摘要：Linguistic avoidance - the conscious or unconscious non-use of a problematic language expression and/or the substitution of another - is difficult to demonstrate empirically in historical terms, as it constitutes a negative action that leaves no positive trace. However, all hope is not lost for the empiricist: the very conspicuousness of the problematic expression’s absence can in fact be measured across a time period. A perhaps unique case in point in English is the avoidance of number-word expressions such as five millions of individuals and, alternatively, five million individuals precisely at the point in history when it became unclear which of the two expressions was the correct one, i.e. when usage was shifting. At this point, newspapers in the US and the UK, following patterns that were similar but not contemporaneous, began to favor numerical expressions like 5,000,000 individuals. The timing of the shift corresponded to the time gap between US and UK newspapers in their respective gradual rejection of the now obsolescent (Obs) five millions of individuals in favor of the present-day English (PdE) construction five million individuals. Then, in both countries, newspapers largely discontinued the use of numerical expressions like 5,000,000 individuals when it had become clear that the PdE construction had prevailed. In other words, the numerical expression was conspicuously substituted to avoid choosing between the Obs and PdE versions of the numberword expression, and this avoidance left a clear empirical trace.
