Summary: Researchers found that everyday speech timing — including pauses, fillers, and subtle patterns — strongly reflects executive function, a key cognitive system that supports memory and flexible thinking. Using AI to analyze natural speech, the study showed that these linguistic features can predict cognitive-test performance independent of age, sex, or education.

Because speech is effortless to collect and free from practice effects, it offers a scalable way to monitor early brain changes tied to dementia risk. The findings highlight natural speech as a promising tool for early detection and long-term tracking of cognitive decline.

Key Facts

  • Speech as a Biomarker: Everyday timing patterns in speech predict executive function across adulthood.
  • AI-Driven Insight: Machine-learning models detected hundreds of subtle linguistic cues linked to cognitive health.
  • Early Detection Potential: Natural speech may enable scalable monitoring for individuals at elevated dementia risk.

Source: Baycrest

The way we speak in everyday conversation may hold important clues about brain health, according to new research from Baycrest, the University of Toronto and York University.

The study found that subtle features of speech timing, such as pauses, fillers (‘uh,’ ‘um’) and word-finding difficulty, are strongly linked to executive function, the set of mental skills that support memory, planning and flexible thinking. 

 The study is one of the first to demonstrate a direct link between natural speech patterns and essential cognitive functions, opening new avenues of research to better understand the mind. It builds on previous research that showed that faster talking speed is linked to preserved thinking in older adults (Wei et al., 2024). 

“The message is clear: speech timing is more than just a matter of style, it’s a sensitive indicator of brain health,” says Dr. Jed Meltzer, Senior Scientist at Baycrest’s Rotman Research Institute and senior author on this study, titled “Natural Speech Analysis Can Reveal Individual Differences in Executive Function Across the Adult Lifespan.”  

Participants were asked to describe complex pictures in their own words while also completing standard tests of executive function.

Using artificial intelligence, researchers analyzed the recordings and identified hundreds of subtle features, including pauses, filler words and timing patterns. These features reliably predicted performance on cognitive tests, even after accounting for age, sex and education.  

Executive functions decline with age and are often compromised early in dementia, but they are hard to track with traditional testing, which is time-consuming and vulnerable to practice effects, the improvements in performance due to familiarity.

Natural speech, by contrast, is an everyday behaviour that can be measured repeatedly, unobtrusively and at scale. It also provides information about processing speed as a sensitive measure of cognitive integrity in an ecologically valid manner, without the need for imposed time limits — something that is challenging to capture with most traditional cognitive tasks.  

Given the ease, convenience and sensitivity of natural speech analysis, it is an ideal choice for repeated assessments, which could identify individuals who are experiencing cognitive decline at a higher rate than expected and may be at high risk for developing dementia.

“This research sets the stage for exciting opportunities to develop tools that could help track cognitive changes in clinics or even at home. Early detection is critical for any cure or intervention, as dementia involves progressive degeneration of the brain that may be slowed,” says Dr. Meltzer. 

The researchers emphasize the need for longitudinal studies, following individuals’ speech over time, to separate normal aging from early signs of disease. They note that combining naturalistic speech with other measures could make early detection of cognitive decline more precise and accessible.   

Funding: This research was supported by the Mitacs Accelerate program and the Natural Sciences and Engineering Research Council of Canada (NSERC).  

Key Questions Answered:Q: How does natural speech relate to brain health?

A: Subtle timing features in everyday speech closely track executive function, a core marker of cognitive integrity.

Q: Why is speech a useful tool for early dementia detection?

A: It can be measured frequently, naturally, and without testing pressures, revealing decline earlier than traditional tasks.

Q: What did the AI analysis uncover?

A: Pauses, fillers, and timing patterns reliably predicted performance on executive-function tests across adulthood.

About this speech and cognitive decline research news

Author: Natasha Nacevski-Laird
Source: Baycrest
Contact: Natasha Nacevski-Laird – Baycrest
Image: The image is credited to Neuroscience News

Original Research: Closed access.
“Natural Speech Analysis Can Reveal Individual Differences in Executive Function Across the Adult Lifespan” by Jed Meltzer et al. Journal of Speech, Language, and Hearing Research

Abstract

Natural Speech Analysis Can Reveal Individual Differences in Executive Function Across the Adult Lifespan

Purpose:

Automated analysis of naturalistic speech has emerged as an effective tool for detecting cognitive decline in dementia but has seldom been used to examine the ordinary cognitive decline occurring in normal aging.

Executive function (EF) declines throughout the adult lifespan but is difficult to track longitudinally due to practice effects, making speech-based assessments particularly attractive. This study examined relationships between EF and speech characteristics.

Method:

We collected two audio picture descriptions from participants in two experiments that also included EF assessments, with 67 healthy older adults (aged 65–75 years) in Study 1 and 174 healthy adults (aged 18–90 years) in Study 2. Language composite scores were computed by aggregating relevant speech features indexing aspects of speech that have been reported to show changes in pathological aging.

Principal components reflecting common covariation in speech features were extracted from a large training data set to compute speech domain scores. The relationships between language composites/speech principal components and EF were assessed while controlling for age, gender, and education.

Results:

In Study 1, older adults’ word-finding difficulties, measured as speech disfluencies, showed significant associations with EF. Study 2 confirms that speech disfluencies can explain individual differences in EF not only for adults above the age of 65 years but also across the adult lifespan. Information units and coherence in speech showed weaker associations with EF and Montreal Cognitive Assessment scores that were not significant after correction for multiple comparisons.

Conclusion:

The findings revealed associations between word-finding ability in natural speech and general EF across the adult lifespan, supporting natural speech analysis as a convenient and sensitive assessment of general cognitive ability.