The following script helps plot clear graphs from Google Ngram corpus.
Although N-gram Viewer can output a graph that depicts the historical changes of frequency for a particular phrase (or word), the quality of graph is quite low. We therefore took use of ‘ngramr’ package (Carmody, 2015) in R programming language to extract the N-gram data and plotted new graphs based on these data. After that, all graphs were integrated into a page to yield a high-quality chart which is Figure 4 (the R script we wrote is provided to help those who would like to replicate it). Please see our paper "Frequency distributions of punctuation marks in English: Evidence from large‐scale corpora" published in English Today(https://www.cambridge.org/core/journals/english-today/article/frequency-distributions-of-punctuation-marks-in-english/A953C2854729CFB44794CF8883200188).