Year: 2025
Type: conference-paper
Abstract: This full research paper focuses on understanding programming plans - knowledge of recurrent code patterns that accumulate as programmers gain experience. This knowledge enables them to read code not as a sequence of tokens, but in terms of patterns they recognize. As a method for studying the development of plan knowledge, time-constrained code recall tasks have been put forward. In such a task, students attempt to memorize and reproduce code they have been briefly shown. However, there is little research on how students' plan knowledge is reflected in their performance on time-constrained code recall. To investigate this, we assigned a code recall task to 27 student programmers: they viewed a code snippet for 15 seconds and attempted to reproduce it from memory. We analyzed which parts of the code they recalled accurately, the order in which they wrote their code, and the code they wrote that diverged from the shown snippet. We found evidence of plan recognition and the use of plan knowledge in reproducing code. Our results suggest the presence of two types of plans: grammatical and task plans. We also see some evidence of reasoning and using higher-level understanding of the code's purpose to reproduce code, indicating code recall could be a process of reconstruction rather than replication. Our findings provide insight into students' plan knowledge, and how it is used in code comprehension and generation. Furthermore, our findings demonstrate that time-constrained code recall tasks can provide insight into students' plan knowledge.
Institutions: Aalto University, New York University
Language: English
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: green
Funder: Academy of Finland
Award: G3584147223
OpenAlex ID: W7123897050
