Negru, Mircea (2026) Cross-situational word learning in conditions of limited time. Bachelor's Thesis, Artificial Intelligence.
|
Text
Negrus2945126BachelorsThesis.pdf Download (705kB) | Preview |
Abstract
The cross-situational word learning (CSWL) process involves aggregating information across multiple ambiguous situations to acquire a new word. There are two leading theories on the specific mechanisms of CSWL. The propose-but-verify mechanism is thought of as a type of conscious learning by which only one hypothesis about the meaning of a word is stored for later testing. On the other hand, the gradual association mechanism is a type of statistical and subconscious learning through which not just one, but many possible word-to-referent pairs are learned simultaneously. We discuss how the time constraint influences the learning strategy. We investigate the interaction between the two competing strategies through an experiment in which participants try to find the meaning of novel words in ambiguous situations under conditions of limited time. A stricter deadline facilitates subconscious learning. Therefore, we expect to see more evidence for gradual association as less time is given. However, more time to make a choice would show more evidence for the propose-but-verify account. The time given to make the choice is the condition. We use 3 different conditions for comparison pseudo-randomized within participants: 2s, 3.5s and 5s. Our results show stronger learning as more time is given. Furthermore, our findings are in line with the propose-but-verify mechanism in each condition.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Supervisor name: | Jones, S.M. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 30 Jan 2026 14:25 |
| Last Modified: | 30 Jan 2026 14:25 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/37221 |
Actions (login required)
 |
View Item |