Spoken language changes constantly. As a result, many German dialects are increasingly getting closer to the standard accent. Even the Queen’s pronunciation is changing: An interview with language researcher Jonathan Harrington.
Jonathan Harrington, Chair of Phonetics and Digital Speech Processing and Director of the institute of the same name at LMU, studies sound change from a variety of angles. He also investigates the conditions under which sound change occurs, drawing on the capabilities of artificial intelligence in the course of his research. He is one of the few scientists in the world to have received three ERC Advanced Grants.
Using Queen Elizabeth II’s Christmas Day speeches as an example, you have studied how her pronunciation has changed over the decades. What changes have you identified?
Jonathan Harrington: The changes are to the standard accent of England, known as Received Pronunciation (RP). Between 1950 and 1980, the Queen’s pronunciation shifted away from a more aristocratic form of RP known as U-RP or upper-class RP toward mainstream RP that is more typical of the middle classes. This is especially so for her vowels.
What is meant by upper-class RP?
It is an old-fashioned, aristocratic form of the standard accent of England in which the ‘o’ in words like ‘lost’ and ‘often’ is pronounced identically to the ‘aw’ in ‘law’. A form of mainstream RP can be heard in many BBC newsreaders today and was typical of the middle classes around 1980. In the year 2000, we published an article in Nature in which we showed acoustically that the Queen’s vowels from the 1980s were closer to mainstream RP than they had been in the 1950s. The ‘u’ sound in ‘moon’ or ‘blue’ in the 1950s was produced with the tongue further back in the mouth – which is why the English ‘do’ and the German ‘du’ were more similar back then than in 1980. Another example: The word-final vowel ‘-y’ in ‘city’ or ‘pretty’ used to be shorter and more like the ‘i’ in the standard German ‘Mitte’. Today it is longer and more similar to the ‘ie’ in ‘Miete’.
What causes this kind of sound change?
One reason is contact between people. In England, social change saw middle-class and sometimes even working-class people taking higher professional positions, with the result that the Queen would have had more contact with them than in the past. Imitation also plays an important role. Experiments in phonetics and speech processing show that we imitate each other when we talk. Having said that, the changes in the Queen’s pronunciation are minimal. You can’t detect them from one year to the next, so she could not have made them intentionally – for example to get closer to the pronunciation of her people.
Mobility and contact between people make sound change more likely. England experienced greater mobility between social strata in the sixties and seventies. But in Germany, sound change in recent decades has happened more as a consequence of the decline of dialects.
Do similar developments occur in other languages?
Mobility and contact between people make sound change more likely. England experienced greater mobility between social strata in the sixties and seventies. But in Germany, sound change in recent decades has happened more as a consequence of the decline of dialects. In particular, we note that standard German of the kind spoken by newscasters at public broadcasting companies, for example, is increasingly influencing dialects. Elementary school children near Altötting whom we recorded once a year for four years still clearly spoke the West Middle Bavarian dialect. But some of their vowels increasingly reflected characteristics of standard German.
New dialects also emerge when people are isolated. This explains the development of new dialects in the periods when America, Australia and New Zealand were colonized, for example. Australian English is heavily shaped by Cockney, which is still spoken in London, and to some extent by Irish English, because these groups accounted for the majority of convicts in the 18th century. And the phonetics of a group referred to as ‘winterers’ – researchers whose pronunciation we recorded before and after a long winter of isolation in the Antarctic – had converged slightly after three months.
What are you studying in your new ERC project ‘SoundAct’?
We want to find out why sound change happens in one language or dialect but not in another. Why have ‘beer’ and ‘bear’ become what are called homophones – different words with the same pronunciation – in New Zealand English but not in Australian English? Whether or not sound change actually takes place depends on the random stochastic interplay of at least three key factors. Alongside cognitive factors, which we analyze in part with computer models, and social aspects, there are also phonetic considerations: Accordingly, we analyze speech sounds that are susceptible to sound change. Umlaut in German is one such example: The word ‘mächtig’ has its roots in the Old High German ‘mahtîg’. Over time, the more open mouth position in /a/ has adjusted to the more closed mouth position for the subsequent /i/ in ‘mahtîg’, causing a mouth position between the two to become established in the word stem.
However, to understand the origin of a sound change, we would need to be aware of these factors before and shortly after the start of the sound change. But this is challenging, because sound change is imperceptible in the initial phase. We work around this problem by analyzing two related dialects that happen to demonstrate a similar sound change that has been completed in the one language but not in the other. For example, we can no longer actively investigate the sound change toward the German umlaut, because this process is finished. But a comparable sound that is still in progress exists in southern Italian dialects around Calabria.
Using a combination of imitation experiments and computer models, we attempt to transform the dialect in which the sound change has not yet taken place, or has only just begun, into a dialect in which the sound change has been completed. And we try to find the factors that facilitate this transformation. To find a general solution, we study pairs of dialects from completely different language groups. In addition to Italo-Romance dialects, for instance, we will also be analysing northern Greek and Japanese dialects, as well as Bantu dialects that are found in Kenya.
What kind of misunderstandings can arise when pronunciation changes?
One example is New Zealand English, where what is known as a vowel chain shift has occurred over the past 70 years. For outsiders, ‘had’ sounds like ‘head’, but ‘head’ sounds like ‘hid’ and ‘bid’ like ‘bud’. Moreover, there is no longer any distinction in the pronunciation of ‘hair’ and ‘here’ or ‘chair’ and ‘cheer’. That is why the name of the national airline sounds not like ‘Air New Zealand’ but like ‘Ear New Zealand’ for most outsiders, even for Australians. Fortunately, though, differences in meaning can often be recognized from the context.
In British English, this is the only way you can today distinguish between ‘not’ and ‘knot’. Four hundred years ago – in Shakespeare’s day – ‘knot’ was still pronounced with an audible ‘k’, like ‘Knoten’ in German.