Can One Word Tell Where You Are From? It Has Decided Who Lived

The word is taken from a passage in the Book of Judges describing a conflict between the Gileadites and the Ephraimites, in which the victors held the fords of the Jordan and required anybody wishing to cross to say a particular word. The Ephraimite variety of the language lacked the initial sound, so its speakers produced a slightly different word, and the text states that forty two thousand were killed on that evidence. Whether the number is reliable is beside the point. The passage records the principle exactly, that a feature of speech acquired in childhood and impossible to fake under pressure can be used as an identity test.

The reason speakers cannot simply produce the sound is well understood:

• •Infants learn to hear the distinctions their own language uses
• •Distinctions their language ignores become genuinely hard to hear
• •A sound absent from the language is not practised by the mouth
• •Adults substitute the nearest sound they do have, without noticing
• •Stress and fear make careful imitation harder still
• •Even fluent later learners usually keep some traces

The pattern recurs through history with dismaying regularity, which suggests it is discovered independently rather than copied. Flemish rebels in Bruges in 1302 reportedly used a phrase that French speakers could not manage. Rebels in Sicily in 1282 are said to have used a similar test. Frisians used a sentence that outsiders could not pronounce. In the Dominican Republic in 1937 the pronunciation of the Spanish word for parsley was used to identify Haitian speakers of Creole, with tens of thousands killed. During the Second World War, American sentries used words containing sounds absent from German and Japanese. The same device appears in conflicts into the present.

Treating pronunciation as proof of origin is unreliable in both directions, and the historical examples illustrate the failure clearly. People who genuinely belong may have grown up elsewhere, may have moved as children, may have a speech difficulty, or may simply speak a variety the testers have not encountered, and every one of those produces a false accusation. People who do not belong may have acquired the accent through long residence, through a parent, or through deliberate practice, which produces a false pass. The test also assumes that the categories it sorts people into are sharp, when speech varies continuously across a region. Accounts of the 1937 killings record exactly these failures.

Most uses are not violent and are simply the ordinary business of placing somebody socially, which happens continuously and mostly below conscious notice. The pronunciation of a local place name marks residents from visitors reliably, since the spelling rarely predicts it. Vocabulary choices for the same object map onto regions and onto class. Certain grammatical habits are read as markers of education. Professional groups use terms that identify insiders. All of these function the same way, offering a cheap and quick signal of who somebody is, and all of them carry the same risk, which is that the signal is about upbringing rather than about anything that matters.

A speech feature learned in infancy cannot be faked reliably under pressure, since a sound absent from somebody's first language is genuinely hard to hear and to produce. That has made pronunciation a test of belonging from the Book of Judges through Bruges in 1302 to the Dominican Republic in 1937. The everyday version places people socially by place names, vocabulary and grammar, on evidence about upbringing alone.