Automated Measures of Syntactic Complexity in Natural Speech Production: Older and Younger Adults as a Case Study

There is no consensus on what syntactic complexity is or how it can be quantified in spontaneous speech. In the cognitive literature, complex syntactic structures have usually been studied using detailed linguistic comparisons. However, when studying spontaneous speech, highly controlled methods are challenging to implement. In this paper, we adopt an approach that considers the cognitive cost of syntactic structures for automatically quantifying syntactic complexity in spontaneous speech. We define syntactic complexity as the frequency of structures that are known to have a processing cost. We investigate those structures in natural speech samples produced in a picture description task by younger and older healthy participants. First, we show that older participants produce significantly fewer complex structures, which are identified manually in the transcripts. Second, to determine how to quantify the syntactic differences between the groups automatically, we examined three automatically derived metrics: 1. Direct assessment of complex syntactic structures; 2. Mean dependency distance; 3. Sentence length. Automated assessment of complex syntactic structures was the most successful metric in distinguishing between older and younger participants. Since this metric can be derived automatically, it can save considerable time, cost and effort compared to manually analyzing large-scale corpora, while maintaining high face validity and parsimony, suggesting that it is useful for studying syntactic complexity in spontaneous speech.