Naming a picture out loud runs a short chain: see the object, recognise what it is, retrieve the word, say it. The question here is whether the meaning stage leaves a signature legible in EEG while somebody is actually speaking.

That last part is the hard bit. Overt speech introduces muscle activity and movement artefacts that most experiments avoid by having participants stay silent, which keeps the recording clean while removing the behaviour of interest. Sixteen native French speakers named line drawings aloud with high-density EEG recording throughout.

The category structure comes from embedding the picture labels with a multilingual text-embedding model and organising them into groups, rather than imposing categories by hand. That is a neat move: it lets the semantic organisation come from how the words relate in language, instead of from the experimenter's intuition about which things belong together.