The article by Cutler and Norris proposes a model of speech segmentation in stress languages, suggesting that strong syllables trigger segmentation of the speech signal, while weak syllables do not. Experiments were conducted to test this hypothesis, where listeners detected words embedded in nonsense bisyllables more slowly when the bisyllable had two strong syllables compared to one strong and one weak syllable. The results support the model, indicating that segmentation at strong syllables delays word detection by requiring the assembly of speech material across a segmentation point. The authors argue that this segmentation serves to detect the most efficient locations for initiating lexical access. The findings challenge traditional speech recognition models and suggest that segmentation processes are crucial for efficient word recognition in continuous speech.The article by Cutler and Norris proposes a model of speech segmentation in stress languages, suggesting that strong syllables trigger segmentation of the speech signal, while weak syllables do not. Experiments were conducted to test this hypothesis, where listeners detected words embedded in nonsense bisyllables more slowly when the bisyllable had two strong syllables compared to one strong and one weak syllable. The results support the model, indicating that segmentation at strong syllables delays word detection by requiring the assembly of speech material across a segmentation point. The authors argue that this segmentation serves to detect the most efficient locations for initiating lexical access. The findings challenge traditional speech recognition models and suggest that segmentation processes are crucial for efficient word recognition in continuous speech.