develop to /v, ð, j/.

Except in special circumstances, any IE word which contained the original consonants developed the new consonants after Grimm’s Law. We can see this if we compare some words from languages in other IE families, which did not undergo Grimm’s Law.

	English	Other IE Forms
Voiceless Stops	(IE p, t, k)
	father	pater (Latin), pitar (Sanskrit)
	foot	pes, ped- (Latin)
	fish	piscis (Latin), pesce (Italian)
	three	tres (Latin, Spanish), trayas (Sanskrit)
	thou	tu (French)
	hound	sun (Sanskrit), canis (Latin)
	hundred	centum (Latin), satem (Sanskrit)
	hemp	kánnabis (Greek)
Voiced Stops	(PrIE b, d, g)
	two	dva- (Sanskrit), duo (Latin), dos (Spanish)
	know	gignoskein (Greek)
	knee	genu (Latin)
	acre	ajras (Sanskrit), ager (Latin)
Voiced Aspirates	(PrIE bh, dh, gh)
	do	dha (Sanskrit), feci (Latin)
	wagon	vah- (Sanskrit), veho (Latin)

All these words are described cognates; that is, they all have a common descent from an earlier parent language, but they are not borrowed from each other. Hence a word like English two is a cognate of Latin duo. However, the English word duo (as in ‘the Dynamic Duo’) is borrowed directly from Latin and is therefore considered a loanword.

Grimm’s Law probably took place because of instabilities in the IE stop system. Each series of consonants was characterized by three distinct features: stopping, voicing, and aspiration. The systemic relation between the sounds can be visualised in the table below:

	STOPPING	VOICING	ASPIRATION
voiceless stops	yes	no	no
voiced stops	yes	yes	no
voiced aspirates	yes	yes	yes

Notice that all three features are needed to distinguish the consonants. However, this is no longer the case after Grimm’s Law:

	STOPPING	VOICING	ASPIRATION
voiceless stops	yes	no	no
voiceless fricatives	no	no	no
voiced fricatives	no	yes	no

Grimm’s Law effectively made aspiration a redundant feature of the system, since none of the resulting sounds were aspirated. Even when the voiced fricatives later became voiced stops, the system was still based only on voicing and stopping. This provides one motivation for the sound shift. Grimm’s Law made aspiration completely redundant. Afterwards, all three series could be distinguished by only two features, instead of three features.

An apparent exception to Grimm’s Law occurs where an IE voiceless stop followed another voiceless stop or the voiceless fricative /s/. For example, compare Latin stella and rëctus with their English equivalents star and right. If Grimm’s Law had affected these forms, we would expect them to have a /θ/ sound instead a /t/ sound. It is important to realize that after voiceless stops and /s/ no voiced sounds could occur. Due to a process called assimilation, any voiced sound which may have occurred in this environment in early IE would have become voiceless under the influence of the preceding voiceless consonant. The same thing happens in the English words robbed /rͻbd/ and stopped /stͻpt/, where the /d/ become voiceless /t/ because the preceding /p/ is voiceless.

We must now address the actual process by which Grimm’s Law took place. Did all the changes happen at once, or did some changes precede others? One possibility is known as the Drag-Chain Model. According to this model, Grimm’s Law was initiated by a change of the original voiceless stops to voiceless fricatives. The original voiced stops were then ‘dragged’ in the resulting ‘gap’ in the system. There are several problems with this model:

1. If the first change were of voiceless stops to voiceless fricatives, the result would have been a ‘gap’, or asymmetry, in the system. Although languages can function with such ‘gaps’, they are generally avoided unless there is some other motivation.

2. Since languages can function with such ‘gaps’, there is no necessary reason why the voiceless stop ‘gap’ should have been filled; some other motivation is required.

3. If this were a drag chain, the voiced aspirates should not have become fricatives; they should have become voiced stops in order to fill the ‘gap’. This did not happen until much later in Germanic, and only gradually and incompletely.

4. The Drag-Chain model does not explain the failure of Grimm’s Law after voiceless stops and the voiceless fricative /s/.

Another possibility is the Push-Chain Model. In this model aspiration was nearly redundant in the IE stop system: it only distinguished voiced stops from voiced aspirates. So these sounds ‘pushed’ apart in order to maximize the differences in their articulation. This caused a subsequent ‘push’ on the voiceless stop series. In other words, the voiced stops and voiced aspirates changed first in response to their close articulation; the change of voiceless stops to fricatives was a secondary change. The Push-Chain Model has several advantages:

1. The initial change of IE voiced stops to voiceless stops makes necessary the secondary change of IE voiceless stops to voiceless fricatives in order to maintain essential distinctions in the system.

2. The ‘pushing apart’ of the voiced stops and voiced aspirates provides an explanation for the production of new fricatives: the new sounds were now distinguishable as stop and non-stop consonants.

3. If the ‘trigger’ for the change came from the ‘pushing’ of the IE voiced stops on the IE voiceless stops, then it stands to reason that that ‘trigger’ would not exist in environments where voiced stops could not exist: that is, after voiceless stops and voiceless /s/. These are precisely the environments where Grimm’s Law does not take place.[1]