Abstract Phonological encoding in speech production is the set of processes that generates the sound form of an utterance. The word's sound form includes the phonemes as well as certain structural properties, such as word stress and syllable number. The current paper presents three naming experiments on the word's CV structure, which is the abstract representation of the alternation of consonants and vowels in the words. The naming paradigm used was the translation task in which subjects had to give the Dutch translation of visually presented English words. The results show that (1) a word's CV structure is stored and retrieved from the mental lexicon during phonological encoding, (2) a consonant cluster is represented differently than a single consonant in a CV structure, (3) words with different alternations of vowels and consonants have different CV structures, and (4) long vowels and short vowels are represented the same in a CV structure. Implications for models of speech production are discussed.