Constructing Pseudowords with Constraints on Morphological Features - Application for Polish Pseudonouns and...

Area of Science:

• Psycholinguistics
• Computational Linguistics
• Cognitive Science

Background:

• Pseudowords are crucial for studying language processing, but require realistic linguistic properties.
• Existing methods may not generate stimuli that accurately mimic natural language phonology and orthography.

Purpose of the Study:

• To develop a Python pipeline for generating natural-sounding pseudoword stimuli.
• To ensure pseudowords possess key psycholinguistic properties and grammatical class indicators.
• To validate the generated stimuli through human participant studies.

Main Methods:

• Developed a Python-based pipeline for pseudoword generation.
• Incorporated features for natural phonological and orthographic resemblance.
• Ensured grammatical class distinctiveness using language-specific cues (e.g., Polish suffixes).
• Included psycholinguistic measures like orthographic Levenshtein distance.
• Conducted two validation studies with a total of 640 participants.

Main Results:

• Generated two sets of Polish pseudonouns and pseudoverbs.
• Pseudowords were found to be orthographically and phonologically plausible.
• Human participants could reliably classify the grammatical category of the generated stimuli.
• Validated the algorithm's effectiveness in creating usable research stimuli.

Conclusions:

• The Python pipeline successfully generates high-quality pseudowords for language processing research.
• These stimuli are meaningful for investigating grammatical and syntactic aspects without semantic content.
• The generated pseudowords offer a valuable resource for psycholinguistic and cognitive science research.