An investigation of programming code textbase readability based on a cognitive readability model.

摘要

The high cost of the maintenance phase of the software development life cycle (SDLC) continues to be a major economic problem for anyone who uses software. The activity of code reading has been found to consume the greatest amount of time spent on program maintenance, accounting for most of the high cost of the maintenance phase of the SDLC. This research on programming code readability forms the necessary foundation for the development of programming code readability formulas and indices that can inform the assessment of the readability of programming code. Improving the readability of programs would potentially reduce the time spent reading and understanding code, especially during the maintenance phase of the SDLC.; Researchers in programming code comprehension (PCC) have investigated code-reading difficulties primarily from the point of view of the cognitive processes of the reader. Despite decades of PCC research, however, only limited progress has been made in identifying and implementing strategies to address the problem of code-reading---reading programming code continues to be a laborious and costly activity. This dissertation has developed and tested an alternative, text-oriented readability approach to the problem of PCC---an area that has received scant attention in prior reader-oriented PCC research.; Building upon cognitive readability theory, this dissertation proposed and developed a Programming Code Textbase Readability Model (PCTRM) and a methodology to investigate the effects of textbase features of programming code on code readability and comprehension.; The evidence of our findings both supports and refines the PCTRM. Our findings indicate that improving readability reduces the time spent reading code. We found that the programming code textbase had a significant influence on readability, largely through the effect of repeated arguments. Programming domain knowledge also significantly influenced readability, slightly more strongly than the textbase.; We also found a significant interaction effect of programming domain knowledge and readability on comprehension. This interaction supports our contention that supplementing programmer training with attention to the quality of the code may potentially reduce time and cost of code maintenance.