For now, Spem has the following phonemes (sounds that are treated by the language as indivisible, and in this case all of them are). Where there's not an unambiguous letter I'll indicate the sound by giving an English example word and highlighting only the relevant letters.
I'll give the IPA symbols for each non-English letter used along with its Unicode hex number. In the case that the language needs to be written in pure ASCII, I plan to also define a transliteration scheme.
|symbol||English example if necessary||Unicode hex number|
|ɵ||put. Note this is a different sound from cup. The rhotacized version in were does not get a separate symbol; I'm deciding not to distinguish the two.||U+0275|
|o||This is the o sound typical of other languages, but pronouncing it as in note is acceptable. I think we should try to prononuce it the way other languages do though because that's only one sound and so presumably faster.|
|y||y (note the relationship to i)|
|w||w (note the relationship to u)|
|j||vision (not joke; that's dj)|
|θ||thing (not this)||U+03B8|
|ŋ||sing. Spem does not discriminate against this minority sound; it appears in some situations where it's not allowed to in English like at the beginning of a word. I've known some people who seemed incapable of pronouncing this, but I know we English-speakers all can because the sound is in English.||U+014B|
You can see I've chosen most of the same sounds as English, and the ones I've left out are ones that are super rare in other languages. There are a few digraphs; in theory these are composed of the sounds of their letters, but I'll list them anyway:
e + i = way
ɑ + i = hi
ʌ + i = tight (note the difference from hi)
a + u = ow
o + i = boy
t + ʃ = ch
d + j = jet
Any combination of vowels not listed there, such as -ui-, is to be prononuced as two syllables.
Pronunciation guidelines and open questions
What r sound should I go with? I like the English one more aesthetically, but I know it's very hard for people to pronounce that didn't grow up speaking English and even English-speaking children often master this sound last, which indicates that it's not a good sound for a designed human language. I could use the Japanese one, which doesn't seem to be as hard for English-speakers to learn as vice versa (the English 'weak d' sound like in pedal is very close to it), and I find it natural to substitute that sound during song.
No squishing duplicate sounds
When you have a word that starts with the same sound as the previous word ends with, they are to be prononuced separately. Squashing them together as if they're a single letter will most likely result in a torrent of ambiguities because I'm not making any effort to avoid it.
Voiceless stops stay voiceless
In English, voiceless stops become voiced if they're between a fricative and a vowel. For example, most people don't notice it but spell is actually pronounced as sbell. Spem should not be pronounced this way; it limits our phenome space. If a word is supposed to be pronouned sbell, it will be written that way. (And hence the name of the language is Spem, not Sbem.)
Intervocalic unstressed stops
Something similar happens to intervocalic (between two vowels) unstressed stops. icky when said by an English speaker is pronounced as something I'd write as ikgy.
I'd like to do away with that practice as well because it's inconsistent and wrong in principle, but that it seems to happen in every natural language I've studied suggests that it might be that this is inherently easier for the human mouth and I shouldn't try to force the proper pronunciation. Thoughts? (I don't think allowing this can reduce the phenome space.)
In English l is pronounced differently if it's before a consonant or at the end of the word. It's normally the dental lateral approximant (Wikipedia mislabels it as alveolar) but if you pronounce a word like 'milk' your tongue probably never touches your teeth. I have the same feelings and the same ambivalence about this as I do about intervocalic unstressed stops.
I don't think we should ever use emphasis to distinguish anything otherwise identical, because emphasis is used to draw attention to a difference, so if it's also regarded as important by the word it can cause confusion. For example back when I studied Latin I asked someone if I was pronouncing a word right, and I stressed the syllable I was unsure about but that wasn't the syllable that was supposed to be stressed so he corrected my emphasis. I of course knew that wasn't supposed to be the stressed syllable.
I do really, really like the thought of using emphasis to completely elide the need for grammatical questions. We could cut syllables and reap so much benefit. But at the same time, emphasis isn't as clear as syllables; it's more likely to be misheard, and if it makes the entire difference between a statement and a question, I don't like that prospect. Let me know what you think in the comments.