Abstract
Lexical stress-the emphasis placed on syllables within words-is a key feature of spoken language and an important marker for diagnosing speech and language disorders. However, current assessments of prosody, including lexical stress, often rely on reading skills, lack of natural contexts, and have limited clinical utility. To address these challenges, we developed two novel lexical stress (NLS) tasks: (1) a receptive task where participants identified pictures based on the lexical stress of the word, and (2) an expressive task where participants produced target words during a picture description task. Both tasks used familiar multisyllabic words with either trochaic (strong-weak) or iambic (weak-strong) stress patterns. We tested the tasks with 40 neurotypical adults and 15 typically developing children aged 5-11 years who speak mainstream US English. Participants' responses were judged perceptually, and expressive productions were also analyzed acoustically. Results showed that accuracy was higher on the NLS tasks compared to the PEPS-C, likely reflecting the use of familiar words embedded in meaningful contexts that reduce metalinguistic and task-related demands, thereby supporting greater ecological validity. Duration and intensity were the most reliable acoustic cues distinguishing stress patterns, while fundamental frequency (pitch) was less informative. These findings provide initial evidence that the NLS tasks are feasible, developmentally appropriate, and psychometrically promising for assessing lexical stress without literacy demands. By embedding items in meaningful contexts, the tasks may offer a more natural and clinically useful approach to evaluating prosody. Future work will expand testing with younger children and clinical populations.