A Data-Driven Approach to Compare the Syntactic Difficulty of Programming Languages

• Published in: Journal of information systems education Vol. 34; no. 1; pp. 84 - 93
• Main Authors: Lokkila, Erno, Christopoulos, Athanasios, Laakso, Mikko-Jussi
• Format: Journal Article
• Language: English
• Published: West Lafayette Journal of Information Systems Education 2023; EDSIG
• Subjects: Adaptive learning; Artificial intelligence; Coding; Comparative Analysis; Computer programming; Computer science; Computer Science Education; Difficulty Level; Error Patterns; Higher education; Individualized Instruction; Information Science Education; Information systems; Introductory Courses; Java; Learning Analytics; Learning Management Systems; Object oriented programming; Pedagogy; Popularity; Probability; Programming Languages; Python; Questions; Students; Syntax; Teachers; Teaching; Textbooks; Transition probabilities; Trends; Undergraduate Students; Undergraduate study; United Kingdom
• ISSN: 1055-3096; 2574-3872

Educators who teach programming subjects are often wondering "which programming language should I teach first?" The debate behind this question has a long history and coming up with a definite answer to this question would be farfetched. Nonetheless, several efforts can be identified in the literature wherein pros and cons of mainstream programming languages are examined, analysed, and discussed in view of their potential to facilitate the didactics of programming concepts especially to novice programmers. In line with these efforts, we explore the latter question by comparing the syntactic difficulty of two modern, but fundamentally different, programming languages: Java and Python. To achieve this objective, we introduce a standalone and purely data-driven method which stores the code submissions and clusters the errors occurred under the aid of a custom transition probability matrix. For the evaluation of this model a total of 219,454 submissions, made by 715 first-year undergraduate students, in 259 unique programming exercises were gathered and analysed. The results indicate that Python is an easier-to-grasp programming language and is, therefore, highly recommended as the steppingnstone in introductory courses. Besides, the adoption of the described method enables educators to not only identify those students who struggle with coding (syntax-wise) but further paves the pathway for the adoption of personalised and adaptive learning practices.