Date created: | Last Updated:

Category: Project

Description: “Alexa, play “Despacito”!” While one might hear this in a night club where the DJ is called Alexa, the more likely environment to encounter this exclamation is at home: Already in 2020, more than 4 billion voice assistants (VAs) were used globally, with an estimated doubling of that number by 2024 (Business Wire & Voicebot.ai, 2020). Especially popular are those VAs which are integrated into smart speakers, like Amazon’s Echo or Google’s Nest. In 2021, around a third of German households owned a smart speaker (OMD Germany, 2021), a popularity which can be explained by their voice control, which makes usage easy and intuitive. This is particularly beneficial for children who might not have learned to write or read yet, and can now use technology like never before, including making Google searches or playing music. This interactivity and the use of natural speech not only makes usage intuitive, according to Reeves and Nass’ (1996) Media Equation Theory, these attributes also trigger social scripts in the users that make them mindlessly treat the device like a social being. They apply gender stereotypes to them (Tolmeijer et al., 2021), feel like having a parasocial relationship with the VA (Han & Yang, 2018), or are polite to it (Schneider & Hagmann, 2022). Despite displaying those reactions, humans are generally aware that devices such as VAs do not warrant social treatment (Nass & Moon, 2000). Children, on the other hand, have a higher tendency to humanize inanimate objects and treat them as social beings (Airenti, 2018), which is why they take a special place in human-technology interaction research as a vulnerable user group. While Media Equation Theory postulates that users display social reactions such as politeness, exclamations like “play “Despacito”!” visualize that it does not always seem to be the case. This particular example is not only uttered in a commanding manner but also lacks said politeness, which shows that users seemingly apply more functional commands without politeness to VAs, too. Audience Design Theory (H. H. Clark & Murphy, 1982) suggests a perceived lack of communicative skill in the device could make the users adapt to maximize the effectiveness of their requests. This could show, e.g., in that Alexa does not answer properly if the request contains unneeded expressions, such “please” or “could you”, which is why users will start speaking more functionally in the future. The way of learning about these deficiencies can easily lead to frustration and anger which people might let out on their device in the form of insults (Riquel et al., 2021). Children, too, have been observed to casually yell insults or threats at the VA (Lovato & Piper, 2015), though mainly for fun or testing how the device would react. With commercial VAs usually not replying to verbally abusive requests (Curry & Rieser, 2018), the worry arises that such rude behavior on the side of the user is reinforced as normal and finds its way into real-life relationships. On the other hand, with long-term integration into a family home, users can learn about the most effective phrasing of a request which can not only affect their politeness and tendency to give commands to the VA but eventually also their tendency to verbally abuse their device. With this expected increase in users’ tendencies to speak less politely and in a commanding manner, and the possibility that verbally abusive communication is normalized, concern has been growing as to the negative effects communication with VAs can bring about. With commercial VAs not rebuking non-polite or rude requests, the public has become worried that such communicational patterns transcend into human-human communication (e.g., Cosic, 2022). Verbal adaptation to a VA during an interaction has already been observed (Horstmann et al., 2023), what has not been investigated, though, is what effect long-term interactions, such as in a family home, have on users’ expression adoption tendency. Just like people who spend more time together are more likely to adapt (Riordan et al., 2014), the same could be expected for VAs. Overall, the goal of this research is to investigate communication behavior between families and their VAs over time. This includes aspects of communication causing concern among research and public, namely politeness, commanding, and verbal abuse. Due to VAs’ social nature, the influence of social constructs will be considered. Apart from the families’ communication with the device, focus is also laid on the extent to which interactions with the VA affect users’ communication outside of the VA interactions. With this, the aim is to learn more about what hides behind the worry of adopting negative communication patterns into human-human interactions. By looking at the family context, two user groups emerge which are essentially different, not only in the ways they use VAs (Garg & Sengupta, 2020) but also concerning their understanding and perception of the device (Airenti, 2018). Especially with children yet having to learn the rules of social communication, it is crucial to understand how their communication with the VA differs from that of their parents. Therefore, the following research question was formulated: RQ1: What effect does adults’ tendency to a) display acts of politeness toward, b) give commands to, and c) verbally abuse the VA have on their children’s respective tendency? With the goal of tackling these research gaps, a 2.5-year online survey was conducted in which 128 families filled in questionnaires. Over six measuring points, participants provided information on their own and their children’s perceived communication behavior, and to what extent they felt like communication with the VA influenced subsequent interactions. Furthermore, social variables were assessed, such as the level to which the participants perceived their device as a social entity, and how close they felt their own and their children’s relationship with the VA to be.