Foxtrot by Bill Amend
Students often don't believe me when I say that if your linguistic analysis is getting really messy and complicated, it's highly likely that you overlooked something that would make it simpler. But it's true. For example, if you are trying to describe when a phoneme is pronounced as one of its allophones and need five different descriptions for one set of environments, you're most likely overlooking what the environments have in common. The other day, my History of the English language class had a data set to analyze in order to determine when the letter f was pronounced like a v. Some of the words were fifta, fif, fiFel, and oFnas (the two fs that are bolded and capitalized are the only two that are pronounced like a v). As a student of linguistics, you should expect that a generalization can be drawn and that you will not have to answer "between these specific letters." Look for what the environments have in common: i, e, o, and n are all voiced phonemes. In fact, the only time f is pronounced like a v is when it occurs between two voiced phonemes.
One of the things I love about linguistics is how methodical it is--there is a method to the madness of language if you look for it. The voicing in the above example changed to match the voicing of the surrounding segments--it makes sense to do that. It wouldn't make much sense to say that the voiced v occurs only in the environments i_e and o_n. That wouldn't tell us what is special about those particular letters to change the pronunciation of another. Generalizing gives us the opportunity to say, "Ah, being surrounded by voiced sounds causes the voiceless phoneme to become voiced." Again, it makes sense. There is a method to the madness. If you're doing a linguistic analysis, go with the KISS method: Keep It Simple, Stupid.