Abstract This study examines the diachronic evolution of the polysemy of the Spanish verb sentir (‘to feel’) by means of a corpus-based dynamic behavioral profile (BP) analysis. Methodologically, it presents the first application of the BP approach to historical data and proposes some methodological innovations not only within the current body of research in historical semantics but also with regard to previous applications of the BP approach. First, whereas the majority of existing studies in quantitative historical semantics are largely based on observed frequencies or percentages of collocational co-occurrence, our study leverages more complex historical data that are based on the similarities of vectors. Second, this study also provides an extension of the methodological apparatus of the BP approach by complementing the traditional hierarchical agglomerative cluster analysis (HAC) with a dynamic BP approach derived from multidimensional scaling maps (MDS). Theoretically, this methodology contributes to a comprehensive perspective on the process of Constructionalization and the nature of networks, which is illustrated on the basis of the development of the Discourse Marker (DM) lo siento (‘I’m sorry’).

中文翻译：

---

建立动态行为模式：历时性研究西班牙多义感

摘要这项研究通过基于语料库的动态行为特征（BP）分析研究了西班牙语动词“感”的多义性的历时演变。从方法上讲，它提出了BP方法在历史数据中的首次应用，并且不仅在历史语义学的当前研究领域内而且还就BP方法的先前应用提出了一些方法上的创新。首先，尽管现有的关于定量历史语义学的研究大多基于观察到的并置共现频率或百分比，但我们的研究却利用了基于向量相似性的更复杂的历史数据。第二，这项研究还通过使用从多维比例尺映射图（MDS）派生的动态BP方法对传统的层次集聚聚类分析（HAC）进行补充，对BP方法的方法学手段进行了扩展。从理论上讲，这种方法有助于对建构化过程和网络性质的全面了解，并基于话语标记（DM）lo siento（“对不起”）的发展加以说明。