I was wondering if any studies have been done in corpus linguistics which look at word length (number of phonemes per word) as a function of the word's frequency of use? Thanks!
EDIT: Never mind, Zipf wrote about it!
Whoops, didn't know I was allowed to answer my own question.
Zipf wrote about exactly this! I knew that he'd formulated Zipf's Law (the relative frequency of a word in a language is inversely proportional to its rank in frequency, so the most common word is used twice as much as the second most common word, three times as much as the third most common, and so on), but I wasn't aware that he also addressed the question of length as a function of frequency: the more frequent a word is in usage, the shorter its length will typically be (as measured in number of phonemes). Greenberg described the same thing with his feature hierarchy Singular > Plural > Dual, where the most frequent forms (singular) are on average expressed with a smaller number of phonemes than the less frequent forms (plural and dual, with dual on average being the least frequent and expressed with the highest number of phonemes). Rosch made a similar observation that 'basic level categories' have the highest semantic generality and are typically shortest in length (e.g. 'dog' is shorter than 'greyhound'). Hawkins takes the correlation between length and frequency as evidence for his parsing constraint Minimize Forms, which he argues is used by the speaker to select the most "efficient" (a concept he defines and quantifies) utterance among competitors.
This is the sort of finding I expected and it makes a lot of intuitive sense, but I wasn't sure if anyone had actually formally described the phenomenon. Turns out several people have! Zipf and to a lesser extent Hawkins are the only sources I've found which gather empirical and statistically testable data about it, but it's been at least noted by several linguists. I'd really love to see what kinds of generalizations can or cannot extend to polysynthetic languages, where the number of unique words in performance data is lower than in more analytic languages, but I can't find any information on that question, sadly.
Sources I've found which explicitly describe length as a function of relative frequency:
Greenberg , J. H. (1966), Language Universals, with Special Reference to Feature Hierarchies (The Hague: Mouton).
Hawkins, J. (2004), Efficiency and Complexity in Grammars (New York: Oxford University Press)
Rosch , E. (1978), ‘Principles of Categorization’, in E. Rosch and L. L. Lloyd (eds.), Cognition and Categorization (Hillsdale, NJ: Erlbaum), 27–48.
Zipf , G. (1949), Human Behavior and the Principle of Least Effort (New York: Hafner).