To like or not to like: Negotiating food assessments of children from families with a low socioeconomic position

The present study explored how primary school-aged children from families with a low socioeconomic position produce 'likes' and 'dislikes' of foods during everyday family meals, and how these (dis)likes are understood and treated by their parents. It is crucial to understand how food preferences develop in the course of everyday life, as it is known that there are socioeconomic disparities in food preference and consumption, and that children from families with a low socioeconomic position have relatively poorer diets. Deploying an interactional approach to food preference, video recordings of 79 evening meals in families with a low socioeconomic position were analyzed using discursive psychology and conversation analysis. The analysis highlighted that children's food likes and dislikes were treated differently by their parents. While likes were routinely not responded to, agreed with or further elaborated, dislikes were predominantly oriented to as food refusals or treated as inappropriate, or non-genuine claims. Children's food assessments, i.e., likes and dislikes, were often disattended by parents when they appeared to be food preference displays. By contrast, assessments that accomplished social actions like refusals and complaints were more often responded to. The analysis also revealed the importance of distinguishing between assessments about food items in general, that were not currently being eaten, and assessments of food eaten here-and-now. All in all, the study evidences that and how assessment sequences open up interactional spaces where children and parents orient to and negotiate relative rights and responsibilities to know, to assess and to accomplish specific actions. Implications for food preference research are discussed.


Background
Over the past decades, the prevalence of obesity has increased rapidly, as well as the prevalence of associated lifestyle diseases (Williams, Mesidor, Winters, Dubbert, & Wyatt, 2015). An imbalance between energy intake and expenditure, due to, e.g. an energy-dense diet and limited physical activity, is a major contributor to obesity (Mitchell, Catenacci, Wyatt, & Hill, 2011). Food preferences developed in childhood greatly influence food preferences and eating behavior in later stages of life (Anzman-Frasca & Ehrenberg, 2018;Issanchou, 2017;Skinner, Carruth, Bounds, & Ziegler, 2002). Therefore, it is important that a preference for healthy foods and healthy eating habits are developed already in childhood. Whereas in low and middle income countries obesity is predominantly a problem of the rich, in high income countries obesity is more prevalent among people with a low socioeconomic position (SEP) (Dinsa, Goryakin, Fumagalli, & Suhrcke, 2012). Studies conducted in high income countries, in which the present research is also situated, have shown that children from families with a low socioeconomic position (SEP) have poorer diets than their higher SEP counterparts (Mech, Hooley, Skouteris, & Williams, 2016; Van der Velde et al., 2019;Zarnowiecki, Dollman, & Parletta, 2014;Zarnowiecki, Parletta, & Dollman, 2014). As such, populations and especially children with a low SEP could gain most benefit from interventions aiming to improve, e.g., dietary behavior. However, populations with a low SEP are least reached by such interventions (Beauchamp, Backholder, Magliano, & Peeters, 2014;Bukman et al., 2014). A complication to this issue is that populations with a low SEP are relatively underrepresented in current research.
Various factors contribute to more unhealthy eating behavior in populations with a low SEP. For example, healthy, nutrient-rich foods are more expensive than more unhealthy, energy-dense foods (Monsivais, Mclain, & Drewnowski, 2010). Individuals with a low SEP perceive high costs of healthy foods as a barrier to healthy eating (Van der Heijden, Te Molder, Jager, & Mulder, 2021). In addition, competing values can also pose a barrier to healthy eating: people with a low SEP acknowledge the importance of healthy eating, but may simultaneously value maintaining social relationships, upholding specific identities, or keeping up traditions, which may involve indulging in unhealthy eating practices (Van der Heijden, TeMolder, Jager, & Mulder, 2021). Food 'liking' is also a major influence on food choice (Liem & Russell, 2019;Wanich et al., 2020). Food liking appears to be patterned by SEP. For example, men with a low SEP reported less consumption and lower implicit (unconscious) liking for fruit than men with a higher SEP, while no differences were found in explicit (conscious) liking, nor for other foods such as cheese and cake (Pechey, Monsivais, Ng, & Marteau, 2015). Moreover, people with a low SEP displayed a preference for an abundance of foods, whereas their higher SEP counterparts focused more on aesthetics than quantity (Baumann, Szabo, & Johnston, 2019).
Cutting edge sensory science and psychological research showed that associating unhealthy foods with tastiness contributed to more unhealthy food choices (Raghunathan, Naylor, & Hoyer, 2006;Mai & Hoffman, 2015). However, studies on how people associate healthiness and tastiness of foods with each other show mixed results. Whereas Raghunathan et al. (2006) showed that people implicitly associated unhealthy foods with tastiness even if they explicitly reported that they did not, Werle, Trendel, and Ardito (2013) showed that people associated healthy foods more with tastiness, and Van der Heijden, Te Molder, De Graaf, & Jager (2020) showed that children as well as parents with a low SEP associated healthy foods with tastiness on implicit level, whereas children indicated unhealthy foods as tastier than healthy foods on explicit level. Thus, although research has shown that there are socioeconomic disparities in food consumption and food preference, and various determinants of eating behavior such as food costs, beliefs, associations, and liking have been identified, it remains unclear how such food beliefs, associations and likings play out in everyday life. In order to gain a more comprehensive understanding of food preference, it is crucial to unravel how food likes and dislikes are produced and responded to, what terms such as 'liking', 'tasty' and 'healthy' actually mean to people in the context of everyday life and what norms and (hidden) moralities are at play in displays of real-life food preferences (Van der Heijden, TeMolder, De Graaf, & Jager, 2020).
Family mealtimes are a suitable setting for the study of food preference in everyday life, as it is a place where food is offered and eaten, and where taste is regularly negotiated (Wiggins, 2013). The family mealtime has been a site to study, for example, how parental feeding strategies (e.g., authoritative, authoritarian, indulgent) and types of encouraging prompts (e.g., reasoning, bribing, pressuring) influence children's eating behavior (Edelson, Mokdad, & Martin, 2016;Fries, Martin, & Van der Horst, 2017;Hughes et al., 2011). However, less research has been done on initiatives by children to display their food preferences. The first study on children's food likings and parents' responses to them showed that there is little interactional space for children to voice their food preferences, i.e., likes and dislikes (Wiggins, 2014). Whereas parents made more claims about what their children like or don't like, children mostly made claims about their own likes and dislikes of food. However, parents frequently countered their children's claims, or treated them as inappropriate (Wiggins, 2014). More research on how children display their food preferences and how this is treated by parents, is crucial to our understanding of how children's food preferences are developed in the context of everyday life, and what it means to children and parents to like or dislike a food.

A discursive psychological approach to food preference
Discursive psychology offers a well-suited entry point for systematically researching the interactional organization of food likes and dislikes, as it engages with how psychological constructs, such as food preferences, are produced and made relevant by people in everyday talk (Wiggins, 2017). Discursive psychologists study food evaluations as interactional practices, rather than as individual cognitive constructs (Wiggins, 2001). Already in 2001, Wiggins noted that most social psychological research on food preference was based on assumptions derived from attitudinal research. Food preference was thereby treated as a cognitive construct that is measurable with, e.g., questionnaires and rating scales (Wiggins, 2001). Two 'problems' arise when food preferences are studied solely in this way (Wiggins, 2001;Wiggins & Potter, 2003). First, the distinction between underlying food preferences (e.g., attitudes) and evaluations of particular food items (e.g., immediate hedonic taste evaluations), which participants find relevant (Wiggins & Potter, 2003), appears to be difficult to make in practice (Wiggins, 2001). Second, by drawing on evaluative terms such as which foods are "liked" or "disliked", "tasty" and "not tasty", "healthy" and "unhealthy", researchers impose specific categories and lexical terms of food evaluations upon their study participants. When used in questionnaires or rating scales, the meaning of such evaluative terms is taken out of context: it is not considered how such evaluations are oriented to and used in daily realities, at a specific point in time, or for any other purpose than expressing a cognitive state or sensory experience. There may be a discrepancy between researchers' and participants' interpretation of the meaning of those terms (Wiggins, 2001;Wiggins & Potter, 2003). In a recent review on healthy eating beliefs and the meaning of food in populations with a low SEP this issue also became apparent, as the results showed that people expressed various meanings of what is 'healthy ' and 'good' eating (Van der Heijden et al., 2021).
Discursive psychological research on food preference offers an alternative approach to attitude-based research on food preference by studying what people do when expressing food evaluations in everyday talk; which social actions are accomplished when people express evaluations of food (Wiggins & Potter, 2003;Wiggins, 2017). A discursive psychological approach to food preference research helps to bridge the gap between psychological and sociological concepts of eating (Wiggins, 2004). Everyday talk is highly structured and organized, and people orient to interactional norms to make sense of each other (Schegloff, 2007). Therefore, by studying how people orient to a construct such as food preference in the normatively organized context of everyday interaction, common-sense taken-for-granted interactional and societal norms (Garfinkel, 1967) regarding food preference could be unraveled. Such norms may not be expressed and oriented to explicitly, and may thus remain invisible in research that relies on self-reports such as interviews (Versteeg, 2018).

Study aim
A deeper understanding of the role of taste and how this is negotiated in the everyday life of families with a low SEP could provide a valuable contribution to future interventions that aim to improve eating behavior in this target group. To date, discursive psychological research on food preferences, or any other subject, has not specifically topicalized socioeconomic position. In addition, only little research focused on discursive practices deployed by children to display their food preference. The present study aims to shed light on how children from families with a low SEP show (dis)likings of food, i.e., provide evaluations of foods, during everyday family meals, and how these (dis)likings are understood and treated by their parents.

The interactional approach to assessments
In order to gain a more in-depth understanding of how children use evaluations of food in everyday talk, this section will outline the discursive psychological approach to attitudes and evaluations in everyday interactions, i.e., assessments, and highlight key concepts therein. It will review contributions both from discursive psychology (Potter, 2021) and conversation analysis (Sidnell & Stivers, 2013), two closely aligned approaches that have been used to study assessments in naturally occurring talk-in interaction. Henceforth, the term assessment will be used as a synonym to 'evaluation'.

Assessments and actions
Discursive psychological research on food preference shows that evaluations of food, i.e., food assessments, in everyday talk are actionoriented. By providing evaluations of food, individuals produce (other) social actions within ongoing (social) activities (Wiggins & Potter, 2003). For example, evaluations of food can serve to compliment the cook, request (more) food, account for (not) eating food, claim knowledge of or experience with certain foods, convince someone to eat a food, or phrase a possible complaint about a food (Wiggins, 2001;Wiggins & Potter, 2003). In addition, conversation-analytic work showed that, as part of social practices, food assessments can also fuel topical talk, reorient people's focus towards the food, or stop an emerging course of action (Mondada, 2009). Evaluations of food are not always fully phrased as such; they can also come as bodily expressions, gustatory mmms, or disgust markers such as 'yuck' (Wiggins, 2002(Wiggins, , 2013. Thus, food assessments can serve as vehicles for other actions in addition to or instead of evaluating food; the former is the so-called double-barreled nature of assessments (Schegloff, 2007). Schegloff (2007, p. 20) notes that the "the action which some talk is doing can be grounded in its position, not just its composition." Thus, how assessments accomplish specific actions derives from the assessment's place within a sequence, as well as the design of the turn through which it is implemented. To illustrate, when a person initiates talk by doing an assessment, e.g., 'this food is delicious' (a so-called first-pair part (Schegloff, 2007;Stivers & Rossano, 2010;Pomerantz, 1984)), (dis) agreement from another person becomes relevant, e.g., 'yes it is', or 'well, actually I think it is gross' (as a second-pair part). By contrast, assessments done in a responsive position, e.g., to answer a question, as a response to a previous assessment, or as an evaluation of someone else's response, have a decreased response relevance (Stivers & Rossano, 2010).
In terms of composition, the action that an assessment accomplishes can partially be grounded in its formulation as an object-side or subjectside assessment (Edwards & Potter, 2017;Potter, Hepburn, & Edwards, 2020;Wiggins & Potter, 2003). Object-side assessments, for example 'this food is tasty', display an evaluation as a 'feature of the world', where the evaluation indexes a quality of the assessable (in this case, the food) (Edwards & Potter, 2017;Wiggins & Potter, 2003). Object-side assessments appear as independent of the speaker (Edwards & Potter, 2017). As such, they are suitable to express for example compliments about food, as they display not 'just' a subjective experience (Wiggins & Potter, 2003). Alternatively, subject-side assessments display a personal stance or subjective experience towards the assessable that is restricted to the evaluation of the speaker (Edwards & Potter, 2017;Wiggins & Potter, 2003), such as 'I like this food'. Subjective evaluations do not necessarily implicate co-participants in an interaction, which limits the suggestion that co-participants should respond to or agree with the assessment (Wiggins & Potter, 2003).
Assessments can also be divided into category assessments and item assessments. In the context of food preference, a category assessment implies a broader category of foods, i.e., 'I like apples', whereas an item assessment refers to a specific food item, such as 'I like this apple' (Wiggins, 2014;Wiggins & Potter, 2003). Each type can accomplish different actions. For example, a category assessment may convey a food preference as enduring over time, rather than only at the current occasion; as such, a category assessment can be used to account for, e.g., refusing a food (Wiggins & Potter, 2003). Item assessments are bound to a specific occasion, which makes them suitable to, e.g., do an evaluation of a specific food item that can be different from a general evaluation of the corresponding food category (Wiggins, 2014;Wiggins & Potter, 2003).

Assessments and epistemic rights
People ubiquitously manage and negotiate knowledge of a certain domain (Stivers, Mondada, & Steensig, 2011;Heritage & Raymond, 2005). The dimensions and morality of knowledge are highly relevant to the study of (food) assessments in interaction, because doing an assessment implies that one has epistemic access to the assessable, and that one has a relative right to assess it (Pomerantz, 1984;Heritage & Raymond, 2005). Three dimensions of knowledge, or epistemics, can be distinguished (Stivers et al., 2011). Epistemic access refers to the extent to which someone has access to specific knowledge. Epistemic primacy refers to asymmetries in people's relative rights to knowledge and to make claims about it. Finally, epistemic responsibility entails the responsibility that people have towards certain knowledge; i.e., people don't only have a right, but also a responsibility to know certain things such as (personal) information or common sense (Stivers et al., 2011). Participants in everyday interaction treat knowledge as a moral domain, and hold each other accountable for the rights and responsibilities that come with epistemic access, primacy, and responsibility (Stivers et al., 2011).
Especially with regard to interactions between children and parents, it seems plausible that there could be an asymmetry between children's and parents' epistemic access regarding certain foods or drinks, and their relative rights and obligations to know and make claims about them. This could account for how parents orient to children's food assessments, and vice versa. To illustrate, if food preference is constructed as a cognitive state of an individual, then what people like and don't like could be treated as within their own epistemic domain to which they have epistemic primacy (Stivers et al., 2011;Wiggins, 2014). However, in her paper on food likes and dislikes by children and parents, Wiggins (2014) showed for example that parents routinely countered or ignored children's (dis)likes and made claims about what their children liked according to themthereby claiming primary rights to their children's food likings, thus, treating their children's food preference as not only within the children's own epistemic domain. In addition, a study on challenging food evaluations showed that if food preferences were challenged or questioned, it was not the taste evaluation as such that was challenged, but rather whether the speaker was entitled (had a relative right) to do the assessment (Wiggins, 2004).

Assessments and the relevance of responding
If we aim to understand how children's food assessments are understood and treated by parents, it is also essential to identify if and how assessments are responded to. Although first position assessments invite (dis)agreement, and thus make a response relevant (but not required), assessments are also frequently not responded to by co-participants in an interaction (Stivers & Rossano, 2010). Non-response to assessments also happens at the dinner table; Wiggins (2014) noted that when children expressed an evaluation of food featuring the term 'love' (e.g., 'I love apples'), parents did not always verbally respond to it, nor discuss it in, for example, the context of what the family might eat in the future.
Thus, there is more at stake when it comes to whether assessments are responded to, or not. To this end, Stivers and Rossano (2010) developed a model of response relevance and identified multiple response-mobilizing features of speakers' turn design, that, as they argue, mobilize coparticipants in an interaction to respond. In addition to the sequential position and produced social action, features of turn design that contribute to mobilizing a response to an assessment are the use of an interrogative lexico-morphosyntax (i.e., using a question word or morpheme, through which an utterance can be understood as a question, for example), the use of interrogative prosody (e.g., a rising intonation; often hearable in the final part of questions, such as 'it is, isn't it?'), engaging in speaker gaze (the speaker gazes at the recipient, to indicate that a response is due), and displayed epistemic asymmetry between the speaker and the recipient (e.g., the speaker treats the recipient as more knowledgeable of something than the speaker is) (Stivers & Rossano, 2010).
All in all, the vast body of discursive psychological and conversation analytic work on assessments reveals how particular constructions of assessments can serve to perform particular social actions, how the design of assessments can make a (particular) response from cointeractants relevant (or not), and how assessing displays speakers' epistemic access, rights and responsibilities. Thus, treating children's food assessments as social actions in interaction, rather than as expressions of individual food preference as a cognitive state, allows us to better understand how children and parents collaborate in the interactional construction and negotiation of children's food preferences. The present paper employs this approach, which we describe in the next section.

Data corpus and study procedure
For this study, we collected video recordings of 79 evening mealtimes. The recordings were made by ten families with a low SEP. Nine of these families were recruited from multiple food banks throughout the Netherlands and one additional family was recruited via snowball sampling. One of the researchers (AH) volunteered to help at the food banks to be able to personally contact the families and inform them about the research. If interested, participants could contact the researchers.
In each family there was at least one primary school-aged child and at least one parent. Table 1 provides an overview of the participating families' compositions. Low SEP was indicated by a low or medium education level of at least one parent, based on the definition of Statistics Netherlands (CBS, 2021), and low household income. Education level and demographical information (i.e., age of the family members) were assessed before conducting the recordings. Household income was not directly assessed, since this may be a sensitive topic for the families and may reduce their willingness to participate. Instead, to avoid stigmatization, low household income was implied by the eligibility of the families to visit the food bank, which requires a household income under the national minimum income limit.
Families recorded the evening meals themselves, without researchers being present. Each family was provided with two cameras to enable recording from different angles and capture verbal as well as non-verbal interaction (e.g., body positions, hand gestures and facial expressions). The cameras were placed on a tripod stand next to the dinner site, which was a table in the kitchen or living room, or a couch in front of a television. Families were encouraged to record only when all family members felt comfortable to do so. Although some reactance to the presence of a camera might be expected from the families, awareness of a camera does not automatically pose an impediment to interaction, nor does it necessarily lead to producing only specific types of talk (Speer & Hutchby, 2003).
After completing the recordings, the families received meal boxes as reimbursement for their participation. All family members were informed about the purpose of the study (i.e., to capture naturally occurring conversation during mealtimes) and were given ample opportunity to ask questions. All recorded family members provided written informed consent (written informed consent for children was provided by their parents). This study was approved by the Medical Ethical Review Board of Wageningen University & Research, the Netherlands (METC-WU, file number: NL64893.081.18).

Analytical procedure
The data were analyzed using discursive psychology (DP) (Potter, 2021;Edwards & Potter, 1992;Potter & Hepburn, 2005;Wiggins, 2017) and conversation analysis (CA) (Schegloff, 2007;Sidnell & Stivers, 2013). These are qualitative, inductive methodologies for the analysis of real-life talk-in-interaction (Sidnell & Stivers, 2013;Wiggins, 2017). DP and CA focus on how (i.e., through which practices) specific actions (e.g., complimenting, offering, or complaining) are accomplished in interaction, rather than approaching talk as a means to access or get insight into cognitive processes (Potter & Hepburn, 2005;Sidnell & Stivers, 2013). DP and CA examine how a speaker's turns at talk are understood and treated by other participants in the interaction, as is displayed in the other participants' immediate next turns or in the subsequent unfolding interaction (Potter & Hepburn, 2005;Sidnell & Stivers, 2013;Wiggins, 2017). Thus, DP and CA focus on the way in which participants themselves make sense of each other in their interactional context. DP draws on analytical principles of CA and there is major overlap between the two. However, as Wiggins (2017, p. 42) explains, DP is distinctive in the examination of "how psychological constructs are enacted and made relevant in interaction, and the implications of these for social practices". For the present paper, the psychological construct under study entailed food preference, i.e., liking and disliking foods, and more specifically we focused on food assessments.
A verbatim transcript of the audio recordings was made by a transcription service. We manually searched the verbatim transcripts for utterances about food. We transcribed these sections in detail, using the Jeffersonian notation (Jefferson, 2004) and including non-verbal expressions (e.g., gazes, hand gestures), derived from the video recordings. We searched the detailed transcripts for food assessments and narrowed our focus to food assessments done by children. We included assessments in which children used an assessment segment in their talk (Goodwin & Goodwin, 1987) in our collection of cases for the present analysis. Thus, assessments in which children used at least one assessment term or phrase with a clear positive or negative valence, such as 'I (don't) like it', 'this food is (not) tasty', 'yummy', 'gross', and similar utterances.
We identified 96 cases of food assessments by children in our data corpus that met the inclusion criteria. We closely examined each individual case and in an iterative analytic process we looked for commonalities and discrepancies between the cases, and identified interactional patterns. Table 2 provides an overview of the identified cases. In the following analysis, we will show how children's assessments were understood and treated in various designs and contexts, as derived from our data corpus of mealtime conversations in families with a low SEP. Due to the explorative nature of the present research and space limitations, we focus our detailed analysis on the most dominant (i.e., frequent) interactional patterns. Note. a In random order. b All names are pseudonyms.

Analysis
Our analysis tracks a range of assessment features (such as the valence of the assessment, and whether the referent food item is currently being eaten or not) and the sequential position of the turn implementing the assessment. We will show that and how these features are relevant for the actions children are performing and for how parents respond to them.

Assessments conveying positive here-and-now experiences
Children's positive assessments that conveyed here-and-now experiences were frequently not responded to, agreed with by parents, and elaborated on with further specifications of the assessment object. We highlight how sequential position in conjunction with turn design furnished children with resources to implement particular social actions within larger courses of action. Whether assessments were responded to by parents or not ultimately depended on the constellation of these features.
We start by showing two excerpts in which a child does a positive assessment about the food eaten at the present moment, that is not responsive to others, and does not receive a response from its parent(s). We found this to be a common pattern throughout the data, identified in 40 cases. Excerpt 1 shows Mum, Dad, Marie (11 years old), Claire (3 years old) and Benjamin (8 years old) sitting on the couch in the living room, eating their dinner. Claire finished her potatoes and Mum offers her some more, asking her which potatoes she wants (as there are two options to choose from). Our target assessment is on line 11. After Mum and Claire established which potatoes Claire wants and Mum provided her with those, a 3-s silence follows and Marie initiates a new sequence with a positive, object-side assessment in line 11. Marie's assessment is in sequentially initial position, as it is not responsive to, nor receipting a prior turn, meaning that it could but in this case does not receive a reply (Stivers & Rossano, 2010). By virtue of being sequentially first, a response could have been relevant here; specifically, it would make a second assessment (agreeing or disagreeing) a logical second-pair part of the assessment (Schegloff, 2007). In addition, the object-side assessment projects the 'tastiness' as a quality of the potatoes itself, as a 'feature of the world', rather than as a personal stance of Marie. As such, the assessment may not be restricted to the epistemic domain of Marie, but may also cover the epistemic domains of others (especially Mum, who prepared the potatoes), which may invite them to respond. This assessment could also function as a vehicle for another social action; object-side assessments are suitable ways to make, e.g., compliments about the food (Wiggins & Potter, 2003), and Mum could easily have treated this assessment as a compliment. Notably, while doing her assessment, Marie is looking down at her plate rather than at the other family members. She does not make anyone accountable for receipting her turn and/or producing a response. In addition, Marie engages in a speaker gaze towards Mum after she does her assessment (line 12), but at that point Mum is looking at her own plate while scooping potatoes onto it, and does not see nor receipt Marie's gaze. Thus, in light of the interactional efforts that accompany Marie's assessment (initiating the sequence, object-side assessment, potential social action, and gazing at Mum after the assessment), a response could be relevant here. However, to be treated as, e.g., a compliment rather than as a subjective experience that does not implicate anyone, it appears that these features alone do not suffice and additional elements are necessarywhich are missing in this interaction. In Excerpt 2, Mum, Mason (18 years old), Sarah (10 years old) and Lucas (11 years old) are sitting at the dinner table. They are just starting their meal with rice, chicken and sauce. Our target assessment is on line 15.
In line 4, Lucas inquires about the food he is being served. Mum glances at Lucas, indicating her receipt of his inquiry, but does not produce a verbal responseprobably because after a short 0.3-s silence Sarah intervenes and gets involved in an argument with Mum, which can be derived from Mum's angry voice in response to Sarah's turn (lines 6-9). Following a 2.5 s silence, Lucas produces a high-pitched but soft 'oh' while looking back and forth from his plate to Mum (lines 11-12). 'Oh' is considered as a change-of-state token indicating a change in a person's current knowledge state (Heritage, 1984(Heritage, , 2018. As such, Lucas conveys he has found out what the food is. After another silence of 2 s Lucas does a positive assessment (line 15). His assessment seems to be a development of his own previous turns, 'oh' (line 12) and inquiry about the food (line 4). This does not yield a response: a 17-s silence follows, in which all family members continue eating. After this, Lucas starts a new sequence (line 17).
Lucas's inquiry about the food on his plate and his repeated looking at his plate exemplify that his assessment is about this food item, here and now. Lucas's assessment is not clearly formulated as a subject-or object-side assessment. Lucas' inquiry about the food in line 4 might be a potential criticism of the food, and after finding out what the food is (line 12), a way to address this veiled criticism could be to offer a compliment. This way, the positive assessment may also function as a compliment to the preparer of the food (Mum). However, from the way his assessment is treated by others, it does not seem to be a vehicle for any other action than a display of his personal experience and preference for this food item. Informing about personal preference is inherent to Lucas's own epistemic domain (i.e., territory of knowledge), and other family members are not implicated by that action. Being treated as such, the assessment does not make a response from others relevant. With regard to features of turn design, Lucas's assessment is declarative rather than interrogative in wording and falling intonation. This does not convey that Lucas might expect a response. Lucas does, however, engage in speaker gaze, as he is looking back and forth to his plate and Mum while doing his assessment. There might also be epistemic asymmetry between Lucas and Mum, as Mum is likely to be responsible for, and more knowledgeable of, the food on their plates. These latter two features could have made a response relevant. However, Mum may not have noticed Lucas's gaze, as she is looking at her own plate.
Thus, Lucas's positive assessment does not yield any response. A close examination of the assessment's social action (a potential compliment, but treated as a display of personal preference) and features of Lucas's turn design, indicate that this assessment did not make a response relevant. The lack of response to his assessment is not treated as problematic by Lucasconsistent with findings of Stivers and Rossano (2010), who established that speakers orient to nonresponse to an assessment as an acceptable alternative.
We now move on to Excerpt 3 in which an assessment in subsequent sequential position also fails to elicit a reaction from co-present interlocutors. We have Mum, Dad, Marie (11 years old), Claire (3 years old) and Benjamin (8 years old) again sitting on the couch in the living room, and they just started their dinner. On the table by the couch are two types of potatoes: plain cooked potatoes and baked potatoes with mushrooms. Mum is going to distribute the potatoes onto the children's plates. Our target assessment is produced by Marie in line 13.
In line 3, Mum starts a sequence by inquiring 'which potatoes do you want', although the inquiry is not clearly directed to anyone in particular, as Mum is looking at the table and the pans. In overlap with Mum's turn-final 'those', Marie indicates her choice for the baked potatoes with mushrooms, but Mum counters it by claiming better knowledge of Marie's food preference, stating that Marie does not like those 'at all' (line 6) because they contain mushrooms. While Mum is subsequently occupied with Claire's food choice in lines 7-11, Marie counters Mum's claim by stating that she will avoid taking any mushrooms (line 9), and continues to scoop the baked potatoes onto her plate. This counter is not responded to. After a 4-s silence, Marie follows up with a positive assessment of the potatoes she supposedly didn't like (line 13). Note the contrastive emphasis on 'do' and 'like' which invites listeners to hear her turn as a further counter to Mum's assertion regarding her food preferences. This assessment is not responded to, as becomes apparent in the next few lines (14-19), where a 2-s silence follows, and Mum occupies herself with serving Claire and Benjamin their potatoes.
Marie does her positive assessment in a responsive position, as a reaction to Mum's assertion, and her assessment is also a vehicle for another action: negating Mum's contestation by doing a counter-claim (the positive assessment). As Marie's assessment is done in subsequent (not initial) sequence position, it does not necessarily make a response relevant. Moreover, Marie's assessment is formulated as a subject-side assessment, displaying her personal positive stance towards the potatoes she is currently eating and thus limiting the relevance for others to respond (Wiggins & Potter, 2003). The assessment is not responded to, and this is not treated as problematic by Marie.
Next, we consider two examples in which a child's positive assessment does receive a response, as it is agreed with by parents, and elaborated on with further specifications of the assessment object. Agreement and negotiating further specifications comprised the most typical responses when parents produced a response to their children's assessments conveying positive here-and-now experiences. In Excerpt 4, Mum, Liam (14 years old), Emma (6 years old) and James (5 years old) just started their meal while sitting on the couch by the television. They are eating pizza with mozzarella.
Mum initiates the sequence by noting that such a large piece of pizza won't fit into Emma's mouth (lines 1-2). In her response, Emma appoints Mum as the next speaker by summoning her to 'pay attention' (line 3) and subsequently chews on her large piece of pizza. In her response, Mum inserts an assessment sequence: she expresses a gustatory 'mmm' (line 6) (Wiggins, 2002). Mum's expression of gustatory pleasure does not seem to display her own experience, thoughrather, it seems enacted pleasure to display epistemic primacy over the here-and-now experience of Emma, the one who is actually tasting the pizza in this moment (cf. Wiggins, 2014;Wiggins, 2019). Emma treats this as an invitation to provide her here-and-now experience, and responds in second position with a positive assessment ('YUMM:Y' in line 7). Building on that, Mum expands the assessment sequence with a full-lexical upgraded positive object-side assessment about the cheese on the pizza, including an interrogative syntax (line 8). With Emma gently nodding, thereby non-verbally confirming Mum's interrogative, the sequence is closed (line 9).
It is notable that in her elaboration of the assessment sequence, Mum specifies the assessable (the 'soft che:ese' on the pizza) via an object-side evaluation ('v:ery tasty'), thereby presenting it as a 'feature of the world' that is not limited to the epistemic domain of either Mum or Emma. This construction, together with the interrogative syntax and turn-final tag, allow for Mum's assessment to be heard as made on behalf of Emma, whom she invites to provide a confirmation (cf. Heritage & Raymond, 2005) that the soft cheese is indeed the referent and the origin of the 'yumminess'. By first enacting Emma's taste experience occasioned by the presence of pizza in Emma's mouth, via 'Mmm', Mum treats Emma's experience as an individual sensory one while the experience is coordinated by both Emma and Mum (cf. Mondada, 2018;Wiggins & Keevallik, 2020). Subsequently, Mum specifies the experience further drawing Emma's attention to a particular characteristic of the food item she is presumably enjoying and thus effectively socializing Emma's food preferences. Thus, this extract may provide a snapshot into the interactional development of what could end up to be treated as a child's food preference and which has the origin in the interactional co-production of assessment sequences.
As a final example of positive assessments that convey here-and-now experiences that are specified further and agreed upon, Excerpt 5 features Mum and Lola (9 years old) watching TV on the couch in the living room and eating potatoes, a salad and shrimps. The target assessment is done by Lola on line 6.
After a long silence in which Lola and Mum are eating and watching TV, Mum initiates a sequence by expressing 'phew' while looking at Lola and waving her hand in front of her mouth. This could be a display of discomfort invoked by her current eating experience, but the utterance is quite opaqueas reflected in the next turn by Lola, who displays that she did not hear Mum correctly or did not understand what Mum meant, with an open class repair initiator ('Hm?' in line 4) (Drew, 1997). Mum does not produce a verbal response as she is chewing on her food, and after a 3-s silence Lola does a declaratively designed positive object-side assessment about shrimps (line 6). Lola's assessment is in responsive position: it occurs as a second-pair part to Mum's apparent display of discomfort, and it displays disagreement with that. After a 1-s silence, Mum responds in line 8 with a "pro forma" agreement to Lola's assessment (Schegloff, 2007, p. 69). She starts with 'yes but', initially agreeing with Lola's assessment, while the initial agreement actually serves to delay the subsequent dispreferred (i.e., non-aligning) response 'also quite ↑spicy'. After another 1-s silence Mum continues to specify the assessment object, by pointing out a specific shrimp ('that ↓one.' in line 8). Lola initiates repair to Mum's specification in line 11, by providing a candidate understanding (i.e., a suggested meaning) of what the assessment object is (specification plus tag: 'that big one right?'). Mum's confirmation in line 12 provides a repair solution ('y:es,'). Thus, Mum and Lola eventually reach agreement on what is actually assessed, and on the specific content of the assessmentafter which the sequence is closed.
Although assessments in responsive position, such as Lola's assessment in line 6, decrease the relevance for co-interactants to respond (Stivers & Rossano, 2010), in this sequence there is some misunderstanding regarding the assessment object (what is actually assessed) and valence (is the assessment positive or negative), and thus whether Lola's assessment about shrimps is then an agreement or disagreement to Mum's initial utterance. Such misunderstanding makes a post-expansion relevant, which is a stretch of talk that comes after the first and second pair-part (Mum's utterance and Lola's response), while still being part of the same sequence (Schegloff, 2007, p. 148). In addition, the object-side formulation of Lola's assessment displays tastiness as a quality inherent to shrimps, rather than as a personal preference and/or subjective experience of Lola, increasing the relevance of co-interactant Mum to respond (particularly to agree or disagree). It is notable that although Lola's assessment is lexically formulated as a category assessment (Wiggins, 2014;Wiggins & Potter, 2003), i.e., about shrimps as a category of food rather than specifically these shrimps, in her pro forma agreement in lines 8-10 and the subsequent specifications, Mum treats the assessment as if it were an item assessment conveying a here-and-now experience, i.e., about these shrimps. This excerpt provides another example of how an assessment sequence is interactionally achieved as a coproduction between interlocutors.
In sum, children's positive assessments that conveyed here-and-now experiences typically yielded two types of responses. A common pattern was that they were not responded to at all (as in Excerpt 1 and 2). These assessments were produced with few or no response-mobilizing features (Stivers & Rossano, 2010) and even though they occupied a turn-initial position and some assessments could have been taken as subtle compliments, parents did not attend to them. By contrast, positive assessments that were featured in sequences initiated by a parent were agreed with or responded to by further specifying the assessment referent, displaying assessment sequences as interactionally achieved co-productions (as in Excerpt 4 and 5). This can be called notable; even when displays of personal food preference do not make a response relevant with regard to features of turn design, parents could have addressed these in terms of taste development and thus could have built their children's food preferences; or could have addressed these in terms of what could be eaten in future meals, for example.

Assessments conveying negative here-and-now experiences
Children's negative assessments conveying here-and-now experiences were more likely to yield a response and more elaboration from their parents than their positive counterparts, as they accomplished different actions. They were typically treated as food refusals, or as inappropriate claims. Excerpt 6 shows an assessment conveying a negative here-and-now experience that is treated as a food refusal. Mum, Leila (6 years old) and Caitlyn (8 years old) are eating at the dinner table. Leila has finished most of her potatoes and carrots, but still has a lot of peas left on her plate. Leila initiates a sequence by summoning Mum in line 1, drawing her attention and appointing her as the next speaker. After Mum gives a 'go ahead' (line 2) (Schegloff, 2007), Leila produces a negative assessment conveying a here-and-now experience about the peas on her plate ('I don't like it', lines 3-4). The declaratively formulated subject-side assessment displays a personal non-preference for the peas, which does not necessarily implicate others, and limits the relevance of a response. However, Mum responds with a 'Hmm', followed by a short pause and a counter claim, 'but then you will still taste' (line 5).
Notably, first, by initiating the sequence and drawing Mum's attention Leila makes a response from Mum to what she is going to say next relevant. Second, in Mum's response it is not the negative assessment as such, i.e., Leila's disliking of peas, that is elaborated upon; rather, what is responded to is the social action that is accomplished, namely a refusal to eat more peas. Mum responds first with a delay token ('Hmm', line 5) which not only forestalls the upcoming dispreferred response, but also conveys she is first searching for a mutually acceptable solution (Jefferson, 1980). The response she finally produces is phrased as a compromise via the use of the contrastive 'but' and the lexical choice 'taste' that directs Leila to eat some of the peas on her plate. In her response in line 7, Leila shows compliance with Mum's directive, thereby orienting to Mum's response as a correct understanding of her negative assessment (i.e., a food refusal).
Also notable here is that Mum continues to elaborate on the amount of peas that Leila has to eat (line 9). She does so in an interrogative format, asking Leila how many peas she has to eat; she offers Leila the opportunity to come up herself with the proper way on how to deal with the peas that are left. Such practice corresponds to what Hepburn (2020) describes as parents' preference for self-direction as a resource for socialization practices. Leila is given the opportunity to develop her own solution to her 'problem behavior', i.e., refusing her peas, rather than being more invasively directed, as correcting others is generally understood as a dispreferred action (cf. Hepburn, 2020).
Excerpt 7 shows a similar scenario. Mum and Lola (9 years old) are eating potatoes, shrimps and salad on the couch by the television.
Lola initiates a sequence by expressing a declaratively formulated, softened subject-side assessment conveying her negative here-and-now experience of the salad (line 4). Mum treats it as an 'informing', as displayed by her response with a change-of-state token 'Oh' (Heritage, 1984), indicating that her knowledge state regarding Lola's disliking of the salad changed from not-knowing to now-knowing (line 5). After a 2-s silence, Mum elaborates the sequence by asking 'now what?', which effectively asks Lola to clarify the implication or upshot of her assessment. Lola treats Mum's turn as displaying some difficulty in understanding due to lack of access to the referent of the assessment. She thus walks over to Mum and points to the content of the plate to facilitate her access to the assessable. It is Mum in line 7 who names pearl onions as the referent of the Lola's negative assessment and then allows Lola to leave them out, thus treating Lola's assessment as a food refusal. Noteworthy here is the work Mum does to separate the pearl onions from the rest of the salad and thus on the one hand concede to her daughter's food refusal, while on the other hand making provisions for Lola to continue eating the salad once the problematic ingredient has been dealt with.
We continue with an example of an assessment conveying a negative here-and-now experience, that is responded to by treating it as a nongenuine claim. Excerpt 8 shows Mum, Eva (3 years old), Michael (12 years old) and Kathryn (7 years old) eating potatoes, spinach and chicken at the dinner table.
While Kathryn and Michael are engaged in some talk about pumpkin heads, Eva initiates a sequence with the disgust marker 'yuck' while grimacing and looking at her plate (lines 3-4). After 2 s, she follows it up with a full-lexical subject-side assessment conveying a subjective negative here-and-now experience. As a declaratively designed, subjective experience that does not necessarily implicate others, the relevance of a response is limited. As Michael and Kathryn are engaged in conversation they do not respond, but Mum looks at Eva. After 1 s Eva repeats her statement, possibly in pursuit of a response. Michael starts a response (line 10), but is interrupted by Mum calling Eva a 'faker' (line 11), thereby treating Eva as a non-credible person, deeming her negative assessments as not genuine. In line 12 Michael continues his response by stating 'you do have to eat it', treating Eva's negative assessment as a food refusal. Interestingly, however, is that in this case it is the brother who orients to the social action of the assessment, while the parent 'dismisses' the assessment.
Finally, Excerpt 9 shows an example where a negative assessment is not responded to. Mum, Dad, Ronald (17 years old), Oliver (15 years old) and Charlotte (13 years old) are eating stew with smoked sausage at the dinner table. There is still some smoked sausage left, which Dad offers to the family. The target assessment is done by Charlotte in line 4.
Here, the sequence is initiated by Dad who offers Charlotte smoked sausage. The offer is implemented via a negative interrogative ('don't you want') that puts some pressure on the recipient to accept the offer or at least provide strong reasons for the refusal (Drew, 2013). After a 3-s silence, Dad treats the lack of response as a refusal, and solicits an account for it from Charlotte (line 3). In response to Dad, Charlotte produces a double object-side negative assessment of the sausage: it is gross and also not tasty. The assessment accounts for rejecting a food offer by Dad. As the object-side assessment is not limited to Charlotte's own epistemic domain, a response from others could be relevant. However, the assessment is done in second sequential position, as a second-pair part to Dad's account solicitation, which limits the relevance of a response. As becomes clear from lines 5-8, her assessment is not responded to.
In sum, children's assessments conveying a negative here-and-now experience were oriented to by parents as vehicles for other actions. Parents frequently elaborated on the social action that the assessment accomplished, as they understood it, while elaboration on the evaluative component of the assessment was very limited. Assessments done in first sequential position initiated courses of action and were more likely to be responded to than assessments done in other sequential positions as they accomplished different actions; e.g., initiating the refusal of a food (first position) versus accounting for not eating a food (in a responsive position).

Assessments about food items in general
Although most assessments were done about foods that were currently eaten, i.e., here-and-now experiences, on few occasions children assessed foods in general, i.e., that were not bound to a current eating experience. These assessments were usually done in the context of a 'larger project' in which families were establishing what foods could or should (not) be eaten on a future occasion. Assessments about food items in general were designed and/or treated differently than assessments conveying here-and-now experiences as they accomplished different actions, and offered children and parents a platform to negotiate epistemic access, rights and responsibilities to assess. In Excerpt 10, Mum, Mason (18 years old), Lucas (11 years old) and Sarah (10 years old) are eating potatoes, sausages and kale at the dinner table.
The sequence is initiated by Mason, who indicates he will cook spinach stew next time (line 1). Both Sarah and Lucas have alternative suggestions: broccoli stew and broccoli soup. Additionally, in line 7, Sarah suggests including chicory, which Mason flatly rejects in line 8. In response, in line 10, Sarah does a declaratively formulated, object-side positive assessment about chicory. The declarative design and responsive position decrease the relevance of a response; however, the objectside formulation indexes the tastiness as a quality of the chicory, which is not restricted to Sarah's personal preference or epistemic domain, increasing the relevance of a response. As a social action, the assessment may serve as a vehicle to account for her suggestion in line 7 as well as a counter to Mason's rejection of it. While Mason remains silent, after a 1-s gap, Mum elaborates by asking Sarah whether the chicory was cooked in the oven (line 12). Sarah confirms this at which point Lucas jumps in and contradicts her (line 14). Mum continues to elaborate in line 15 by inquiring with whom she learnt to eat that. Sarah replies by shrugging her shoulders (line 16); possibly indicating that she cannot give Mum an adequate answer to her question. Lucas takes over and explains that he ate chicory cold (raw).
Mum's elaboration on Sarah's evaluation of chicory in the oven comes at a point in the conversation where there is a conflict in the offing between Sarah and Lucas (line 15). With her inquiry, Mum steers the conversation in a different direction and further development of the conflict is avoided. In addition, by asking Sarah about her eating habits, Mum orients to those as within the epistemic domain and authority of Sarah, while also displaying her own entitlement, perhaps even her responsibility, to know about them. Interestingly, the matter inquired here is specifically with whom Sarah learnt to eat chicory. This displays Mum's orientation to the accountability of Sarah's assessment as depending on who taught Sarah to eat chicory (as it was apparently not Mum herself), rather than, for example, where or how she got acquainted with chicory.
Finally, Excerpt 11 shows the same family on a different occasion, about to start their meal consisting of spinach, meat balls, boiled eggs and potatoes. Mum forgot to fry the bacon that she intended for the current meal. She just mentioned that the bacon is in the kitchen, but that she is not going to prepare it anymore for the current meal. Our target assessments are done by Lucas in line 3, and by Sarah in lines 5-6.
In line 1, Mum states that they will eat the bacon tomorrow with a sandwich and egg. After a 1-s delay, indicating an upcoming nonpreferred response (Schegloff, 2007), Lucas responds with 'alright (1.0) b:ut', and a positive object-side category assessment about bacon, 'bacon is a plus,' (line 3). Lucas's response is a "pro forma" agreement (Schegloff, 2007, p. 69). He initially agrees with Mum's proposal, delaying his subsequent 'dispreferred' response that does not align with that. As a social action, Lucas's turn including his assessment about bacon serves as a vehicle to display disagreement with Mum's statement that they will eat the bacon tomorrow (instead of today). Despite the object-side formulation of the assessment, the relevance of a response is limited by the responsive position of the assessment, its declarative formulation and the social action, that has little implications for others. Mum does not respond and the sequence is closed.
After a 3-s silence Sarah initiates a new sequence in line 5 by summoning Mum and inquiring her about her memory of Sarah's preference for pancakes with bacon. Sarah formulates her preference as a subject-side assessment ('I like'), and the assessment is about a food item in general (pancakes with bacon, which are not currently being eaten). Mum confirms Sarah's inquiry in line 6, and Sarah follows up with a request to make those pancakes for her some time (line 8). As becomes apparent from the next lines, Mum does not respond to this request as Lucas intervenes in the conversation, and eventually Mum starts a new sequence directed at Mason. However, it is not whether the request is granted or not that is of interest here, but rather how Sarah designs it.
The sequence-initial position, the interrogative format of Sarah's turn and the appointment of Mum as next speaker make Mum increasingly accountable for producing a response (Stivers & Rossano, 2010).
By inquiring about Mum's memory, Sarah solicits recognition, thereby displaying herself what she expects Mum to remember (cf. Shaw & Kitzinger, 2007). Sarah's inquiry and Mum's confirmation constitute a pre-sequence (Schegloff, 2007), as this sequence precedes a projected next action: a request by Sarah to have pancakes with bacon some time. Thus, the action of Sarah's assessment is double-barreled: not only does it solicit recognition in a summons-answer pair, it also functions as a pre-sequence to 'clear the grounds' for a later food request.
By showing what she expects Mum to remember, Sarah displays her understanding that Mum not only has epistemic access to Sarah's food preferences, but also has an epistemic responsibility towards them, i.e., she should know about them. Whereas in Excerpt 10 Mum actively asked Sarah about the origin of her preference for chicory, in Excerpt 11 Mum displays her understanding of Sarah's preference for pancakes with bacon in her confirmation in line 7, which diminishes the relevance for Mum to inquire further about Sarah's preference for those pancakes. Moreover, Sarah's formulation of soliciting recognition and her subsequent food request show how she orients to interactional rights and responsibilities of the family roles that are in place (i.e., child and mother). To elaborate, first, by displaying the apparent need to 'clear the grounds' before doing a food request, Sarah orients to Mum as having more authority in deciding which foods are consumed, as Mum is probably the person buying groceries and preparing meals while Sarah 'just' receives the food. By building up her food request with this presequence, Sarah displays her understanding that it would be 'inappropriate' for her to request Mum for a specific food out of the blue. Sarah's orientation to her role of a 'good child' is also illustrated in the formulation of the request itself: 'do you = want' indicates a freedom of choice for Mum to accept or decline the request, displaying Sarah's understanding of having relatively inferior rights to make decisions about food, and Mum having superior rights. Moreover, 'some time' softens the request, making it less demanding by leaving the time frame for granting the request undefined.
In sum, children's assessments about food items in general that were not bound to a current eating experience accomplished different actions than assessments conveying here-and-now experiences. When a response was conditionally relevant, assessments about food items in general were designed and/or yielded responses in which children's and parents' relative rights and responsibilities to knowledge were oriented to and negotiated, as well as relative rights to do specific actions. Thus, children's assessments about food items in general opened up an interactional space to negotiate matters of morality at the dinner table.

Discussion
The present research examined how children from families with a low SEP produce (dis)likes of food, i.e. food assessments, during everyday family meals and how these were treated by their parents. Our analysis highlights that children's food assessments conveying positive here-andnow experiences were routinely not responded to by parents, and to a lesser extent agreed with or further elaborated. Children's food assessments conveying negative here-and-now experiences were typically oriented to by parents as vehicles for other actions, particularly food refusals, or were treated as non-genuine claims. Finally, children's assessments about food items in general, i.e., not invoked by a current eating experience, were designed and treated differently as they opened up an interactional space to negotiate matters of epistemics and morality relative rights and responsibilities to know, to assess and to accomplish specific actions. All in all, children's food assessments were increasingly likely to be responded to and elaborated upon when they accomplished social actions other than preference displays and thus implicated other co-present individuals (particularly parents).
The findings of the present study add to discursive psychological literature on food assessments in family mealtimes, as the analysis of the present study shows similarities as well as discrepancies to the first study deploying an interactional approach to children's food likes and dislikes (Wiggins, 2014). In Wiggins (2014), it was concluded that parents frequently claimed epistemic primacy over their children's food preferences, and countered (dis)likes displayed by children, or treated them as inappropriate. Our findings regarding children's assessments conveying negative here-and-now assessments (i.e., dislikes) show a similar pattern, since they were regularly treated as inappropriate, non-genuine claimswhich could be an indirect way of parents to claim epistemic primacy over their children's food preferences. However, regarding children's assessments conveying positive here-and-now experiences our findings deviate from Wiggins (2014). When children in the present research expressed that they liked a food, epistemic matters were not brought to the surface of the conversation, and were not treated as non-genuine claims. Instead, these assessments mostly did not yield any response or, when responded to, parents and children collaborated to establish the specific referents of the assessments. A possible explanation might be that liking a food is not understood by parents as a vehicle for an 'urgent' social action, and does not make a response relevant (for a more elaborate explanation on 'urgency', see the section on response-mobilizing features below).
Furthermore, Wiggins (2014) noted that in some cases, children's assessments featuring the term 'love' (e.g., 'I love apples') were not responded to by parents, nor discussed with regard to possible future meals. The present analysis indicated that non-response is very common when children do assessments conveying positive here-and-now experiences (likes), but not when they express negative experiences (dislikes). Thus, adopting the terminology applied by Wiggins (2014) and elaborating on the previous point, this provides another indication that parents treat children's likes as 'nouns' (physiological and/or psychological states) that limit the relevance of a response, and dislikes and assessments about food items in general as 'verbs' (actions) that do make a response relevant.
Moreover, our findings partly correspond to Wiggins' (2014) noticing that subjective category assessments (e.g., 'I love apples') were regularly treated as expressions indicative of underlying food preferencesbut that when children did those assessments, parents seemed to treat those assessments as actions that children were performing, rather than as an underlying food preference. This observation becomes particularly apparent in our findings regarding children's dislikes, which parents regularly treated as a food refusalthus, parents' responses receipted the social action of the assessment. However, our findings regarding children's likes do not correspond to this noticing, as these were predominantly treated as displays of food preference which do not require a response.
The findings of the present study also add to conversation analytic literature on how responses are mobilized (e.g., Stivers & Rossano, 2010;Schegloff, 2010;Couper-Kuhlen, 2010;Eilittä, Haddington, & Vatanen, 2021). A clue to why some of children's assessments were responded to and others were not, may be found in children's relative rights to engage in interactions with adults, and the 'urgency' of the social action accomplished with the assessment (e.g., Sacks, 1995;Butler & Wilkonson, 2013;Eilittä et al., 2021). Sacks (1995) argued that children have relatively inferior rights to engage in conversation with adults. However, Eilittä et al. (2021) studied children's rights to engage in multiparty interactions with adults in cars and reported that if and how children's summons are receipted is highly dependent of the position and composition of the child's summons, rather than a priori determined rights to engage in conversation. In particular, children were likely to receive a response when no other conversation was going on, but not when the adults were already engaged in another conversationexcept when the reason for the child's summons was urgent (Eilittä et al., 2021). Translating this to children's food assessments in mealtime conversations, children may be more likely to receive a response when their assessment is understood as a vehicle for an 'urgent' social action. Our research showed that parents treated a food refusal as more 'urgent' than a subjective pleasurable experience.
The data corpus for the present study consisted solely of families with a low SEP. At least two ideas can be highlighted regarding this matter. Both will be addressed briefly. First, interactional research such as discursive psychology and conversation analysis asserts that interactional patterns transcend linguistic and cultural diversity (Schegloff, 2007). DP and CA research are grounded in the ethnomethodological assumption that social issues, such as power, oppression, racism, and indeed, social class or socioeconomic position, are not pre-determined existing entities; rather, they are socially constructed by people through their talk-in-interaction (Garfinkel, 1967;Kitzinger, 2000). As such, within the research fields of DP/CA, it is controversial to pre-categorize participants based on pre-defined categories such as gender, race, or indeed, socioeconomic position. It is argued that, in an attempt to firmly ground the analysis in the interactional data, participants' characteristics and/or particular contexts in which an interaction takes place should only be addressed in an analysis when oriented to, i. e., made relevant by, the interactants themselves (e.g., Schegloff, 1997;Wiggins, 2002b). However, there is an increasing volume of conversation analytic work in which contextual factors are not explicitly oriented to by the interactants themselves, but are addressed in the analysis, as they are observed as being relevant for the interaction by the analyst (e. g., Cameron, 2008;Flinkfeldt, Parslow, & Stokoe, 2021;Whitehead, 2020). Kitzinger (2000) argues that only describing particular forms of talk as belonging to specific categories when the interactants orient to it as such, would be very limiting. In fact, how various assumptions are routinely incorporated into everyday conversations without anyone noticing or responding to them, could be particularly interesting and relevant for analysis (Kitzinger, 2000).
The authors' approach to the present research is in accordance with the latter viewpoint. In the present research, children and parents did not explicitly orient to socioeconomic position in their everyday interactions. In line with research by Kitzinger (2000) and Whitehead (2020), we argue that this does not mean that it could not be relevant for the interaction. Rather, it might be that assumptions specific for families with a low SEP are routinely incorporated into their everyday conversations. As they fit into the worldview of other family members, they are not explicitly noticed or responded to. What is not oriented to by interactants could, in fact, reveal taken-for-granted social and interactional norms. Moreover, considering that 'socioeconomic position', and even more so the categorization in high, middle or low, is invented by scientists based on measures such as education, income and occupation (Shavers, 2007), it cannot be expected that participants will make this relevant themselves in everyday interaction. It is not a participants' category, to which one naturally belongs, such as to a specific gender, race or age group. Thus, although not explicitly articulated in everyday conversation, there is still a possibility that interactional patterns between children and parents might differ between families from various SEPs. Conclusions regarding whether and to what extent families from various SEPs treat for example food likes and dislikes differently and which actions are accomplished, should be based on actual research in families with various SEPs rather than being assumed based on theoretical approaches. To clarify, we do not imply that interactional patterns identified in children and parents from the present research are somehow 'dictated by' the low SEP of the families, as if it were a causal relationship. Rather, interactional patterns identified in this research are present at least in, and might be specific for, families with a low SEP. The present research provides a starting point for more in-depth investigation of how families with various SEPs talk about food.
A second, related point is that the SEPs of the families analyzed by Wiggins (2014) were not collected or not available. As such, a direct comparison between interactional patterns based on SEP is unfortunately not possible at the moment. We propose that it would benefit discursive psychological and conversation analytic research to enquire into the distribution of practices across socio-demographic categories, with a focus on underrepresented populations. This could unravel whether interaction patterns are more prominent in, or more explicitly oriented to in, e.g., specific age groups, people with a certain worldview, people with a certain socioeconomic position, or otherwise. This discussion has received more attention recently, for example in this recent blog within a community of interactional researchers, in which the wide availability of data from white participants with an Anglocentric worldview is discussed (Sciubba, Shrikant, & Williamson, 2021). Moreover, participants with a low SEP are likely to be underrepresented in research where no specific attention is paid to the SEP of the participants, considering that it generally takes a lot of time and effort to include families with a low SEP in research (Stuber, Middel, Mackenbach, Beulens, & Lakerveld, 2020).
Nonetheless, on a note of cautious interpretation, we would like to offer a brief reflection of our findings compared to earlier research on perceptions of taste among mothers of low, middle and high SEPs (Van Otterloo & Van Ogtrop, 1989). Van Otterloo and Van Ogtrop (1989) interviewed mothers and found that whereas mothers with a low SEP believed that their children's taste was a fixed given, mothers with a higher SEP believed taste was moldable and could be developed over time (Van Otterloo & Van Ogtrop, 1989). Following the theoretical assumption within cognitive psychology that behavior is influenced by cognition, we might expect that parents with a low SEP, who believe taste is fixed and thus cannot be changed, and parents with a higher SEP, who believe taste is moldable, would exhibit different behaviors towards their children's expressions of likes and dislikes. More research is needed to establish whether this is the case; however, a comparison between the findings of our study and findings from Wiggins (2014), who did not specifically topicalize SEP, provides very early and preliminary evidence that this might not be the case, as the findings from both studies did not show strikingly opposite results but rather complement each other.
Moreover, if parents believed that taste is fixed, then logically there would be no difference in how they treated likes and dislikes. But the present research showed that these are treated very differently, as likes are predominantly not responded to and dislikes are predominantly treated as food refusalsthus, in natural conversations, likes and dislikes are not two sides of the same coin. Parents may believe that taste in the sense of food preference is fixed when they are asked (Van Otterloo & Van Ogtrop, 1989), possibly derived from assumptions that taste is routed in, e.g., biology, habits, or a combination thereof, and is accessible through and reflected in language. However, when observing real life situations like the present research, we see that parents treat children's displays of taste as actions within feeding activities that have to be dealt with in situdislikes are for example treated as potentially causing feeding problems, that have to be 'solved' right away. This is especially relevant considering that feeding children is a primary task and responsibility for parents. This confirms that people treat language-in-use as action, with real consequences for their behavior in everyday life. Moreover, this shows that we should not rely solely on measures built on people's accounts of what they believe, because these do not capture the action dimension of language and thus run the risk of failing to provide a comprehensive understanding of human behavior, e.g., how people treat displays of taste, and why parents hardly respond to children's food likes but address dislikes.
All in all, our findings suggest that children's likes and dislikes are not symmetrical alternatives (as depicted by for example a 5-point Likert scale) and thus we might want to reconsider the way we operationalize and measure food preferences. As already noted by Wiggins and Potter (2003, p. 515), measuring food preference on rating scales forces participants into "a particular language game of semantic differentials and numerical judgements", and "does not test the possibility that food evaluation in natural situations may be done in as parts of very different practices". The latter is clearly shown in the present research. Moreover, as also noted by Wiggins (2001) and Wiggins and Potter (2003), a discrepancy may exist between researchers' and research participants' interpretation of what constructs are actually measured in food preference research, using specific evaluative termswhile researchers may be convinced that they measure a stable cognitive state or inner sensation, participants may refer to the last time they ate a particular food, actions accomplished with specific evaluation terms, and/or the interactionally achieved evaluation of that food.
Everyday interaction is highly organized following common-sense taken-for-granted interactional norms (Garfinkel, 1967). It is likely that parents are not always aware of when they respond to their children's assessments, and how they negotiate epistemic access, rights and responsibilities surrounding some of their children's food assessments. Therefore, a practical implication derived from the present research could be the development of a dialogue training for parents. The aim of such dialogue trainings is to increase awareness of speakers' own discursive practices and interactional patterns, by aiding them step-by-step to systematically reflect upon (un)intended effects of their own talk-in-interaction (Mogendorff, TeMolder, Van Woerkum, & Gremmen, 2016). The Discursive Action Method (DAM) (Lamerichs, Koelen, & Te Molder, 2009;Mogendorff, Te Molder, Van Woerkum, & Gremmen, 2016) or Conversation-Analytic Role-Play Method (CARM) (Stokoe, 2014) provide suitable formats that can be adapted to a dialogue training for parents. A dialogue training, developed from DP/CA analyses, provides a useful means to increase parents' understanding of the environment in which food is discussed (including likes and dislikes), served and consumed. Critical reflection by parents on how matters of taste (i.e., likes and dislikes) are dealt with in everyday life could, for example, be a component of future interventions aiming to improve eating behavior in families with a low SEP. However, as discussed earlier, it is not substantiated that the findings of the present research apply exclusively to families with a low SEP. As such, the dialogue training could also be suitable for families with a middle or high SEP. However, since families with a low SEP consume relatively poor diets compared to families with a higher SEP (e.g., Van der Velde et al., 2019) and consequently the most health benefits could be gained in families with a low SEP, such a dialogue training could be particularly suitable for families with a low SEP. Based on the present research, interactional practices that could be addressed in a dialogue training are, e.g., how parents respond to children's food likes and dislikes and the (un)intended effects on the interaction. The present analysis showed that children's food likes and dislikes were predominantly not responded to or treated as food refusals or non-genuine claims, respectively. They could, however, also provide an opportunity for dialogue. Rather than resisting, children's assessments could also invite exploring: they could, for example, be treated as invitations to discuss matters of children's taste, taste development, or shared decision making with regard to future meals. In such a training, also potential implications of children's and parents' everyday talk-in-interaction on how children learn about taste preferences and their relative rights to articulate those might be discussed.
As any research, the present research has strengths as well as limitations. The present research is the first to deploy an interactional approach to studying food preferences, i.e., likes and dislikes, in children with a low SEP. The study gives a unique insight in how children from families with a low SEP express (dis)likes of foods, and how these are understood and treated by their parents. It contributes to a deeper understanding of how food preferences are interactionally constructed in the course of everyday life in families with a low SEP. Although the study presents a detailed analysis of the most common parental understandings and treatments of children's food assessments as they occurred in the data corpus, it does not present an exhaustive list of possible responses. In addition, inherent to in-depth qualitative research, the findings are not intended to be generalized to other populations and settings. Finally, our collection of cases featured only a few assessments about food items in general. Although these cases showed a pattern, more cases could yield more robust claims regarding how assessments about food items in general are designed and treated, and how they offer a platform to negotiate epistemic access, rights and responsibilities.
The present research indicates niches for future research. First, more research is needed to establish whether and which differences may exist in interaction patterns between families with different SEPs, for example how they deal with likes and dislikes, and how this relates to their articulated beliefs about taste. Such insights would be beneficial for research in the field of discursive psychology as well as cognitive psychology. Although interactional research, such as discursive psychology and conversation analysis, asserts that interactional patterns transcend linguistic and cultural diversity (Schegloff, 2007), as elaborated upon earlier in this section, it becomes increasingly acknowledged that context such as cultural or socioeconomic background may shape how constructs are made relevant in interaction (Kitzinger, 2000;Whitehead, 2020;Pomerantz, 2021). In addition, our findings indicate an apparent gradient in the relevance for a parent to respond to an assessment, depending on the type of action accomplished. Some assessments that served as vehicles for other actions such as food refusals yielded elaboration from parents, whereas other assessments that might for example function as a compliment, were not responded to. More research could unravel the nature of why some actions, implemented by assessments, are treated as relevant to respond to, while others are not. A clue might be found in children's relative rights to engage in interactions with adults, and/or the 'urgency' of the social action accomplished with the assessment (e.g., Butler & Wilkonson, 2013;Eilittä et al., 2021;Sacks, 1995). Furthermore, future research could explore if and how the age of children influences how parents orient to and treat their food assessments. Lastly, in addition to the liking and disliking of food, it could also be explored which discursive practices are deployed by children and/or parents during everyday family mealtimes with regard to other aspects of food, such as the food's healthiness.
To conclude, we would like to emphasize once more the importance of conducting food preference research in populations with low SEP, especially in children. Children with a low SEP consume poorer diets than children with a higher SEP (e.g. Van der Velde et al., 2019), and populations with a low SEP are least reached by lifestyle interventions (Beauchamp et al., 2014;Bukman et al., 2014). Part of the problem may be that lifestyle interventions are not sufficiently tailored to the complexity of people's everyday life (Bouwman, Te Molder, Koelen, & Van Woerkum, 2009;Bukman et al., 2014). Simultaneously, populations with a low SEP are relatively underrepresented in current research. It is clear that there are socioeconomic differences in food preference and consumption; however, very little is known about how food preferences develop in the course of everyday life. The present research showed that in everyday mealtime conversations, children's food likes, dislikes and assessments about food items in general accomplish different social actions and parents treat them in different ways. While likes were routinely not responded to at all, were agreed with or further elaborated, dislikes were predominantly oriented to as food refusals or treated as inappropriate, or non-genuine claims. This underlines the action orientation of language-in-interaction, and shows that likes and dislikes are not two sides of the same coin when expressed in everyday life. This, in turn, highlights the potential limitations of relying solely on cognitive measures in food preference research, as the present research illuminates that there is more to 'food preference' than merely the representation of psychological states. As many parental elaborations on their children's food assessments refer to the accomplished social action and/or relative rights to know and to assess, rather than to the evaluation of the food as suchit becomes apparent that although '(non) tastiness' is discussed at the dinner table, it appears that these discussions are frequently not about taste at all.

Author contributions
AH, GJ and HM designed the study. AH, BH and HM analyzed the data. AH drafted the manuscript. All authors approved the final article.

Funding
This work was supported by the Edema-Steernberg Foundation, Wageningen, the Netherlands. The funding source had no involvement in the study design, the collection, analysis and interpretation of data, in the writing of the report and in the decision to submit the article for publication.

Data code and availability
The research data underlying this manuscript, audio and video recordings, are confidential and cannot be made publicly available. Upon request, the recordings are available in an anonymized format for peerreview. The lead author has full access to the data reported in the manuscript.

Ethical statement
This article involved audio and video recordings of children and parents. All family members were informed about the purpose of the study (i.e., to capture naturally occurring conversation during mealtimes) and were given ample opportunity to ask questions. All recorded family members provided written informed consent (written informed consent for children was provided by their parents). This study was approved by the Medical Ethical Review Board of Wageningen University & Research, the Netherlands (METC-WU, file number: NL64893.081.18).

Declaration of competing interest
None.