Is lexical stress mostly consistent across accents of Standard English?

According to Wikipedia, lexical stress in Standard English* is "phonemic" (whatever they think they mean by that), using the minimal pair insight/incite as an example. My hypothesis is that, across Standard English, lexical stress is mostly consistent. I am however not very familiar with the many different English accents, so I wonder whether this is true.

Note that I'm specifically not considering prosodic stress, or how stress is realised (inflection, vowel length, etc.) in different accents. I'm only asking about one part of the puzzle: if Standard English speakers are asked to mark the stress of words, would they mostly agree**, regardless of their accent?

* I know, this is a vague term. I'm mostly concerned about what's called "native English" in North America, the British Isles, Australia, and New Zealand. If you believe that's too narrow of a definition, please don't hesitate to comment on that.

** If you need a metric for agreement: the correlation of the placement of primary lexical stress for each word (only distinguished by exact spelling and part of speech, to account for details like heteronyms and unexpected spellings), weighed according to that word's frequency (in a corpus of your choice, which usually also comes with a preferred PoS inventory).

Solution 1:

I will defer to someone who can provide more explicit references, but I'm going to say "yes, it is mostly consistent", for two reasons:

• "J. C. Wells: Accents of English" (by John C. Wells, the famous British phonetician), in summarizing the phonetic features of the various major accents of English, mentions word stress only twice:

  • In the West Indies, "Words such as realize are stressed on the last syllable."
  • In India, "word stress sometimes deviates from that of other accents (acquire [ˈɛkʋaɪɹ])."

  
  Additionally, it mentions that RP has a "[w]eak suffix in -ary: momentary /ˈməʊməntrɪ/; but not in -ile: hostile /ˈhɒstaɪl/". Under some analyses, these would be differences in secondary stress between different accents; under other analyses, unreduced vowels are not inherently considered stressed, so these would just be differences in vowel reduction, not in stress. (Wells apparently subscribes to the latter camp, else he would presumably have written /ˈhɒsˌtaɪl/.)

• Anecdotally: when I've listened to non-American accents, I've frequently noticed differences in stress placement in various words, such as protester and elsewhere, but my experience is that these are by far the exception rather than the rule.

Of course, even within an accent, there is often variation between different speakers, especially in less-common words.