English prepositions like on, in, and at have many meanings. We say "on the desk," but also "on the phone" and "on time."

Do we need to memorize each one? Some researchers say no. Each preposition has one core picture.

For example, "on" means touching something. Other meanings grow from that picture. In an experiment, students who learned with pictures did better on a translation test. But pictures are only a start.

Why does "on" appear in "on the desk," "on the phone" and "on sale"? Cognitive linguistics suggests that these meanings are not random. Each preposition has a core spatial image, and other meanings extend from it through metaphor. For "on," the core is contact or support, so a picture can be "on" a wall, and a person can be "on" the phone line.

In a 2018 study, 64 Cantonese-speaking learners practiced prepositions on a computer. Those who received feedback with image diagrams did better on a translation test than those who were only told whether they were right or wrong, although there was no difference on a fill-in-the-blank test. A 2024 replication with 81 learners found similar results, and the advantage lasted on a delayed test.

English says "on the desk," but also "on fire," "on the phone" and "on sale." Faced with this, learners often memorize each phrase separately. Cognitive linguistics offers another route. In this view, each preposition has a core spatial scene, such as contact or support for "on," and its other senses grow out of that scene through metaphor.

Even time expressions follow the pattern, with "at" for points, "on" for days and "in" for containers like months. Tyler and Evans mapped these networks systematically in 2003, and Boers had already proposed bringing the idea into classrooms.

In 2018, Wong, Zhao and MacWhinney had 64 Cantonese-speaking learners practice with a computer tutor that gave image-based, rule-based or simple right-or-wrong feedback. Every group improved. The image group beat the right-or-wrong group on translation, though not on fill-in-the-blank tests. A close replication with 81 Chinese-speaking learners in 2024 produced broadly consistent results, and the advantage held on a delayed test.

Boers's 2013 review sounded a warning. Many such studies are small, some effects are modest, and part of the benefit may come from deeper processing in general. Images make a good entry point, but learners still need to meet prepositions in many real sentences.