This study aims to identify the most frequently erroneous linguistic and surface structure forms committed in computer-based speaking (CBS) and computer-based writing (CBW) performances by Korean university students. To examine the errors, a total number of thirty students were involved in two different tasks (CBS and CBW) based on the same question “Do you prefer living off campus or on campus? Use specific reasons and details to support your answer” on a certain given time. Their CBS and CBW samples were collected and analyzed based on the two main error analysis taxonomies: linguistic and surface structure. The linguistic taxonomy included the syntactic complexity (number of sentences, embedded sentences, tokens, and types) and word level errors(-ed/-ing, singular, plural forms, and subject-verb agreement), whereas the surface structure taxonomy focused on deletion, addition, repetition, substitution, ordering, and run-on sentences. The participants’ CBS and CBW samples have been implemented as research instruments while manipulating the descriptive statistics and t-test analysis purposes. According to the overall analysis, the students committed the following errors. First, the results of the linguistic error analysis focusing on the syntactic complexity revealed that they produced more tokens and types in CBW than in CBS and also yielded more sentences and embedded sentences in CBW than in CBS. Likewise, the analysis under the word level errors was carried out and identified that the most prominent errors in terms of the morphemes (-ing/-ed) and subject-verb agreement committed in CBS than in CBW. However, fewer errors in singular and plural forms in CBS were found than in CBW. Second, the results under the category (Surface Structure Error Taxonomy) were carefully examined and identified the following findings. More addition errors were identified in CBS than in CBW. However, the errors on deletion, repetition, and substitution in CBS and CBW showed no significant differences statistically.