This paper presents computational methods for quantitative description and analysis of expressive performance strategies in violin performances. We present general techniques for extracting beat-level tempo and loudness data, and the Local Maximum Phrase Detection (LMPD) method. The LMPD method equates local maxima in the loudness curve with interpreted phrases, and defines measures of phrase strength (clarity), phrase volatility (standard deviation), and phrase typicality (concurrence with norm), for characterizing each phrase. The methods are developed in the context of, and applied to, eleven recorded performances of the Andante movement from Bach's Sonata No. 2 in A minor BWV 1003 for solo violin by master violinists. For each performance, we present tempo and loudness summary statistics of the entire piece, its sections, and each phrase. In our experiments, we find that loudness is a more consistent indicator of phrasing strategies, suggesting that phrase structure may impose stricter constraints on dynamic than on tempo variation. The results of the LMPD method show that Kremer's performance exhibits the highest, and Enescu's the lowest, phrase volatility; Milstein's shows the highest average phrase typicality, and Enescu's the lowest; and, Grumiaux plays with the highest, and Menuhin the lowest, average phrase strength.