In the present study, formant frequencies are measured in General American English vowels taken from a read story which is well-controlled with respect to consonant context, lexical stress, and speech style, including words in a carrier phrase, continuously read speech, and spontaneous speech. The vowels carry primary or secondary lexical stress. In total, the database consists of approximately 1500 vowel tokens from four speakers. Vowels from each speaker are analyzed separately. Results indicate that consonant context has a greater effect on formant frequencies of non-reduced vowels at the midpoint than lexical stress or speech style.