Context-independent = monophone
Context-dependent = biphone, triphone, generalized triphone, quadphones, quinphones. The context may or may not include the position of the phone within the word (word-position dependent), and word-internal and cross-word contexts may be merged or considered as separate models. (Markov Models for Pattern Recognition: From Theory to Applications, pg. 310, Ruslan Mitkov, 2009)
The term uni-gram refers to the prior probability of the words in the lexicon and the term zero-gram model is used to denote the uniform probability distribution over o lexicon of given size. (Markov Models for Pattern Recognition: From Theory to Applications, pg 114, Gernot A. Fink, 2014)