/ɪ/ sound when not stressed

Question

I've seen that some words in English are pronounced with the /ɪ/ sound when the vowel is not stressed.

Some examples include: pocket /ˈpɒkɪt/, comet /ˈkɒmɪt/. But hundred /ˈhʌndrəd/. Is there any pronunciation rule here when the words end in /t/?

Answer 1

Some people do use /ɪ/ in "hundred", according to the American Heritage Dictionary (hŭnʹdrĭd.) But it's the only dictionary I've found that says this. (According to snailplane, the Longman Pronunciation Dictionary also records this as a possible pronunciation—I don't have access to this, so I can't say any more about that.)

As John Lawler says, unstressed vowels are generally more variable in pronunciation than stressed vowels. Some speakers have almost no phonemic distinction between /ɪ/ and /ə/ as reduced vowels (vowels with neither primary nor secondary stress). This is called the "weak vowel merger" or "rabbit-abbot merger". Like many mergers, it exists at various stages: some speakers might have these sounds merged in perception but not in production, or vice versa. This merger is very common in American English. I don't know much about its prevalence in British English; I believe it is not unknown there. However, the most famous dialect of British English that has been described in detail, "Received Pronunciation" ("RP"), did not have this merger. Even in non-merging dialects like RP, however, it seems there is some variability between /ɪ/ and /ə/ in certain words.

Wikipedia mentions "roses" and "Rosa's" as possible homophones, but in fact the situation for these is more complicated. Many speakers who have a merger in general between /ɪ/ and /ə/ still use a different type of schwa word-finally, and this is maintained in inflected forms of words that end in schwa. We might represent the word-final variant as [ɐ], since it is generally realized as phonetically lower than non-final schwa. For speakers like this, there is a fairly clear, if small phonemic contrast between "roses" ([roʊsəz~roʊsɪz]) and "Rosa's" ([roʊsɐz]) even though there is no clear phonemic contrast between "rabbit" ([ræbət~ræbɪt]) and "abbot" ([æbət~æbɪt]). I'm one of these speakers ("rabbit"-"abbot" merged, but not "roses"-"Rosa's" merged). I go into a little more detail about the phonetics of this in my answer here: Exodus word pronounciation - Why it is different from spelling

Final consonants can affect the articulation of a preceding reduced vowel, but I don't think the effect is strong enough to be of much use in determining the identity of a reduced vowel in dialects where this is phonemic. It's more relevant for historical development, and allophony in dialects where the distinction is not phonemic. The concept of a "phonemic" distinction in fact implies that the vowel used cannot be predicted entirely from its phonetic or phonological context. The word "abbot" that I mentioned above is a counterexample to any rule saying something like "/ɪ/ must be used before /t/".

Whether a word has /ɪ/ or /ə/ also does not seem to be fully explainable by spelling or even etymology. The word "hundred", which seems to generally have /ə/ in non-merged dialects, contrasts with words like "hatred", "kindred" and "naked" which are all listed as having /ɪ/ by the OED. According to the phonetician John Wells, the word chicken is apparently generally pronounced /ˈtʃɪkɪn/, in contrast to words like thicken, quicken, stricken, sicken which all end in /ən/. Wells says "I have no idea why that should be the case. Given the identical spelling and the similar morphology (a fairly transparent -en suffix), you would expect them to have the same vowels."

So, it seems the only reliable way to determine the pronunciation is to look in a dictionary that records the pronunciation of a dialect that maintains the distinction between /ɪ/ and /ə/ as weak vowels. I have heard that the Longman Pronunciation Dictionary that snailplane mentioned, which was actually written by John Wells, is a good resource for questions like this.