HTK requires at the very least 30-40 'sentences' of 'phonetically balanced' words, in addition to your Grammar words.

Note: if your Grammar is large enough, it may be 'phonetically balanced' by itself - and not require any additional words to be added to your pronunciation dictionnary to properly compile in HTK.  However, in most cases, you will need to add additional words in addition to your Grammar words in order to compile an Acoustic Model with HTK.