杏吧原创

As languages evolve over centuries and millennia, the most frequently used words tend to remain unaltered, while rarer words are more likely to change.

This tendency was long suspected, but has now been proven rigorously for the first time by two new studies. The results show that the tools of evolutionary biology can be applied to study the evolution of cultural artefacts like language.

Evolutionary biologist and his colleagues at the University of Reading, UK, used a comparative database of Indo-European languages to trace the words used to express 200 different meanings in 87 different languages.

A few meanings, for example the number 鈥渢wo鈥�, used related words, known as 鈥渃ognates鈥�, in every language they studied. Some meanings, for example 鈥渢ail鈥�, were expressed by many apparently unrelated words across the 87 languages, indicating great change in the 6,000 to 10,000 years since the origin of Indo-European languages.

Cultural coordinates

The researchers then turned to large databases of four modern Indo-European languages 鈥� English, Russian, Spanish and Greek 鈥� to determine the frequency of use of the 200 meanings in these languages. They found the more frequently the meaning is used in speech, the less change in the words used to express it.

This suggests that more important linguistic elements are more resistant to change. Supporting this, Pagel also found that less important parts of common speech, such as conjunctions and prepositions, are more likely to change than crucial ones such as numbers, pronouns, and so-called 鈥渟pecial adverbs鈥� like 鈥渨ho鈥�, 鈥渨hat鈥�, 鈥渨here鈥�, 鈥渨hen鈥�, 鈥渉ow鈥� and 鈥渘ot鈥�.

鈥淲e use those words over and over,鈥� says Pagel. 鈥淚t鈥檚 what I like to think of as coordinates: who did what to whom, and how many did they do it with? And those words stay put. They just don鈥檛 evolve, while the other ones seem to go all over the place.鈥�

The researchers were able to mathematically predict the likely 鈥渕utation rate鈥� for each word, based on its frequency. The most frequently used words, they predict, are likely to remain stable for over 10,000 years, making these cultural artifacts, or 鈥渕emes鈥�, more stable than some genes.

鈥楻egularised鈥� verbs

A second study, led by , a mathematician at Harvard University, shows a similar effect within a single language, English. Lieberman鈥檚 colleagues Joe Jackson and Tina Tang combed through grammar textbooks on Old English 鈥� the language of Beowulf 鈥� to compile a list of 177 irregular Old English verbs.

These are verbs, such as 鈥渟ing鈥� do not follow modern English鈥檚 standard rule for forming past tenses by adding 鈥�-ed鈥�, in this case simply becoming 鈥渟ang鈥� or 鈥渟ung鈥�. Of these verbs, 145 were found to still be irregular in Middle English, and 98 remain irregular today, reflecting a tendency of verbs to become 鈥渞egularised鈥� over time.

When Lieberman鈥檚 team included information on the frequency of use of each verb, a clear pattern emerged. The most frequently used verbs (such as 鈥渂e鈥�, 鈥渉ave鈥�, 鈥渃ome鈥�, 鈥済o鈥� and 鈥渢ake鈥�) remained irregular. The less often a verb is used, the more likely it was to have been regularised. Of the rarest verbs in their list, including 鈥渂ide鈥�, 鈥渄elve鈥�, 鈥渉ew鈥�, 鈥渟nip鈥� and 鈥渨reak鈥�, 91% have regularised over the past 1200 years.

The researchers even think they know the next irregular verb to make the switch: 鈥渨ed鈥�. Officially, the past tense of the verb stays as 鈥渨ed鈥�, as in 鈥渘ewly wed鈥�. But do not get too wedded to that notion: the -ed suffixed version is already alive and well.

Journal reference: (DOI: 10.1038/nature06176, DOI: 10.1038/nature06137)