Start to do VS Start doing: A Diachronic Corpus-based Analysis

This paper aims to analyze the occurrence of the constructions “Start to + V.inf” and “Start + V.ing” as in “start to do” and “start doing” through time. To better see the change through time, the Corpus of Historical American English (COHA), which contains the language data from 1810 to 2010, is chosen. The corpus has a total number of 385 million words collected from 115,000 texts, and each decade has roughly the same balance of data from fiction, popular magazine, newspaper, and non-fiction books.