S/Key isn't intended for "spoken phrases" but, rather, *textual* phrases. The "words" that S/Key employs aren't chosen to be unambiguous in spoken context -- especially over a (perhaps) noisey channel. Instead, they are expected to be "typed in". They are just words because words are easier to transport from a slip of paper through your brain and into your fingertips.
A user prepares a list of the next N "single use passphrases" and, typically, carries them on a sheet of paper in his/her wallet (they aren't very easy to commit to memory; and, the cost/benefit tradeoff is lousy -- why memorize something that you will use EXACTLY ONCE??). Once each password is used, it is considered "consumed" and used (on the server) to indirectly indicate the
*next* valid password (which is present on the user's list).As such, the passwords can be damn near anything chosen from any "dictionary". RFC1760 indicates a *particular* dictionary that doesn't happen to be very well suited to memorization *or* disambiguation.
E.g., sample words: flo ira jot mao ... And, no rules regarding which words are acceptable in which "positions" (indeed, they are position independant; so "flo ira jot" and "ira jot flo" are each valid -- AND DIFFERENT!)
When choosing words for *spoken* transmission, you need to apply intelligibility tests (at the very least) to ensure you haven't allowed two "easily misheard" words to enter into the same context.
See, for example, To use the chart, imagine listening to someone speaking the words on any given line. Would you be able to correctly identify *which* word was spoken -- regardless of the characteristics of the channel, speaker's "accent", etc. (recall, there is no *context* associated with these words -- at least not in S/Key -- so nothing to help you decide whether "wig" or "rig" is being spoken!)
This is the value of the scheme I indicated elsewhere that causes the dictionary to VARY based on word position in the "phrase". E.g., if you treat "rig" as a verb then you know the proper interpretation for: "six blue _igs went walking..." must be "WIGS" because a noun is required in the position in doubt. (note you wouldn't allow WIGS and PIGS to coexist in that word position for similar reason)