Music: The international language?

• 09 July 2005
• From New Scientist Print Edition. Subscribe and get 4 free issues.
• Philip Ball
• Philip Ball is a consultant at Nature

THE Proms, London's annual classical music festival, invariably concludes with Elgar's Pomp and Circumstance march, which concert-goers accompany with a lusty rendition of the anthem Land of Hope and Glory. It is a champagne-fuelled, teary-eyed orgy of patriotism. There seems to be something distinctly English about that tune. But why?

Because music does indeed have a national character, says Aniruddh Patel of the Neurosciences Institute in San Diego, California. Put simply, music echoes speech.

Patel and his colleagues came to this conclusion after comparing the rhythms and pitch variations of English and French music and speech, focusing on the classical music of the late 19th and early 20th centuries. This was a time when many composers were actively seeking to express their own nationality, so national musical characteristics might be expected to be particularly prevalent. Also, by choosing a relatively recent period, the researchers did not have to worry about less well-known patterns of spoken language from the more distant past.

To create a suitable selection of music to compare, the researchers had to avoid some musical traps. Songs were excluded because lyrics inevitably force the music to comply at least loosely with their rhythms, and perhaps also with their changes in pitch. And for classical compositions based on a folk melody, such as Vaughan Williams's Fantasia on Greensleeves, it is likely that the original tune once accompanied a song. (In case you are wondering, the words of Land of Hope and Glory were added long after Elgar wrote Pomp and Circumstance. In fact the words required the addition of an extra note to the melody.) Compositions such as chorales and serenades whose titles implied a link to the vocal tradition were excluded too.

The team also filtered out other potentially complicating factors, including tunes whose rhythms were fixed by dances such as waltzes and gavottes, and compositions intended to mimic the style of a specific foreign tradition, such as Ravel's Bolero. Eventually they settled on a selection of 318 musical themes by 16 composers. In the English team were Elgar, Holst, Bax and Vaughan Williams, while the French selection included the dreamy fin-de-siècle tunes of Debussy, Ravel and Fauré.

With their choice complete, Patel and his colleague Joseph Daniele started their analysis by looking at rhythm. The rhythms of everyday speech are notoriously hard to codify, but there is a measure called the normalised pairwise variability index (nPVI), devised by linguists Esther Grabe of the University of Oxford and her colleagues Low Ee Ling and Francis Nolan. This index measures the variation in length between successive vowels in a spoken phrase. Grabe and others have shown that the average nPVI of British English is significantly higher than that of French. That is, adjacent vowels in English tend to have rather different durations - long and then short, say - whereas in French the durations are more similar.

So can the same pattern be seen in musical rhythm? When Patel and Daniele examined the patterning of note duration in their music samples, they saw no clear national bias in nPVI among individual composers: the ranges for Debussy and Fauré, for example, both lie within that spanned by Bax. But when they averaged the values for all composers from each country, they found a significant difference. As with speech, the nPVI of the English music was higher than that of the French selection.

A composer's nPVI doesn't always reflect our expectations. The index places Elgar as the most "English" of all the composers studied. But Holst, whose Jupiter theme from The Planets suite has been proposed as an alternative to Britain's dirge-like national anthem, has an nPVI score typical of French composers. By this criterion, his music is more French than Ravel's or Debussy's.

Pitch proved more difficult to analyse. Music tends to be "quantised" in pitch - composed from a discrete scale of notes. But while spoken English sounds as though it has more pitch variation than French, quantifying the changes in pitch is tricky.

Patel and his colleagues John Iversen and Jason Rosenberg turned to the work of linguist Piet Mertens of the Catholic University of Leuven (KUL), Belgium. Mertens has noted that the pitch perceived by listeners for a given syllable is largely defined by the average pitch of the syllable's vowel. So a spoken sentence can be reduced to a series of steps between these pitches.

Applying this to French and English, Patel's team found that although the average change in pitch between two syllables is the same in each language, there is more variation in English. And when the researchers analysed their selection of music, they found precisely the same distinction: the variation from the average interval tended to be greater for English composers.

Again, the ranges spanned by composers of each nationality overlapped considerably. Saint-Saëns, who is rhythmically very French, is melodically English, while the reverse is true of his fellow French composer Vincent d'Indy. Oddly, Holst falls squarely in the French camp on both counts, even though his music feels subjectively English. Patel cautions against drawing too strong a conclusion from this, though. "The Holst sample is small," he says, "and if more of his music was sampled, it might turn out that his melodic and rhythmic values would be higher."

But why should music share these acoustic similarities with speech? Patel thinks that the latter probably shapes the former. Composers, he suggests, absorb the speech patterns - the contours of pitch and rhythm - that they have heard since childhood, and unconsciously build these into their music. And people listen with the same conditioning, making British and French people particularly attuned and responsive to the sounds of Elgar and Debussy respectively.

Patel's latest work shows that the contingencies of history, as well as the exigencies of nationality, can play a part. Even though German is a language with a relatively high nPVI, the rhythms of German and Austrian music from the baroque and classical eras of the 17th and 18th centuries have a relatively low nPVI. Patel suggests that is because it was strongly influenced by the music of Italy - a country with a low-nPVI language. But starting in the 17th century, the nPVI value of German and Austrian musical rhythms increased steadily as the countries found their own musical voice. "It looks like the connection is not limited to English and French music of the period I studied," says Patel. "But more empirical work is needed."

His methods are now being used to study the rhythms in speech and music of other cultures, such as Welsh and Japanese. "One of our group's interests is in Thai music," says Patel, "since Thai has a high linguistic nPVI, and Thai culture has a well-developed classical instrumental musical tradition." What's more, Thai is a tonal language (see "Musilanguage"), and Patel is keen to discover if that is reflected in the country's music. He also hopes to study the music from cultures in which the music is not written down, such as those in many African countries, to find out whether the link with language patterns still emerges.

Patel thinks that this analysis might find practical applications. For example, his method of measuring rhythm and tone might allow a computer to assess the intonations of non-native speakers, so they can get feedback on their accents in language labs.

And could the new research help to guide composers of patriotic tunes? A deeper understanding of emotional responses to music probably awaits brain-imaging studies, but Patel's work is already suggestive: perhaps those mawkish renditions of Land of Hope and Glory at the Proms are simply playing on the patterns of speech that English people heard in the cradle.