I'm pretty pleased with the tree I put together for the distinctive features of English, but I still have a couple concerns (at least) that didn't get covered as well as I'd hoped. There are two apparent classes of sounds not adequately described by my tree: palatals and retracted, optionally half-round sounds that I'll call "dark".
I do use a feature [pal] to separate /S/, /Z/, /y/, and the front vowels from the rest of the phonemes, but I feel like I'm leaving one sound out. Like, /y/, /r/ also turns alveolars into postalveolars. This is reflected in the spelling of words like "shred" and "shrill" (but not borrowings like "Sri Lanka"), but it typically goes unnoticed in the (presumed) phoneme sequences /tr/ and /dr/. Only when a speaker tries to pronounce a nonce word like "chree" (only to find it identical to "tree") does this palatalization process become obvious.
The problem with this observation is that, in my current mode of analysis, I'm forced to make a choice between describing /r/ as [-pal] and inventing some new feature for /r/ and leaving /y/ unvalued for [pal], /w/ unvalued for [rnd], or /l/ unvalued for [lat]. None of these options sound very good to me. There could very well be some redundancy in English speakers' mental representation of the phonemes, but I'd prefer to avoid that in the formal representation (a preference shared by most, if not all, scientists).
The second problem has not so much to do with the patterning of sounds in words as with the sounds themselves. The most notable characteristic of the "dark" sounds is that a slight degree of lip rounding is normal but not obligatory. Of the consonants, /S/, /Z/, /r/, and possibly /l/ fall into this category. Of the vowels, /u/ is obviously dark, but /A/ could be as well. (/A/ is typically unrounded, but this could be a result of the low tongue position.) /w/ and /o/ might be [+dark], but they're also [+rnd], so it's hard to tell. What do all these sounds have in common? In my tree representation, not much. The most notable inconsistency is that for /S/ and /Z/, [+pal] corresponds to [+dark], but for /r/ and /u/, it's the reverse.
If there are any phonologists (or aspiring phonologists) out there who have a few minutes, your comments on these problems would be greatly appreciated. Just click the (very tiny and easy to miss) green icon below or the "0 Comments" link that follows it. You don't have to have a Google account; you can even be anonymous if you want. Please!