American pronunciation of "with"

Question

What is the most common American way of pronouncing "with"?
I'm asking specifically about "th" combinations - dictionaries give both the unvoiced (wɪθ)
and the voiced (wɪð) ones?

Personally, I've heard it pronounced most of the times as wɪð (note: I'm not American) and this is why I'm constantly surprised when online pronunciation tutorials encourage students to pronounce it unvoiced.

So, I guess my questions are:

1. When a native American speaker pronounce "with", does he usually just "absorb" one pronunciation form and stick with it for the rest of his life or the pronunciation can vary depending on the context?
   
2. If the answer to the last question is "yes" (speakers stick to one pronunciation form) - which regions in the states pronounce it voiced/unvoiced?
   
3. Again - if speakers don't change pronunciation depending on the context, what is the most common way of pronouncing "with"?
   
4. What is the "correct" pronunciation according to the notorious "General American" pronunciation?

Answer 1

In "General American" the pronunciation varies depending on the letter that follows:

In front of consonants, it usually sounds like wɪθ: "go with (wɪθ) Nancy"

with the possible exception of "L", where it usually sounds closer to wɪð: "go with (wɪð) Linda"

In front of vowels, it's usually wɪð: "go with (wɪð) Adam"

I can't speak for any of the various regional accents or dialects, but where I live the regional speech patterns tend to emphasize consonants, so it frequently comes out as wɪð when speaking slowly. But if they're talking normal to fast, they still tend to revert to the above rules.

Answer 2

(i). This picture provides an answer for 4). AmE stands for General American, BrE for British RP.

Source: "J C Wells, Longman English Pronunciation Dictionary, New Edition"

(ii). For 1), let's consider a native speaker A chooses the voiced /ð/ instead of voiceless /θ/ for <with>, then A can also very likely to pronounce <with> using /θ/. Why is that? Because in English a voiced consonant (or lenis consonant to be more precise) is frequently devoiced (e.g. it sits right before a voiceless consonant (this phenomenon is called energy assimilation), or it's before a pause (if it's at end of sentence, for example), or it's after a pause (at the beginning of speech, for example)), except in some special environments (e.g. the sound is trapped between two voiced consonants or vowels or between a voiced consonant and vowel). On other hand, if a native speaker B chooses /θ/, he will definitely stick with it for the rest of his life (except some special "affected" occasions, for example, where B tries to be like A). Why is that? Because in English, energy assimilation doesn't apply for the case where a voiced consonant turns an adjacent voiceless consonant into voiced one.

(iii). 2) is answered by (ii) already.

(iv). 3) is answered by the combination of (i) and (ii).

Notes:

• I'm not American.
• Most of what I wrote here can be found in "B. Collins et al.,Practical English Phonetics and Phonology, Routledge" or "J C Wells, Longman English Pronunciation Dictionary, New Edition".

Answer 3

I'm a younger speaker from Chicago. I think I speak with a fairly nonspecific General American or Midwestern accent. I pronounce the word "with," pretty invariably, with a voiceless interdental fricative in the syllable coda: [wɪθ]. I think this is pretty standard. In some cases, possibly intervocalically or in rapid speech, the fricative may take on the voicing of surrounding sounds.

[ð] certainly does not sound wrong to my ears, but it does not necessarily sound standard either. I associate pronunciations like these (quite possibly erroneously) with Southern speakers as well as older people. In addition, it also seems like a sort of "teacherly" pronunciation for use in educational contexts, which may be why you're under the impression that it's more common. A very similar example is the abbreviated title "Ms." The same speakers I mentioned are liable to pronouncing it as [mɪz] when in fact the general standard is really [mɪs].

A previous answer argues that the fricative in "with" is usually voiced before vowels. I don't think that's generally the case - at least in variety of English. To explain this discrepancy, it might be worthwhile to bring up the increasing prevalence of hard attack (epenthesis of a glottal stop at the onset of words beginning with vowels in connected speech) among younger speakers. For example, I would pronounce "with Adam" as [wɪθ ˈʔæ.ɾəm].