In his studies of New York City speech patterns William Labov came across the phenomenon of hypercorrection when he researched the use of particular variables. Following the regular class pattern we would expect a socioeconomically higher group to use more standard forms than the socioeconomic group directly below it. Labov found a deviation from the regular class pattern: Beginning with reading style, in the two most formal speech styles the lower middle-class exceeds the upper middle-class in the frequency of using the standard or prestige variant. The lower middle-class thus ‘hypercorrects’. According to the regular class pattern, the lower middle-class is always expected to score lower on the scale than the upper middle-class.
The same phenomenon was detected by Peter Trudgill in his studies of Norwich speech patterns. Here, however, it was the lower working-class showing hypercorrect speech behaviour for certain variables. This class exceeded the middle working-class in the frequency of using the prestige variant in the two most formal speech styles (reading passage style and word list style).
Hypercorrection constitutes an over-correct speech behaviour of the lower middle-class (USA) and the lower working-class (England) in favour of more standard variants, i.e. those forms which enjoy social prestige. Hypercorrection means that those variants are applied in an abnormally high frequency.
In line graphs hypercorrection is revealed in a typical cross-over pattern.