top of page

Case Study: How Well Do You Know Your Speech Patterns?

Words by: Kat Peterson



How well do we know what affects our families most? If you had to guess, what factors do you think have the strongest effect on your family’s speech patterns: amount of sleep, stress at work, sickness at home, or downtime to relax? You might say it changes by the day or by the person. We wondered if families who participated in our study had an intuitive sense of what factors play the biggest roles in how much caregivers spoke to their children over time.


To do this, one family was asked on a scale from “Not at all” to “Very much” to indicate how much each factor (for example, positive mood) affected the average amount caregivers spoke. Their predictions are shown in Figure 1. For the most part, the family predicted that the majority of the variables would have a neutral effect or no effect on how much they spoke. However, as shown in Figure 2, almost all variables turned out to have some effect—whether positive or negative—on the amount caregivers spoke!



How to read graphs below: On the left axis of the graph (y-axis) are the variables. On the bottom axis (x-axis) is the standardized beta, which shows the strength of the effect each variable had on the rate of caregiver word tokens per minute. The gray vertical line indicates a neutral effect. Any points to the right of this gray line indicate a positive correlation effect (increase in average words per minute). Any points to the left of this gray line indicate a negative correlation effect (decrease in average words per minute).


Figure 1: What the family predicted would relate to their speech patterns.

Figure 2. What our models said related to the same family’s speech patterns.


In this blog, we will take a deep dive into a comparison between the family’s predictions and their actual results. Below I have selected 6 variables that had the highest difference between the family’s predictions and the results.



The Number of People Present.


The family predicted that the number of people present during bath time would have a moderately positive effect on how much caregivers spoke. In other words, the expectation was that the greater the number of people present, the more the caregiver would speak. However, the results showed that more people present during bath time actually had a negative effect on caregivers’ average word use.


Changes in COVID-19 Cases.


One of the variables that the participants guessed would have a neutral to no effect on their language patterns was the change in COVID-19 cases from the day before. In fact, the results show that this variable was related to the caregivers speaking the least! In other words, on days the COVID-19 cases were rising, these caregivers talked the least.




The weekend was another variable the family predicted had little to no effect, but their results show that weekend bath times tended to have a moderately negative effect on how much caregivers spoke. This means that during recordings made on the weekends, caregivers spoke less.


Recording Duration.


The family also guessed that the length of the bath time recording had a relatively neutral effect on their speech patterns, but our analysis shows recording duration to have one of the most positive effects on caregivers’ speech, meaning they spoke a lot more (per minute) the longer bath times lasted!


Negative Affect.


Although the family was correct in predicting that positive affect or emotion would have a positive effect on their language patterns, they were incorrect in guessing that negative affect or emotions would have a negative effect. Their results show that they actually talked more when they were feeling negative—so interesting!


The Percentage (%) of Time Spent on Leisure Activities.


Finally, the family predicted that the higher percentage of time they spent on leisure activities, the lower the amount of caregiver speech. Their case proved the opposite; the percentage of time spent on leisure activities turned out to be one of the variables most positively correlated with caregiver speech!


**Additionally, the variable “week after financial assistance” had the highest positivity correlation for this family, but the caregiver’s prediction for this variable was not available.


What does it all mean?


In all, this comparison between an individual family’s predictions about which variables affected their language and their results from the study demonstrate just how subtle the effects of internal and external factors can be on behaviors like how much we talk to our children. The family predicted that many of the variables would have close to no effect on their patterns, but the results suggest that most variables had at least some kind of effect!


What does it mean that our models were so different than the family’s predictions? One possibility is that parents speak to their children so naturally they aren’t that aware of days they are talking more or less! Another possibility is that parents do have a good sense of how they are talking to their children—like maybe how positive and engaged their speech is—and what affects those qualities, but that this is not captured in our rough measure that simply counts the number of words parents used.


What do you think? Do you notice that you talk more to your child some days and less other days? Feel free to send us a message or let us know in the comments!




bottom of page