Duolingo-inspired pretesting with words and pictures improves vocabulary learning

TL;DR

Guessing word-image or image-word pairs with feedback improves vocabulary learning, similar to pretesting effects in memory research.

Contribution

Demonstrates that pretesting effects occur in word-image/image-word guessing exercises for language learning, even without prior knowledge.

Findings

Pretesting improved vocabulary recall and multiple-choice test performance compared to reading-only conditions.

Participants preferred pretesting over reading for learning second-language vocabulary, especially with word–image formats.

Pretesting benefits emerged even when learners had no prior knowledge or cue-target associations.

Abstract

Contemporary language learning applications such as Duolingo and Rosetta Stone often introduce vocabulary through guessing-with-feedback exercises in which learners match words and pictures. We investigated whether that process might yield a pretesting effect—that is, the phenomenon where guessing with correct answer feedback (pretesting) enhances memory. Across four experiments, adult online learners engaged in multiple-choice pretesting to learn Spanish word translations shown in word–image (Experiments 1–2) or image–word (Experiments 3–4) format. Relative to a read-only condition, pretesting yielded statistically significant performance improvements on subsequent cued recall (Cohen’s d = 0.18–0.40) and, in most cases, multiple-choice tests (d = 0.25–0.67), regardless of whether test formats were separately presented or intermixed. Participants also reported preferring pretesting over reading for learning second-language vocabulary, especially for word–image learning. Together, these findings extend the pretesting effect to visual and verbal materials, offering theoretical insights and substantiating word–image and image–word guessing-based approaches of language learning.

The online version contains supplementary material available at 10.1186/s41235-026-00708-y.

When learning a second language, many individuals now turn to modern language learning applications such as Duolingo and Rosetta Stone. These applications offer guessing exercises involving words and pictures, in which users either see a second-language word and guess the corresponding image, or see an image and guess the corresponding word. The correct answer is then shown as feedback. Are these exercises effective for learning vocabulary? Research on the pretesting effect—the memory benefit of taking practice tests on unlearned material and receiving feedback on the correct answer—suggests they might. It has been unclear, however, whether this effect extends to the word-and-picture guessing exercises commonly found in language learning apps. Moreover, the theoretical mechanisms by which pretesting might benefit learning in such circumstances remain to be clearly established. In the present study, adults with no prior knowledge of Spanish learned vocabulary words via word–image or image–word guessing exercises. With word–image learning, they either selected the image matching a Spanish word (pretesting) or studied a word–image pair (reading). With image–word trials, they either selected the correct word for an image (pretesting) or studied an image–word pair (reading). Correct answer feedback was always provided. On subsequent cued recall and multiple-choice tests, participants typically performed better on words learned through pretesting, demonstrating that pretesting effects can occur through guessing activities involving words and pictures.The present findings have important theoretical and practical implications: They extend research on the pretesting effect to visual–verbal materials, plus provide evidence that the guessing exercises used in popular language learning applications can support effective vocabulary learning. Moreover, these findings raise the possibility that pretesting effects can emerge in cases where learners lack prior knowledge and preexisting cue-target associations, which are consistent with theoretical accounts of the pretesting effect which suggest that pretesting enhances attention and encoding of correct answers, rather than solely relying upon activating prior knowledge.

The online version contains supplementary material available at 10.1186/s41235-026-00708-y.

Figures