Metraipolis › archive › dc_a449eb357fc4

Pseudowords as probes: Large Language Models show little of the sublexical sensitivity that governs human pseudoword processing — arXiv:2610.07936

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.07936
License (as archived): CC0-1.0
Archived: 2026-10-07T11:23:16.007Z
SHA-256 of the archived text: 47de933949ea203ba20cbd1476dab72ff88104e82c478172abbd8bae51864bb0
Stable URL of this page: https://metraipolis.com/documento/dc_a449eb357fc4

How to cite

Pseudowords as probes: Large Language Models show little of the sublexical sensitivity that governs human pseudoword processing — arXiv:2610.07936 — arXiv (en), https://arxiv.org/abs/2610.07936 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_a449eb357fc4 (Metraipolis archive dc_a449eb357fc4, sha256 47de933949ea203ba20cbd1476dab72ff88104e82c478172abbd8bae51864bb0)

BibTeX

@misc{dca449eb357fc4,
  title = {Pseudowords as probes: Large Language Models show little of the sublexical sensitivity that governs human pseudoword processing — arXiv:2610.07936},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.07936},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_a449eb357fc4; sha256 47de933949ea203ba20cbd1476dab72ff88104e82c478172abbd8bae51864bb0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.07936. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Systematicity, the probabilistic mapping of form to meaning, permeates language at all levels, and sublexical cues have been shown to govern human pseudoword processing. Yet whether LLMs exhibit comparable sensitivity to these cues remains unclear. We tested five LLMs on two Italian two-alternative forced-choice pseudoword experiments and compared their responses with a human behavioural baseline. LLMs aligned more reliably with humans when real-word options provided a lexical familiarity cue than in the pseudoword-only condition, where they fell substantially below fastText, a character-n-gram model. In addition, the sublexical cosine-similarity cue that reliably drove human--fastText agreement did not consistently transfer to human--LLM alignment, and reasoning-token expenditure bore no consistent relation to human processing difficulty. These findings suggest that LLMs do not necessarily share the sublexical cues that govern human pseudoword processing; we discuss tokenization and training-data coverage as candidate explanations.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2610.07936 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-07T11:15:39.440Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).