A word or two?

Résumé

The tendency for people to write compounds as two separate words, i.e. decompounding, is, for Scandinavian, often attributed to influence from English. However, English writers also both accidentally compound and decompound words. This article introduces serendipity as a statistical signal of surprise, i.e. deviance from expectations. Examples show that this measure can decide many cases of accidental compounding or decompounding by estimating which alternative is over-represented. Interestingly, the least frequent alternative can be clearly over-represented, thus providing a signal that is different from probability estimates, and linked to change in probability.