Technical Note 2: Understanding Sound Changes
Do sounds change in a regular fashion?
Probably the single most significant insight of 19th century investigations of the history of languages, driven largely by research into Indo-European languages, was that the sound changes that take place over time are indeed regular, subject only to a limited number of exceptions. General statements about the evolution of particular sounds or groups of similar sounds are indeed often termed ‘sound rules’ or ‘sound laws’ to emphasis their unambiguous nature.
It took decades of research for such a confident statement to be made.
Historically, it had been thought that there was quite a lot of variability between words in how sounds might evolve – some indeed believing that ‘each word has its own history’. For example, it was thought that the Greek and Latin words for ‘god’ were so close – ‘theos’ for Greek, the first element of ‘theology’, and ‘deus’ for Latin (as in the phrase ‘Deus ex Machina’) – that they had to be related, even though there were no other examples of a Greek initial ‘th’ corresponding to a Latin ‘d’.
Nowadays, it is accepted that the identical meaning is not a sufficient reason for such a conclusion. All linguists would now agree that a Greek initial ‘th’ never corresponds to a Latin ‘d’. In fact, the best-known Greek word formed from the same root as ‘deus’ is ‘Zeus’, the father of the Gods in the Greek pantheon, both words deriving from a root that seems to be about shining, as in the bright sky, and hence the Sky-God. (In later blogs, you will find the inherited English word honouring a Germanic God that shares the same root, and also the English borrowings that share the root of ‘theos’).
For identical reasons, we have to reject some other seemingly ‘obvious’ matches, such as English ‘day’ and the similar-sounding Latin word for ‘day’, visible in the opening syllable of ‘diurnal’ (like ‘deus’, from the root about shining), or English ‘have’ and the Latin verb with the same meaning that we see in ‘habeas corpus’ (literally, ‘you are to have the body [of the detained person brought to Court]’). As we shall discover, in inherited words an English initial ‘d’ never equates to a Latin ‘d’ or an initial ‘h’ to a Latin ‘h’.
As you can see from this, linguists give greater weight to sounds than to meanings in assessing how words relate. This is because sounds have been shown to evolve in a much more regular fashion than do meanings. Of course, many meanings are highly stable (for example, words about family relationships or numerals). But at the other extreme, think how rapidly and fundamentally adjectives can shift meanings, as in ‘wicked’ becoming in some cases a positive rather than a negative term, or ‘cool’ being used to describe what might also be called a ‘hot’ trend. Such changes are not amenable to ‘rules’, let alone ‘laws’.
There are however four main circumstances where full regularity of sound changes can be affected, leading to outcomes that vary from what might otherwise be expected.
- The first and by far the most widespread reason for the same sound to evolve in more than one way is its phonetic context. Here is an example.
In the first blog we saw that an English ‘f’ in initial position seemed to correspond to a ‘p’ in both Latin and Greek, and it turns out in this case English, along with the other Germanic languages, is very much the exception among other languages which descend from Proto-Indo-European. That does not guarantee that this relationship will hold in a different phonetic context. For example, it is common for consonants preceded by ‘s’ to evolve differently from those in initial position or between vowels. And indeed we find that Proto-Indo-European *sp remains unchanged in English. That’s why ‘spew’ and ‘sputum’, the first inherited and the second borrowed from Latin (it’s the past participle of Latin verb for ‘to spit’, so ‘what is spat out’), both begin ‘sp’. Both are descended from the same original root, which also began with *sp-, and in both Latin and English that combination of sounds happens to be conserved.
We will see many examples of the importance of phonetic context in later chapters. To keep the story manageable, we are going to pay most attention to consonants in initial position, including simple clusters like ‘sp’, as the sound-rules governing these are in general less complex than for vowels, or for consonants in other positions.
Text boxes in later blogs will give you background on four less predictable changes that reflect phonetic context. These (and similar) text boxes will also be included in a Technical Note which will be built up as the boxes appear in subsequent blogs, so that they are easy for you to refer to later.
- A second reason why seemingly anomalous sound changes may occur is because languages frequently have varying dialects even while maintaining mutual intelligibility. If a sound is pronounced differently between dialects and a word from a dialect is then imported into the main language itself, the sound in that particular word may show an evolution that contrasts with the normal change (or stability) of the sound in question.
For example, some South-Western dialects of English changed initial ‘f’ to ‘v’. Modern English, which is based on a dialect with no such change, happens to have borrowed the word for a female fox from the West Country, which is why we see the otherwise inexplicable pairing of ‘fox’ and ‘vixen’.
The same can happen between languages in close proximity: we have already seen in the last blog examples of Old Norse and Dutch words which have been absorbed into English, conserving ‘sk’-sounds that would have been palatalized to ‘sh’-sounds in an inherited word.
- A third reason is usually termed ‘analogy’.
This covers a wide range of changes that have in common the overriding of an otherwise ‘regular’ pronunciation by something different, driven by some phonetic, grammatical or semantic reasoning.
For example, let’s take the Proto-Indo-European words for ‘four’ and ‘five’.
The original word for ‘four’ started with a sound that is broadly conserved in Latin, and whose nature is evident from English borrowings from Latin such as ‘quarter’, ‘quart’ and ‘quartet’: in fact it is the labio-velar conventionally notated as *kʷ. The original word for ‘five’ started with a *p as reflected in our borrowings from Greek like ‘pentagon’.
While we have already seen evidence that a PIE *p in initial position is likely to produce an ‘f’ in English, as in ‘five’, the ‘f’-sound in ‘four’ cannot be explained by any ‘sound rule’. So the initial sound of ‘four’ is generally thought to have developed, originally in the common Germanic language mentioned in the previous chapter, by analogy with the number five, no doubt as a consequence of people counting numbers in sequence.
- Finally, words that are based on actual sounds may well resist sound changes.
‘Cuckoo’ is a well-known example. This bird name sounds very similar in most modern Germanic, Romance and Slav languages, whereas, as we shall see later, an inherited *k would be expected to have a very different outcome in each of these three groups. This type of exception is unusual and generally obvious.
To conclude this Technical Note, here is another example of apparently contradictory meanings from a single original concept. The English word ‘guest’ (a borrowing from Old Norse) is related to ‘hostile’(a borrowing from Latin). What might be the original concept that gave rise to two such oppositional terms?
Answer in Blog 6.