What phonemes could an intelligent fox produce? This assumes that their vocal tract is the same as real life foxes. They have human intelligence and have built a civilization that has reached the pre-Industrial era.
This might sound like a ridiculous answer, but search YouTube for talking dogs. Listen to the sounds they make and how close they are to human speech and you'll get a feel for what's possible with most canines.
The real problem is lips and cheeks, the inability to shape the inside of their mouths to articulate phonemes. Vocal cords just provide the vibrations, so they'd have to rely on a limited vowel set -- /a/, /e/, /ə/, /o/, and /u/ -- and a few simulated consonants if I can presume full use of a prehensile tongue: /s/, /t/, /h/, /k/, /d/, /g/, /l/, and /j/ -- possibly /n/ and /ŋ/ if they can express sound through their nasal cavity without causing other issues like a sneezing fit, which I'm not 100% sure of.
Ventriloquists work around some of the same limitations, so you can probably use that as a starting point.
This is only a partial answer and I feel I just did some research that you could have carried out yourself.
I Googled fox vocalisations and fox sounds. I won't list the results as there are many.
My finding is that a fox's vocalisations are very similar to those of a dog or wolf.
The easiest semi-consonant sound for them to make is approximately "w" as in "woof-woof"
They could approximate the difference between "aaa" and "ooo"
The language is going to be largely tonal. Foxes are able to change the pitch of their voice considerably.
The timbre will be very important. They can make pure sounds, harsh barking sounds, growls, etc.
If you allow them to make cat-like noises such as purring (they have a similar mouth and throat anatomy) then this increases the possibilities.
Repetition as in "yip - yip - yip" is a likely feature and at the very least could be used for counting.
I imagine that tail and ear posture would be very important. They could indicate anger, fear, pleasure and so on. These could be used simultaneously with vocalisations, either to modify meaning or possibly even to change grammar.