Within the field of Instructed Second Language Acquisition (ISLA), the focus of investigation is examining how instructional methods affect language learning, and the development of implicit and explicit knowledge in particular (Ellis et al., 2009; Loewen, 2020). Since fluency relies on implicit knowledge (DeKeyser, 2003, 2015; Doughty, 2003; N. Ellis, 2002; R. Ellis et al., 2009), ways to foster its development should be at the forefront of research carried out in the field. At the same time, the proliferation of technology in modern society has made its way to language classrooms, and while it has been argued that separate theories of technology-based language learning are not necessary (Egbert & Hanson-Smith, 2007; Hubbard, 2009) and that frameworks based on ISLA suffice (Chapelle & Sauro, 2019), more research is necessary to understand the impact that technology can have on language learning. The field of Computer-Assisted Language Learning (CALL) often lacks the rigorous outcome measures used in ISLA studies (Dixon et al., 2021) and with limited studies assessing the impact of CALL on grammar acquisition (Dixon et al., 2021; Gillespie, 2020), there is a need for more systematic research aligning CALL methodologies with established ISLA frameworks. This quasi-experimental mixed-methods study examines the effect of implicit instruction in three modes of instruction on the development of implicit and explicit knowledge of cleft sentences and sentences with partial inversion by advanced learners of English. Participants were adult learners in the first semester of a C1-level course in either a face-to-face (N=44), blended (N=56) or virtual context (N=54). The study followed a Test-Teach-Test structure: participants completed a batch of tests (a Timed Grammaticality Judgement Test, a Discourse Completion Task, an Error Correction Task) before, immediately after and four weeks after an 8-week teaching intervention. A questionnaire was also used to collect qualitative data at pretest and delayed posttest regarding preferences of mode of instruction and type of grammar instruction. Participants in the treatment groups received implicit instruction of the target forms. For each mode, there was also a control group (N=15; N=17; N=15) where the forms were not taught, enabling comparisons to be drawn between modes as well as treatment and control conditions. Data from outcome measures were analysed using descriptive and inferential statistics and qualitative findings were used to further understand the results. The findings show that implicit instruction was effective in developing implicit knowledge and that implicit knowledge strengthened over time, aligning with previous research (Goo et al., 2015; Kang et al., 2019; Spada & Tomita, 2010). Technology-mediated instruction, particularly in virtual contexts, was found to be at least as effective asthan face-to-face instruction, supporting claims about the potential of CALL environments (Dixon et al., 2021; Grgurović et al., 2013) but the blended treatment group showed inconsistent results, which could be attributable to digital competence (Cabangcala et al., 2021) and mixed perceptions of the mode of instruction (Stockwell, 2013). Furthermore, explicit knowledge was found to develop from implicit knowledge, lending support to the Reverse Interface Hypothesis (Godfroid, 2023; Kim & Godfroid, 2023). The study contributes to ISLA and CALL by demonstrating how implicit instruction can be effectively implemented in various instructional modes to foster implicit and explicit knowledge development, and it highlights the importance of considering learner perceptions and digital competence in technology-mediated instruction.