Thank you all for your suggestions. I ended up following the gist of graff's advice. I used AIFF (it turned out to be a more convenient choice than raw or WAV in this case) and I used brother Joost's Audio::SndFile to read in the AIFF files and add the silences as packed zeroes. (Granted, this does not produce a pregnant Cagean silence, fraught with meaning and anticipation, but for language flashcards it'll do...)

It all worked like a charm. Mind you, graff's post is the sum total of my education on these matters. (PCM??? Phase-controlled magnetron? Partido Comunista Mexicano? Please call me? ...). I have no idea of what I'm doing... (If someone happens to know of a particularly good "sound-for-cretins"-type book, let me know.)

Now I'm off to learn about LAME and what I can do with it... (Thanks to hawtin for the tip.) Cheers!