In an attempt to assess speaking performance in ELT, some researchers use the number of spoken words as the index of fluency; the more words a learner utters, the more fluent she is. But this is a very misleading, even wrong, conception. If fluency is just a quantitative matter, it becomes often indistinguishable from verbosity: too much utterance of unnecessary words. If fluency is to be a positive notion, you should not define it simply as the number of uttered words to avoid the confusion between fluency and verbosity.
Common European Framework of Reference (CEFR) clearly regards fluency as a qualitative notion. In "Common Reference Levels: qualitative aspects of spoken language use" (Table 3, pp. 28-29), fluency is defined as follows from the highest level (emphasis added by me):
C2: Can express him/herself spontaneously at length with a natural colloquial flow, avoiding or backtracking around any difficulty so smoothly that the interlocutor is hardly aware of it.
C1: Can express him/herself fluently and spontaneously, almost effortlessly. Only a conceptually difficult subject can hinder a natural, smooth flow of language.
B2: Can produce stretches of language with a fairly even tempo; although he/she can be hesitant as he/she searches for patterns and expressions. There are few noticeably long pauses.
(Descriptions for B1, A2, A1 are omitted here, as they describe fluency only in negative terms).
'Fluency' as a qualitative notion defined by CEFR is described in terms of naturalness, spontaneity, flow, smoothness, effortlessness, or fairly even tempo. Spontaneous, natural and smooth flow (the highest C2 level) involves the comfort a speaker experiences in her creative choice of appropriate words. A natural, smooth flow of language which is almost effortless (the second highest C1 level) is more about the ease felt by a speaker when she says what she plans to say. An utterance with a fairly even tempo (the upper intermediate B2 level) indicates the undisturbedness in a speaker's performance, the maintained pace rather than the sheer speed of an utterance. In all cases, the notions are about quality felt by a speaker (and her interlocutor(s)), not about quantity objectively measured by an uninterested third person (or machine).
We may not have to quote from CEFR. I should have told you first that the simple numerical definition of 'fluency' betrays its meaning as defined in a dictionary (See Merriam-Webster Online Dictionary for example). Furthermore, as I said, the definition may even allow verbose language use to be regarded as 'fluent'.
Verbosity is never welcomed by anybody. Conventional wisdom in speaking (and writing) is let every word count. Economy of language is critical in the real world, for without it, you'll lose your audience. Or as Relevance Theory defines, you have to make your utterance 'relevant' for your audience. Your audience deserve to benefit from your utterance in proportion to the effort they made to process it. Expressions need to be concise. Wordy sentences with little content is never to be hailed. Assessment that may confuse fluency and verbosity must be discouraged.
Of course, the numerical measurement may be justified at a certain stage in ELT; learners may be temporarily allowed to focus on the number alone, if only to get motivated by its increase. But that measurement shouldn't last. After all, the learners cannot compete with native speakers for the sheer number of words they can utter in a fixed time. Learners should not be encouraged to be like a wordy, pointless bore; they should aim to be an intelligent speaker who uses the minimal amount of words to get the maximum effect. In the long run, quantitative measurement that disregards qualitative aspects of fluency can only be harmful.