There are vowel and speech synthesis modes. From the MI website…
Vowel and speech synthesis
A collection of speech synthesis algorithms.
HARMONICS: crossfades between formant filtering, SAM, and LPC vowels, then goes through several banks of LPC words.
TIMBRE: species selection, from Daleks to chipmunks. How does it work? This parameter either shifts the formants up or down independently of the pitch; or underclocks/overclocks the emulated LPC chip (with appropriate compensation to keep the pitch unchanged).
MORPH: phoneme or word segment selection. When HARMONICS is past 11 o’clock, a list of words can be scanned through by turning the MORPH knob or by sending a CV to the corresponding input. One can also patch the trigger input  to trigger the utterance of a word, use the FM attenuverter to control the intonation and the MORPH attenuverter to control speed.
AUX: unfiltered vocal cords’ signal.
It includes whole words, which you could sample or slice to get the specific sounds you want.