Analysts are fond of discussing ambiguity in music, and often imply that it contributes to a piece's value. However, the claim of ambiguity is rarely tested, and it is not clear how the ambiguity inherent in a score might be conveyed to an audience by a performer, if indeed it is. In two famous cases, music theorists have claimed that performers project one interpretation or the other of two possible readings of the musical structure: Nattiez re. Brahms' Intermezzo Op. 119 No. 3, and Lerdahl & Jackendoff re. Mozart's Symphony in G minor K.550. The claims concern phrasing in the first case, and metrical weight in the second. The wide availability of recordings of these pieces allows us to test these claims by close examination of timing and amplitude, both factors known to be manipulated by performers to project phrasing and weight. This paper will report the findings of a small-scale study to test these particular claims. If the claim that these cases are ambiguous is true, we should see a wider distribution in variations of timing and amplitude at these points in the pieces than at other points where the phrasing or metrical weight is not claimed to be ambiguous. Furthermore, if the performers project one interpretation or the other, we should see a bimodal distribution with two peaks corresponding to the two interpretations. The research will use Sonic Annotator (effectively a batch-processing version of the widely used Sonic Visualiser) and standard computing tools, and could stand as an example of what any music analyst could do to test analytical claims through computational analysis of audio.