Does constituent disambiguation facilitate compound interpretation?

In English, nominal compounding is a very productive means of word formation and hence of lexical expansion. Recent additions to the Oxford English Dictionary online include e.g. bucket list – 'a list of things that a person hopes to experience or achieve before they die (or kick the bucket)' – and trout pout – 'unnaturally swollen lips resulting from the injection of excessive collagen into the lips in a cosmetic procedure intended to enhance their appearance'. As these examples demonstrate, although the compound meaning bears some relation to the meanings of the constituent nouns, the relation between them is unexpressed and must therefore be inferred by the hearer or reader. In some cases, the constituent nouns themselves seem to suggest a default reading; for example a compound like polo umpire might be expected to lead to context-free consensus about the meaning, because an umpire officiates in the playing of games, and polo is a game. However, in many cases, compound interpretation depends on other sources of information, including not only deixis and explicit explanation, but also context and world knowledge. This study explores the role of immediate linguistic context in disambiguating newly-coined compounds. All noun-noun strings occurring within a sentence were extracted from the prose fiction section of the British National Corpus. This section was chosen to reduce the risk of selecting compounds whose interpretation would rely on deixis, specialised technical knowledge or highly time-specific information as might be expected in e.g. newspapers. To create a set of hapaxes, the sample was then reduced to items that occurred only once in the whole corpus, and not at all in ukWaC, a much larger corpus of more than 2 billion words. A random selection of 80 hapaxes was then examined in their sentential context. This led to the hypothesis that, in the majority of cases, the compound could be disambiguated on the basis of the