The notion of iconicity plays an important role in recent cognitive-functional linguistics. While the idea is intuitively appealing and there appear to be a number of phenomena accountable in terms of iconic motivation, this presentation offers a critical assessment of the past functional accounts based on this idea. Functional accounts by Haiman (1983) and others refer to formal properties such as `full' vs. `reduced' form, `linguistic distance', `compactness', and `structural complexity'. My complaint is that by referring to these formal properties, these functional grammarians are not being sufficiently functional. For one thing, form differs from one language to another. For another, functionalists must ask what formal complexity really means. I argue that a principle even more functional than Haiman's form-based formulation of iconic motivation better accounts for the form-function correlation of language.