Table 1
Examples of English target words and non-words. Non-words had either one consonant or one vowel change.
| TARGET | C CHANGE | V CHANGE |
|---|---|---|
| astronaut | asgronaut | astrinaut |
| chocolate | chocorate | chocolite |
| elegant | elegart | elegunt |
| father | fatheg | fathor |
| imaginative | imapinative | imagunative |
| magic | magil | magec |
| obsolete | obdolete | obsilete |
| river | rives | rivur |
| understand | undepstand | undorstand |
| zebra | zetra | zebru |
Figure 1
Mean percentage (and individual data points) of trials in which ChatGPT choose non-words that kept the consonantal frame over non-words that kept the vowel frame as more similar to a target word in both English and Spanish. Dotted line is chance level.