Extensive empirical research across diverse natural languages, writing systems, and spoken dialects consistently finds that word frequency distributions obey Zipf's law, refuting the idea that exceptionless counterexamples exist.
The claim states that there are natural languages that do not obey Zipf's law. However, multiple large-scale cross-linguistic studies (e.g., papers 0, 3, and 10) consistently find Zipf's law to be a universal feature across dozens of languages, spoken dialog, and discourse levels. Therefore, the claim is refuted by current linguistic evidence.