The eating model in contemporary Taiwan: Destructuration or a new insight into compressed food modernity in Asia

the ‘compressed ’ food modernity that Taiwan experienced. This study is the first attempt to provide comprehensive data about the eating model in Taiwan. Further empirical studies, particularly in other Asian regions, are expected to advance our thinking about a complex relationship between food modernity and well-being.


Eating model in reflexive food modernity
At first glance, food modernisation has seemed to enrich our dietary lives by increasing food production and dietary information.However, a deeper analysis has revealed an unexpected consequence of the so-called 'gastro-anomie', namely the situation in which contemporary eaters are loaded with too many choices, causing them to be puzzled about what to eat and how to lead good dietary lives (Fischler, 1990;Poulain, 2017).
Moreover, the promotion of nutritional knowledge has facilitated the rationalisation of our dietary lives, but the tendency toward nutritionism or medicalisation has also relegated the aesthetic (non-nutritional) aspects of eating to secondary concerns (Scrinis, 2008).To take an example from Japan, the standardisation of the modern family led to the development and universalisation of modern dietary norms, such as having three meals a day, eating together with family members and the central importance attached to home cooking (Cwiertka, 2007).However, now that family structures and food systems have experienced dynamic changes, these modern dietary norms, without the necessary social conditions for their practice (notably housewives), have also become the source of moral dilemmas rather than simply being informative of dietary well-being (Ueda, 2023a(Ueda, , 2023b)).
These contemporary social conditions have often been called 'reflexive' food modernity.Reflexive modernity refers to the situation in which one needs to live with the radical speed of modernisation and, at the same time, confront its side effects (Beck, 1992;Giddens, 1990).This perspective has guided food scholars to tackle contemporary food issues, such as food risk and anxiety, the individualisation of eating, the prevalence of eating out and the globalisation or localisation of food cultures (Poulain, 2002(Poulain, , 2012(Poulain, , 2017(Poulain, , 2019;;Figuié et al., 2019;Louis & Poulain, 2018).
Due to its multifaceted nature of modernity, it is hard to give a single definition to food modernity.Instead, it is possible to identify some shared understanding (for its review, see Ueda, 2023c): (i) Traditional dietary norms, which used to be reproduced within families and communities, are no longer considered legitimate and are constantly examined in the light of new information about dietary practices.(ii) This weakening of the regulatory function creates the above-mentioned state of 'gastro-anomie', in which eaters confront an overabundance of changing and contradictory norms and resultant food anxiety.(iii) New regulatory bodies, such as the nutritional sciences, heritages/traditions, politics and the nation-state, are then required to alleviate such gastro-anomic symptoms, which creates a picture of various food modernity phenomena, including the medicalisation, heritisation (the return to traditional food practices) or politicisation of food, and the standardisation of eating models.(iv) These modern trends keep advancing and eventually arrive at some contradictory stage at which eaters need to confront the side effects of such increasing modern trends and their gastro-anomic situations are exacerbatedwhich modernity stage is called as reflexive food modernity (or the second food modernity to differentiate itself from the first food modernity at the initial stage).
Our eating model, which refers to a more total focus on the temporal, spatial, social, qualitative and affective aspects of eating than the scopes in traditional dietary assessment, is subject to such food modernity trends and thus serves as a good analytical theme for understanding the nature of reflexive food modernity (Herpin, 1988;Poulain, 2002aPoulain, , 2017a;;Warde, 2016). 1 While its ambition to capture the 'totality' of eating (Morin, 1973) is often compromised to some extent in its operationalisation, Poulain (2002a) ventured to demonstrate the systematic analysis of eating model in contemporary France by setting five major evaluative dimensions, including meal structure, time (timing and duration), social environment (who to eat with), locale (where to eat) and type of choice (who decides the meal).Inclusion of these wide-range evaluative dimensions was necessitated to respond to the social controversy about various dietary changes such as meal skipping, solo eating, eating out and simplified mealswhich have been subsumed under the broad concept of 'destructuration' of the French eating model.While standardised eating models are the product of the first food modernity (e.g., nation-state, nutrition education, modern family), its destructuration, as the prefix 'de-' connotates, refers to rupture from the 'standard' eating model and calls for its reconsideration; thus, it can be viewed as a symptom of reflexive food modernity (Poulain, 2002a).This French study invited further exploration of eating models in other regional contexts (Fischler & Masson, 2008;Mestdag, 2005;Mognard et al., 2023;Ueda, 2022aUeda, , 2022b)).

Compressed food modernity in asia
Although we have thus far assumed the (Western) universality of food modernisation, scholars have recently started investigating the plurality of modernisation, which has differing logics regarding change, the duration of the process and its consequences (Beck & Grande, 2010).Of particular importance to Asian countries is the thesis of 'compressed modernity', that is, 'a civilisational condition in which economic, political, social and/or cultural changes occur in an extremely condensed manner in respect to both time and space, and in which the dynamic coexistence of mutually disparate historical and social elements leads to the construction and reconstruction of a highly complex and fluid social system' (Chang, 2016, p. 33).This thesis has been empirically tested by scholars of two academic traditions that are relevant to the current study.
The first group of studies was carried out within the framework of the Asian Food Barometer, led by Jean-Pierre Poulain.The objective of this project was to develop basic data on eating models in Asian countries for comparative analysis.The complete reports of this project have so far been published only in Malaysia (Poulain et al., 2014) and Indonesia (Khusun et al., 2023).The Asian Food Barometer project is currently centred on southeastern Asian countries, while only partial aspects of eating models are addressed for East Asia (Japan, China and Hong Kong), notably the act of eating out (Mognard et al., 2023).Crucially, no comparative data is currently available about Taiwan.
The other relevant group of studies is the Asian comparison of family and intimacy, led by Japanese sociologist Emiko Ochiai.Although food is not the central issue of this large project, quite relevant to our study is their theoretical output regarding the concept of the 'semi-compressed modernity' that is facing Japan, while characterising Taiwan as one of the typical cases of compressed modernity (Ochiai, 2009(Ochiai, , 2014(Ochiai, , 2019)).One considers two fertility drops (i.e., demographic transitions) as indicators of the first and second (i.e., reflexive) modernities.The former drop represents the prevalence of the modern family, while the latter marks the deconstruction of such a family model due to multiple factors such as de-housewifisation.In this case, Japan can be situated in a middle position, with a stable period -Ochiai called it elsewhere the 'Golden Age of the first modernity' -of about 20 years until the second two modernity (1950− 1970s), as compared with about 50 years in Europe (around the beginning of the 20th century to the 1950s) and almost no period in other East Asian countries. 2Based on this demographic, Ochiai (2014: p. 224) diagnoses that 'Japan's semi-compressed modernity had made it difficult for Japan to extricate itself from its first modernity and take a step forward toward a new stage of social development'.This thesis is convincing, given the largest gender gap and the highest rate of relative poverty among single mothers in Japan of all the high-income countries (WEF, 2022;OECD, 2022), both of which are characteristic trends of the second modernity (Giddens, 1990).Since food is an integral part of family life, further work is needed to understand how this situation applies to the analysis of contemporary eating.

Eating well, dietary well-being studies
Somewhat independently of these sociological studies on food and family, there is also a growing body of literature on well-being in the context of food.Although the various disciplines (e.g., nutrition, psychology, economics) apply differing approaches to well-being with different but interchangeable vocabularies (e.g., 'food-related wellbeing', 'dietary well-being', 'eating well'), their shared objective is to reassess what good eating models means to the contemporary population (Dodge et al., 2012;Lindert et al., 2015;Meiselman, 2016;Ueda, 2022b).These studies are highly relevant to reflexive food modernity, because 'reflexive project of the self' (Giddens, 1991), that is, a constant reconstruction of one's own (dietary) life and self-identity, becomes the central question in the world without any legitimate eating model.
Importantly, some researchers have pioneered international comparisons of dietary well-being, their commonality and cultural differences (Ares et al., 2015(Ares et al., , 2016;;Jaeger et al., 2022;Lappalainen et al., 1998;Rozin et al., 1999).While there are non-negligible cultural differences, such as in preferences for particular food items or in attitudes towards alimentary pleasures, these studies have also confirmed a 1 Eating model (modèle alimentaire) is the analytical framework that was originally developed in the French school of the sociology of food.The core idea behind this framework is to highlight socio-cultural aspects of eating and to overcome the limited scopes in traditional dietary assessments.Eating model thus contains some similarities with other encompassing terms such as 'foodways' and 'practice of eating', which are more or less popular in the Anglo-Saxon worlds, and with other everyday terms such as 'meals' and 'diets' which are the constituents of eating model (for further information, see Poulain, 2002b;2017b). 2While the first fertility drop continued to take place during the period of 1950-1980s, the second fertility drop (below replacement level) occurred with almost no intervals in the mid-1970s in Singapore, in 1980s in Taiwan and South Korea, and in 1990s in China (Ochiai, 2014;Lee, 2009).
H. Ueda and Y.-C. Chiu cross-cultural similarity in perspectives on dietary well-being.In general, eating well 'tends to be conceptualised as a balanced and varied diet consisting of healthy foods, fresh and natural foods, plenty of fruit and vegetables, and food containing vitamins and minerals' (Povey et al., 1998, p. 181).This universality was constructed partly due to 'the fact that health associations are strongly related to dietary guidelines, which are similar in different parts of the world' (Ares et al., 2016, p. 70).In other words, they are the invention of the modern nutritional sciences.
Now that food (in)security is a global agenda that needs to be addressed from the multiple spectrums of availability to access, utilisation and 'well-being' (FAO, 2009), this type of cross-cultural study can contribute to enriching our understanding of dietary well-being and the possibility of international solidarity.However, note that previous comparative studies were centred mostly on Western countries, while Asia was occasionally included and, in such cases, further investigation was called for (Jaeger et al., 2022;Rozin et al., 1999).Among studies on dietary well-being in Asia, Ueda (2022aUeda ( , 2022b) ) provided perhaps the most comprehensive insight into eating well in Japan.As a part of our long-term mission to extend the regional scope of well-being studies to Asian countries, in the current paper we aimed to replicate Ueda's study in the analysis of the eating model in Taiwan.

Contemporary food issues in Taiwan and national statistics
Although no study is available about a comprehensive eating model in contemporary Taiwan, many researchers have reported certain aspects of reflexive food modernity.During the past two decades, a series of food safety scandals have damaged Taiwanese consumers' trust in the national food system.More than 80% of Taiwanese were anxious about the safety of everyday foods, and these consumers were even more distrustful of so-called 'false' food certifications (Chiu & Yu, 2019;Fu, 2013).Behind food anxiety lies an increased distance between farms and tables, and, moreover, between kitchens and tables (i.e. the decline of home cooking practice).Rice used to be the major agricultural product for the Taiwanese food system, but its production was reduced by half from the 1970s to the 1990s, and contemporary Taiwanese consume the smallest portion of rice of all the East Asian countries (FAO, 2023).Moreover, since its integration into the World Trade Organisation framework in 2002, national agriculture has been under increasing competition with foreign products.Recent global instability due to the COVID-19 pandemic and Russia's invasion of Ukraine has also challenged its national food systems.
Population nutrition and health have also become serious issues.The Nutrition and Health Surveys in Taiwan (hereafter, Nutrition Surveys) have revealed that, for men, obesity rose from 10% to 28%, diabetes from 3% to 12% and hyperlipemia from 20% to 29% from the mid-1990s to the mid-2000s (Pan et al., 2011).A possible underlying reason for this health problem is the evolution of eating models, notably the increase in the practice of eating out (e.g., Huang et al., 2017).The most recent Nutrition Survey (2017-2020), confirmed a high prevalence of eating out.It was reported that the working population (aged 19-44 years old) who 'eat out five times or more per week' amounted to 56% for breakfast, 60% for lunch and 31% for dinner (Pan, 2022).
Recently, new food policies, such as food education, have been initiated in Taiwan to address all these food issues characteristic of the reflexive food modernity facing Taiwan (Ueda, 2023d).Due to this sociopolitical context, the development of basic data about dietary norms and practices among the population is more necessary than ever.For this purpose, some relevant national statistics are available, but they do not capture the eating model comprehensively.Table 1 presents a list of national statistics in Taiwan that are relevant to certain evaluative dimensions of the eating model.
Perhaps the most established survey in Taiwan is the Nutrition Survey, which has been conducted every few years since 1980.This survey is composed of dietary and physical assessments but is basically centred on nutritional status.Unlike the Japanese one, the dietary assessment part does not include a wide range of question items that cover the totality of eating.
Conducted annually since 1976, the Survey of Family Income and Expenditure (hereafter, Family Survey) is to some extent informative for meal content and place of eating, which can be calculated based on the combination of food products consumed.However, ambiguity remains regarding the interpretation of the data, such as the breakfast/lunch/ dinner distinctions and the frequency of each meal.Of particular note is the Food Consumption Survey (hereafter, Food Survey) which began in 1998 and has been conducted annually as part of the Council of Agriculture's digitalisation policy.While it has the strength of covering various dimensions of eating models (meal frequency, timing, place of Annual consumption per capita of certain foods (such as rice and meat) Note 1: Evaluative dimensions corresponded with those addressed in Japan's eating model survey (Ueda, 2022a).Source: Developed by the authors based on the most recent statistics as of March 2023.
H. Ueda and Y.-C. Chiu eating and quality), some caution is required when evaluating the sizeable annual data variations.Although acknowledging other useful statistics (e.g., Census of Commerce, Statistics of Finance, Food Balance Sheet), these three surveys currently play a major role in informing us of eating models in Taiwan.We shall discuss the concrete modality of each survey and question item, if needed, when describing and interpreting the results.

Research objectives
Theoretically framed by the food modernity perspective and its mission to explore plural food modenities in under-represented regions, the objective of the current survey was to replicate Ueda's (2022a) previous survey and elucidate the eating model in contemporary Taiwan.The latter was analysed in terms of norms and practices, as well as their cultural commonality with and differences to Japanese eating model.In particular, we aimed to assess whether the destructuration of eating model can be observed in Taiwan and, if so, how far such trend is progressing and being influenced by socio-demographic status, namely gender and age, in comparison to the Japanese situation.

Questionnaire development
The web-based questionnaire was the same as the counterpart survey in Japan (Ueda, 2022a(Ueda, , 2022b) ) except for some cultural adaptation (to be specified below).The translation of the questionnaire into traditional Chinese went through the following steps: (i) The first draft (Chinese) was translated from the source version (Japanese) by one author who is bilingual (native: Japanese, learned with proficiency: Chinese); (ii) the first draft was revised by the other author who is native in Taiwanese and all disagreements were resolved by discussion to develop the provisional version; (iii) in so doing, these authors integrated two types of cultural adaptation, the first changes concerning concrete examples to facilitate participants' answers (e.g., food items, food shop categories) and the second revisions concerning question items about socio-demographic status to be linked with Taiwanese national statistics; (iv) the provisional version was tested for its validity by performing preliminary interviews using this questionnaire with 30 Taiwanese of various backgrounds (including Japanese-Chinese bilinguals) who were recruited in the snowball sampling method; (v) the provisional version was modified based on these preliminary interview insights and the final version was developed through discussion and validation by the authors.
The structure of the questionnaire was the same as the original survey in Japan (see Ueda, 2022aUeda, , 2022b, for full descriptions), so here we limit ourselves to describing only the essential information about the survey protocol.The questionnaire consisted of a first part with open-ended questions about 'eating well', a second part that comprehensively covered 10 dimensions of eating well and a third part on the respondents' socio-demographic status.The second and third parts were the target of analysis for this paper.The lexical analysis for the first part was excluded due to space limitations.
Five dimensions of eating models, initially proposed by Poulain (2002a), were later expanded/sub-divided to 10 dimensions in the subsequent empirical studies (Ueda & Poulain, 2021;Ueda, 2022b), which are applied to our study as well: (1) meal frequency, (2) place of eating, (3) timing of meals, (4) meal duration, ( 5) persons to eat with, (6) place of procurement, ( 7) quality of food, ( 8) pleasure of eating, ( 9) meal content and (10) nutritional status.Each evaluative dimension was explored in terms of dietary norms and practices.Here, 'norms' signify the person's desired/valued level of achievement, whereas 'practices' refer to their actual achievement.These achievement levels could be determined by asking questions such as 'How many times per day would you like to eat?' (norms) and 'How many times did you eat yesterday?' (practices).Only data about dietary norms regarding nutritional status was not obtained because it can be safely assumed that being more nutritious and healthier is always better.
This focus on the norm-practice discrepancy is not entirely new because it has already been viewed in previous studies as a contemporary symptom of reflexive food modernity (Poulain, 2002a, p. 53) or the 'overly optimistic perceptions' that inhibit health promotion (Povey et al., 1998, p. 182).Adding to existing interpretation, the norm-practice gap can be viewed as one's capability to eat well or the freedom to achieve one's valued way of dietary life (Ueda, 2022a).We shall revisit this complex nature of dietary well-being in relation to the empirical findings.Each question's wording and treatment method are described in the results section and in the corresponding tables.

Data collection
The procedure for conducting this web-based questionnaire was the same as that for Japan's survey (Ueda, 2022a(Ueda, , 2022b)).The participants were recruited via an e-mail invitation letter ('Dietary Life Survey') sent by the research company, GMO Research, Inc, which has a wide coverage of service in Asia.The email included an explanation of the modality of the survey and a link to the questionnaire sheet.Once the participants had agreed with the objective and content of the survey, they provided their informed consent by clicking on the survey's front page and then completed the questionnaire sheet.The survey was conducted in March 2023.As stated in the invitation letter, the participants received incentives (roughly 1-5 dollars).As is the case with the counterpart survey in Japan, the inclusion criteria were not being a student and being aged 20-69 at the time of participation. 3  For this study, 976 Taiwanese participants were recruited from the database of the research company, and 56 ineffective responses (e.g., 'I forgot what I had yesterday') were excluded.Overall, 920 effective responses were used for the analysis.This survey was in accordance with the Declaration of Helsinki and approved by the Ethical Committee of the Graduate School of Environmental Studies at Nagoya University.
As indicated in Table 2, the socio-demographic profiles of the respondents largely represented the national average of the population in Taiwan.Nevertheless, note that the socioeconomic status (SES), namely, educational levels and socio-professional categories (excluding the income level), of the respondents was relatively higher than the national average, mostly due to the nature of the web-based questionnaire that require a certain level of computer literacy and internet access.Therefore, some caution is needed when generalising the findings as a social reference in Taiwan.On the other hand, this sample deviation might not be an obstacle to making comparisons with the Japanese study (Ueda, 2022a(Ueda, , 2022b) ) because the latter web-based survey also involved a population with relatively higher SES for the same reason.

Statistical analysis
Statistical analysis was carried out to investigate the following two issues.The first was to confirm the norm-practice discrepancy, whereas the second was to assess the effects of socio-demographic status on the achievement levels of eating well.Among various socio-demographic variables, gender and age, which were highly relevant in the Japanese study (Ueda, 2022a), were selected for testing in our study.
Regarding the norm-practice gap, the Wilcoxon signed-rank test (a nonparametric test for paired samples) was used for the quantitative parameters, while McNemar's test was applied to the categorical/binary data (i.e., meal skipping, eating together, meal content [each dish pattern]).In terms of the sociodemographic effects, multiple regression 3 The population aged over 70 was excluded due to their general limited computer literacy.Also, our study was focused on non-student population, because empirical studies on their dietary conditions have been relatively poor when compared to student population in our targeted regions.
H. Ueda and Y.-C. Chiu analyses were performed for the quantitative dependent variables (such as meal duration: 10− 120 min), whereas binominal logistic regression analyses were performed for the binary ones (that can be expressed Yes: 1 or No: 0).As independent variables, gender and age factors were converted into five dummy variables in all the analyses: 'women', '30s', '40s', '50s', '60s' (baselines: 'men' and '20s'), all of which were entered into the model (forced entry method).While we set 5% as the significance level for testing the norm-practice gap, we made a Bonferroni adjustment to 1% as the significance level for testing the sociodemographic effects in order to avoid the problem of multiplicity (Tsuchiya, 2014).The R software (ver.3.1.2) was used for all the analyses.

Results
The result of each evaluative dimension is described in relation to actual practices and their relationship with norms.Each social reference (such as median or average) in Taiwan is also compared with the Japanese social reference (Ueda, 2022a), as well as Taiwanese national statistics (if available).Unless specifically mentioned, the most recent statistics are referenced.

Meal frequency
The survey involved two approaches to understanding meal frequency (Table 3): a direct question about daily meal frequency (0− 3 times) and weekly meal frequency (0− 7) calculated from the 'place of eating' question.Note that the first question was unavailable in the Japanese survey and was thus derived from a survey with the same methodology that was targeted at Japanese single mothers, a specific population with a high rate of meal skipping (Ueda, 2023a(Ueda, , 2023b)).A conceptual distinction was introduced between habitual skipping (less than three meals per day) and occasional skipping (less than seven meals per week). 4The Food Survey can provide data for both types of meal skipping.
In reality, 37% of the Taiwanese population were habitual meal skippers.The high prevalence of meal skipping can be instantly understood if one compares it with, at best, 28% of Japanese single mothers, one of the most relevant social groups to skip meals habitually.This relative prevalence of meal skipping in Taiwan was also confirmed on a weekly basis; for example, 37% were occasional skippers for one of the three meals compared to 25% in Japan.The 'by-meal' distinction also informed us of the order of the prevalence of meal skipping, from  Note 1: For the national reference in Japan, see Ueda (2022a).Only the daily data was retrieved from the survey of single mothers (Ueda, 2023a(Ueda, , 2023b)).Taiwan-Japan comparison and is subject to a critical reading of the standard meal frequency in the given society.
H. Ueda and Y.-C. Chiu breakfast to lunch and dinner.We calculated a higher prevalence than the Food Statistics (daily meal skipper: 15− 25% during the recent decade), which might have been due to the methodological difference.The Food Survey did not distinguish the norms and practices, so it is assumed that the respondents' answers tended to be 'proper' (i.e., three meals per day) and this psychological reframing thus underestimated the reality of meal skipping.
Further analysis of the dietary norms revealed an interesting phenomenon.Although it is true that the norm-practice gaps were nonnegligible in Taiwan (p < 0.05), the gaps regarding habitual meal frequencies were relatively smaller than those in Japan, which means that the 'three meals per day' norm is relatively weakened in contemporary Taiwan.

Place of eating
Eating places were classified into four groups by 'where to cook' and 'where to eat', the first two of which ('cooking and eating at home' and 'cooking at home and taking/eating out' ['taking out from home']) were together called 'Eat-In', while the latter two ('taking in' and 'eating out') were related to the outsourcing of meal cooking and together called 'Eat-Out' (Table 4).
The Eat-Out practices amounted to 2.5 times per week for breakfast, 3.4 times for lunch and 2.6 times for dinner, all of which were more frequent than in Japan.Interestingly, no norm-practice gap with a statistical difference was observed, representing the absence of dilemmas about home cooking and eating out.
To make the data comparable with the Nutrition Survey, we performed some recalculation.If all three meals were considered (0− 21 times per week), our survey showed that 14% had 0 time for Eat-Out, 30% had 1− 6 times, 31% had 7− 13 times and 25% had 14 or more times.These numbers were more-or-less consistent with the Nutrition Survey (11%, 21%, 35% and 32%, respectively). 5

Timing of meals
If the median value is taken as a reference here, the actual starting times were breakfast at 7 a.m., lunch at 12 a.m. and dinner at 7 p.m. (Table 5).The analysis of distribution (the first and third quartile in this case) revealed that the timing of meals in Taiwan was later for breakfast and earlier for lunch and dinner than in Japan.There were normpractice gaps for all three meals, representing that they wished to start meals early and with regularity but ended up with some delay.
This result was generally consistent with the Food Survey during the recent decade, which identified that about 75% of the population could start dinner before 7 p.m. (and almost no one after 8 p.m.).This early dinner constituted a sharp contrast with the social conditions in Japan, where about 20% of the population, mostly male workers, were left with no choice but to start dinner after 9 p.m. due to long working hours (Ueda, 2022a).a p < 0.05. 5 Other comparative data related to the prevalence of daily Eat-Out users (five or more times per week): We identified 30% for breakfast, 42% for lunch and 30% for dinner (aged 20− 49), while the Nutrition Survey identified 56%, 60% and 31%, respectively (aged 20− 44).While these two surveys agree with the high prevalence of eating out in Taiwan, we were unable to identify convincing reasons for the relative disagreement regarding eating out for breakfast and lunch.
H. Ueda and Y.-C.Chiu

Meal duration
If the median value is taken as a reference here again, the respondents spent 20 min on breakfast, 30 min on lunch and 30 min on dinner (Table 5).Given the data distribution, the meal durations in Taiwan were generally longer than in Japan.Nevertheless, the Taiwanese population still considered them short because norm-practice gaps were observed for all three meals.

Persons to eat with
Solo eating was the primary focus in the socialisation dimension.We identified 64% solo eaters for breakfast, 42% for lunch and 27% for dinner (Table 5).There were more solo eaters for breakfast but fewer for lunch than in Japan.Another cross-cultural difference was the absence of observed norm-practice gaps for eating alone for breakfast and lunch in Taiwan and thus the absence of dilemmas about this socialisation aspect experienced by the Japanese population.

Places of procurement
In descending order of frequency, the respondents procured fresh foods from supermarkets, traditional markets (professional retailers) and discount/hypermarkets (Table 6).This ranking structure was consistent with their norms, but, in reality, their access to traditional markets was constrained, while they relied more on convenience stores and online delivery than they had initially wished.The frequent use of supermarkets is consistent with Japan, but the similarly popular use of traditional markets in Taiwan constitutes a sharp contrast with the shopping conditions in Japan.
Despite their differences in measurement methods, national statistics (such as Food Survey, Census of Commerce and Statistics of Finance) were generally in agreement that supermarkets, traditional markets and convenience stores deserved almost equal importance (whether sales or preference) in food procurement in Taiwan; however, they did not distinguish between fresh and processed products.As far as the provision of fresh foods is concerned, the role of convenience stores is likely to be lower.In that sense, the results of our study were more-or-less consistent with the national statistics.

Quality of foodstuffs
'Good quality' is a polysemic concept that needs to be approached with reference to a multidimensional spectrum.Our approach was framed by convention theory (Allaire & Boyer, 1995), the sociological and economics theory that has influenced the recent food quality policies in Europe and other regions.This international coverage (e.g., Ponte, 2016) serves for the cross-cultural scope of our study.According to convention theory, quality can be understood in relation to the values of differing spheres (cités in French).Each sphere has unique criteria for defining quality: price for the commercial world; efficiency for the industrial world; spatial and temporal proximity for the domestic world; public good for the civil world; fame and recognition for the world of opinion; and originality (and 'naturality' as one of its forms) for the world of inspiration (hereafter called the 'natural' world).The question items to investigate food quality were selected based on previous studies (for details, see Ueda & Poulain, 2021;Ueda, 2022a).
With the mean value (more than 4.5 points) taken as a threshold here, the respondents most valued the natural qualities ('fresh', 'pure' and 'seasonal') and the commercial quality ('good price'), followed by the other quality criteria (industrial: 'standardised' and 'nutrients', domestic: 'producers are visible' and civil: 'certified') (Table 6).This ranking structure was mostly consistent with the case in Japan.The Food Survey also annually reported the relevant qualities to the natural world (e.g., fresh, natural), but, to disagree with our survey, the Food Survey somehow always ranked price as one of the least valued concerns by Taiwanese consumers.
In practice, about 70-80% of the respondents could use the foodstuffs with their valued qualities more than 'every other day'.Their usage of the foodstuffs with the civil ('socially good' and 'certified') and domestic qualities ('visible' and 'regional') was relatively low, but the general usage levels in Taiwan were better than in Japan.We shall later discuss their access to 'good quality' food and its interrelationships with the food environments characteristic of contemporary Taiwan.

Pleasure of eating
Among the various food behavioural steps, the respondents most valued eating (intake), followed by shopping and cooking (Table 6).A closer look at this intake dimension revealed that the gustative ('appreciating the taste of dishes') and corporeal pleasures ('feeling full') were most valued, followed by the relational pleasures with others and nature.This result was similar to the dietary norms in Japan.However, in Taiwan there was no relative decline in the achievement level of experiencing the connection with nature that was particular to Japanese eaters.

Meal content and meal pattern
In general, meal content can be evaluated in terms of nutrients, food groups or dishes.We chose the dish-based evaluation due to the need for simplicity, following the method frequently used in economic surveys (Kudo et al., 2017).This method was used to calculate meal patterns based on several predetermined categories of dishes (see the footnote of Table 7 for concrete categories).It was also used for the previous survey in Japan (Ueda, 2022a).For example, if one had 'rice, fish soup and stir-fried spinach' as the meal content, it was then calculated as the 'staple, dish and soup' pattern.
Regarding breakfast, the 'staple only' deserves special attention.A little more than half the Taiwanese respondents managed breakfast with this simplified meal and its rate was much higher than in Japan (Table 7).On the other hand, lunch and dinner tended to be richer in content than breakfast, in which frequencies of appearance were about 10-30% for each of the various meal patterns.If the 'staple and two or more dishes' pattern was taken as a reference,6 the meal content in Taiwan appeared to be more-or-less simpler than that in Japan; however, it requires careful interpretation.
A closer look at the total number of dishes (including staple) revealed that the Taiwanese actually had richer meals in terms of content, with an average of 3.0 dishes for lunch and 3.3 dishes for dinner, compared to the Japanese, with 1.9 for lunch and 2.6 for dinner.Moreover, in contrast to the rice-centred meal patterns in Japan, further analysis of staple content in Taiwan revealed that the frequency of flour-based meals (e.g., noodles, buns, crepes) was 1.3 times higher and 1.8 times higher than rice-based ones, which represented relative diversity in staple content.In sum, the meal patterns were more flexibly structured (or destructured) in Taiwan than in Japan.In Taiwan, there were frequent cases such as those who had 'no staple but three or four dishes', which were relatively rich in meal content but led to the low practice rate of the 'staple and two or more dishes' structured pattern calculated for the above-mentioned analysis.To give a typical example, the Japanese eat jiaozi 餃子 (Chinese dumplings) with rice, but the Taiwanese do not usually need rice with jiaozi.Therefore, contrary to the first impression, it can be safely said that the dinner content in Taiwan is richer than in Japan.
Another interesting cross-cultural difference is related to dietary norms.While similarly having norm-practice gaps, the Taiwanese did not suffer from such strong dilemmas about simplified meals as the Japanese eaters (for example, see the 'staple and two or more dishes' dinner pattern).We shall revisit these phenomena below and extract important implications for reflexive food modernity.

Nutritional status
Following the Japanese study (Ueda, 2022a), the nutritional status was estimated with the diversity variety score (1-6) that was calculated from the number of six food groups ('grains', 'beans, meat, fishes and eggs', 'dairy products', 'vegetables', 'fruits' and 'nuts and oils' [in accordance with Taiwan's nutrition policy]) taken every other day or more.If the scores were categorised into three groups, from 'high nutrition' (6 score) to 'middle nutrition' (3-5) and 'low nutrition' (0-2), we identified 22.1%, 63.3% and 14.5% for each category.Note that the score was used here to understand the respondents' relative nutritional balance, not the objective status (although the 'low nutrition' category referred to the level at which some dietary intervention might be required).

Socio-demographic effects
Due to space limitations, we selected the six most relevant items to be tested for socio-demographic effects on the achievement levels of eating well, namely habitual meal skipping (daily), solo eating for dinner, Eat-Out (eating out and taking in) for dinner, food procurement from traditional markets, the total number of dish components for dinner and nutritional status.The results of the statistical analysis are summarised in Table 8.
Meal skipping was more characteristic of the women than the men, and the young in their 20s were more likely to skip meals than the middle aged and elderly in their 50s and 60s.The young also practised eating out more often and procured less from traditional markets than the respondents in the 40-60 age group.Nutritional status was higher among the elderly (60s) than among the young.No statistical difference (with 1% significance level) was observed between solo eating and meal content.In sum, the gender-based inequality was not as large as in Japan, where men had much lower capabilities to eat well than women (Ueda, 2022a).However, the age-based inequality was still large in Taiwan, with the young people being most deprived of capabilities in various dimensions of their dietary lives.

The modality of eating well
The central question to be answered here was what eating well meant to the population in Taiwan.To address this, we have so far explored the 10 dimensions of their eating model.The valuable food functionings which constitute eating well for the contemporary Taiwanese can be identified based on our findings and are summarised in Table 9.
Existing cross-cultural comparison of dietary well-being (Ares et al., 2015(Ares et al., , 2016;;Jaeger et al., 2022;Lappalainen et al., 1998;Rozin et al., 1999) has been presupposed, to some extent, on the idea that a meaningful comparison of individuals' well-being can be made.Notably, the content of eating well in Taiwan had much in common with that in Japan (Ueda, 2022a) and this finding also reconfirms the idea of cultural commonality that can be utilised as a basis for food policy making (Sen, 1985;Ueda, 2022b).Nevertheless, the findings also revealed important cross-cultural differences, which we shall discuss below.
Having clarified the content of eating well in Taiwan, it is important to mention its epistemological nature.In the first place, this eating well needs to be distinguished from general dietary guidelines that are developed based on nutritional evidence.Instead, it is a social reference, which serves as a dominant norm in contemporary Taiwan that is open to public reasoning (Sen, 2009) and needs to be continuously scrutinised and modified through social debate and participation.As mentioned in the introduction, the modern nutritional sciences have enriched our dietary lives through scientific rationalisation but, at the same time, its focus on the nutritional dimension has also created conditions of gastro-anomy and impoverished other dimensions of eating (Fischler, 1990;Poulain, 2017a;Scrinis, 2008).What is needed in our reflexive food modernity is to regain a good balance between the scientific and social rationalities (Giddens, 1991).The contribution of this study is clearly in the latter paradigm.
In this reflexive context, the development of basic data on eating well is more necessary than ever.It can serve as a social reference for developing food policies and for helping individual eaters to reconstruct their own eating models.In Taiwan, national statistics relevant to eating models have failed to capture the totality of dietary lives with their limited focus on specific dimensions of eating.Our survey was the first attempt to provide a comprehensive data about the eating model in Taiwan.

Reflexive food modernity within eating well in Taiwan
The next major question to be answered was how, more concretely, the content of eating well was related to the reflexive food modernity in Taiwan.First, in line with other cultural contexts (Poulain, 2002a;Ueda, 2022a), the destructuration of the eating model was observed in contemporary Taiwan 7 : two out of five consumers habitually skipped meals; one out of four ate out 14 times or more in a week; and three out of five ate alone for breakfast.The analysis of norm-practice discrepancies also revealed an interesting phenomenon: the destructuration extended to not only their practices but also their dietary norms.In this sense, the nature of destructuration was somewhat different in Taiwan from those in France and Japan (Poulain, 2002a;Ueda, 2022a), in the latter of which countries contemporary eaters experienced dilemmas due to a conflict between their dietary ideals and realities.
We shall further elaborate on this point by taking eating out as an example.In Taiwan, everyday eating out has continued to increase since the 1980s.Fig. 1 presents the historical transition of eating out activities in Taipei City (the only longitudinal data available for this purpose), but it has been a nationwide phenomenon regardless of urbanisation levels and socio-professional category. 8Multiple conditions influenced the development of the eating out culture in Taiwan.The post-war Note 1: Binominal regression analysis was performed for the first two items, while multiple regression analysis was done for the rest.Note 2: Adjusted R 2 : 0.036 (Eat-Out), 0.047 (procurement), 0.001 (dishes) and 0.025 (nutrition).a p < 0.05.b p < 0.01. 7To obtain the complete picture of such destructuration, it is necessary to gather longitudinal data about eating models; however, such data are currently unavailable in Taiwan (including national statistics as mentioned in section 1.4).Although our insight is limited about dietary norms and practices in the previous Taiwan, a growing body of studies about meal skipping, solo eating, the prevalence of eating out (e.g., Chang, 2021;Huang et al., 2017;Yang et al., 2006;Yen et al., 2021) imply some rupture from the previously structured ones. 8We have confirmed this nationwide phenomenon based on the Taiwan Provincial Government's Family Income and Expenditure Survey during the 1980-90s, which covered both rural and urban areas (excluding directlycontrolled municipalities) and socio-professional categories (including agricultural households).
H. Ueda and Y.-C. Chiu industrialisation led to a shift from agricultural work to office work that demanded the outsourcing of meals.Globalisation also increased the instability of the national economy that supplied, under economic recessions, a cheap labour force for restaurant industries with low entry barriers.On the consumption side, the downsizing of households and women's participation in the workforce increased the cost of household chores and thus accelerated the outsourcing of family meals (Wang, 2021).In Taiwan, demographic and family structural changes have taken place in a far more 'compressed' manner than in Japan and, therefore, the 'post-war family system', the nationwide family model premised on housewives' responsibility for family care work, was not so a large barrier to social restructuring (Ochiai, 2014).This modernisation process has an inevitable implication for the abovementioned destructured dietary norms regarding family meals.Contemporary Taiwanese (including women) are relatively liberated from the dietary norms about home cooking.It is suggestive that in current food education policy in Taiwan, careful attention is paid to this gendered nature of dietary norms and attempts are made to reconstruct eating well without relying much on family burdens.This constitutes a marked contrast to Japan, where the post-war family system has yet to be overcome and dominant food discourses are still premised on 'family responsibilities' (Ueda, 2022a(Ueda, , 2023d)).
On the other hand, it is also inappropriate to leave this destructuration of the eating model unaddressed.To link our findings with previous literature, the prevalence of eating out, habitual meal skipping and the simplification of meal content are possible causes of lifestyle-related diseases and malnutrition (Pan, 2022).The increased distance between farms and tables was the prerequisite condition for consumers' growing food anxiety (Chiu & Yu, 2019) and increased demand for 'quality' information.The consumption of flour-based staple foods, which has become popular since the post-war immigration of mainland Chinese to Taiwan (Chen, 2010(Chen, , pp. 1895(Chen, -2008)), now occupies a relatively important space in daily eating compared to rice-based ones.This dietary change has had difficult implications for restructuring the country's national agriculture, in which rice has historically been the major product.In contemporary Taiwan, it has become an urgent matter to reconstruct dietary norms to reconcile them with these societal issues.
In this respect, our findings point to the still important roles played by certain social institutions to ensure 'intimacy' (Giddens, 1991) at the table.Even within the overall destructuration of eating models, dinner is still a relatively important source of intimacy that many people value and could practice, starting at regular hours and spending time slowly eating together with others.Regarding food environments, traditional markets and professional retailers serve as key sources of fresh and affordable food products.These institutions also make a face-to-face commitment to building and reconfirming consumers' trust in food systems (Giddens, 1990, p. 83), a role that cannot be played by 'faceless' supermarkets.The failure to do so will lead to an increasing demand for quality and information display (ibid: p.33).This is one of the reasons for which various criteria of 'good' foodssuch as 'producers are visible' (the symbol of traditional markets) and, at the same time, 'certified by trustable institutions' (the contemporary invention) -coexists in the minds of Taiwanese eaters.Moreover, our findings revealed sociodemographic inequalities in food capability.In particular, the young generation had limited access to these intimate institutions, compared to the middle class and the elderly.Prior to making a judgement about whether to normalise the current eating model for the next generation, it is important to first ensure appropriate educational opportunities for the young to think reflexively about the (de)valuing of this eating model.

Global implications
How can we extract global implications from this national context in Taiwan?Ueda (2022a) previously reported a large commonality in the content of eating well between Western countries and Japan.Our study has also confirmed the similarity between these countries and Taiwan, despite some differences in the level of destructuration (see Table 9).These findings support the universal thesis of reflexive food modernity and open up the possibility for global dialogues on dietary well-being.
This does not mean that reflexive food modernity is a monotonous trend, but rather invites further analysis of regional differences in its logic, processes and consequences (Beck & Grande, 2010;Chang, 2016;Poulain, 2019).Overall, there were two interesting differences between Taiwan and Japan in terms of the content of eating well.To emphasise it again, the first feature in Taiwan was the small norm-practice gaps, that is, the relative absence of psychological conflict between dietary ideals and the destructuring realities, which was particularly problematic among the Japanese (Ueda, 2022a).In relation to this, the second difference was that gender-based inequalities were not observed in Taiwan (with the exception of meal skipping), as opposed to Japan, in which men were more deprived of capabilities than women in many dimensions of eating (Ueda, 2022a).

Table 9
Illustration of eating well in contemporary Taiwan.This comparative insight from Taiwan relativises Japanese conditions and highlights the historical fact that the current gender-based inequalities in dietary well-being and the norm-practice dilemmas in Japan were constructed due to its post-war family system and the delay of its social restructuringa consequence of 'semi-compressed' modernity (Ochiai, 2014(Ochiai, , 2019)).Contrary to the popular (mostly Western) view that Asia is one region, there exists an important difference between compressed food modernity (in our case, Taiwan) and semi-compressed food modernity (Japan), which provides a convincing case for the recent academic scholarship on the plurality of food modernisation (Figuié et al., 2019;Khusun et al., 2023;Louis & Poulain, 2018;Mognard et al., 2023;Poulain, 2012Poulain, , 2019;;Poulain et al., 2014).

Limitations and prospects
The first limitation concerns the sampling in our survey.While being a mostly national representative sample, our respondents tended to have higher SES (education and profession) than the national average.Given the recent studies on health inequalities in Taiwan (Lee & Jones, 2007;Lo et al., 2012), it is probable that the dietary standards identified in this survey might have been slightly higher than the actual standards and that, if this is the case, the actual destructuration of the eating model in Taiwan might have been more advanced than we assumed.
The second limitation relates to the reconcilability of our results with national statistics.Although our results were mostly consistent with national statistics, there were some disagreements, notably on the prevalence of meal skipping and eating out, and attitudes to price.There were numerous differences in their survey modalities (e.g., web-based questionnaire/face-to-face interviews, norm/practice distinction, question wording); thus, all we can do for now is to respect both results and use them flexibly as food policy resources depending on their purpose.
Another limitation concerns the comprehensibility of the results for practitioners and the public.We set 10 evaluative dimensions and gathered both norm-and practice-related information.This comprehensive data is useful for the best possible understanding of one's dietary well-being, but this methodological choice also poses new questions regarding the complex nature of well-being: how can we make a shift from our profile-based indication to develop an integrated indicator of dietary well-being?How can we reconcile two different sets of data about the subjective and objective aspects of well-being in such a measurement?Having clarified dietary norms and practices in two countries, how can we judge which one is in a better condition?Further studies and social debates are needed to address these questions, which have been rarely discussed among Taiwanese food scholars and practitioners.
Lastly, we have so far followed Ochiai's characterisation of Japan as semi-compressed modernity in terms of demographic transition.Now that the dietary difference between Taiwan and Japan was revealed, it is plausible to say that the larger norm-practice gap and gender inequality in food capabilities are two of the effective indicators of semicompressed food modernity.Further works are needed for its full characterisation.

Conclusion
In contemporary societies with diverse values attached to eating that often conflict with each other, it is important to scrutinise what eating well means to a given population.While such endeavours have been pioneered mostly in Western countries, Asia has been rarely explored.It was positioned merely as a useful case to highlight the features that contrast with the Western equivalents.With the ambition to narrow this knowledge gap, particularly on the commonalities and differences within Asia, we conducted a web-based questionnaire among Taiwanese and elucidated their dietary norms and practices for the total dimensions of eating models; that is, not only meal content, but also the temporal, spatial, social, qualitative and affective aspects.Similarly to the other parts of the world, the so-called 'destructuration' of eating models, such as the irregularisation of mealtimes, the individualisation of eating and outsourcing of family meals, characterised Taiwanese eaters.
However, we have also pointed out that the destructuration does not necessarily mean a decrease in dietary well-being (i.e., norm-practice gaps), a condition that was the consequence of the country's 'compressed' food modernisation.However, on the other hand, the destructuration is also linked with imminent nutrition and food system problems.It is a difficult but inevitable task to reconstruct the nation's dietary norms in light of the current dominant norm and its tension with social conditions.The weakened norm-practice discrepancies in the Taiwanese eating model can also be viewed as the 'restructuration' of dietary norms already in progress.Proactive food policies are needed to assist such a reflexive construction of Taiwanese eating model, in particular, through the enrichment of national statistics and relevant database, organisation of social debate and food education.As a country increasingly facing these challenges, Taiwan presents an important lesson for other countries.To advance our thinking about the complex relationships between dietary well-being and modernisation, further work needs to be done in other cultural contexts, particularly in other underrepresented Asian regions.
Note 2: B: breakfast, L: lunch, D: dinner.Note 3: The McNemar test was used to test the norm-practice gap (only bolded items for simplicity).ap < 0.05.

Table 1
National statistics relevant to the eating model in Taiwan.

Table 4
Place of eating.

Table 5
Timing and duration of meals and persons to eat with.

Table 6
Procurement, quality and pleasure of eating.
H.Ueda and Y.-C.Chiu

Table 7
Meal content and pattern.

Table 8
Multiple regression analysis of sociodemographic effects.