Vous êtes sur la nouvelle plateforme d’Érudit. Bonne visite! Retour à l’ancien site

Speaking Out: Research & Essays on Speaking in the Food Voice

“Like, Pasta, Pizza and Stuff” – New Trends in Online Food Discourse

  • Stefan Diemer,
  • Marie-Louise Brunner et
  • Selina Schmidt

Corps de l’article

Food Discourse Moves Online - From Recipe to Blog to Skype

Food is strongly connected to culture and lifestyle and it has been a central topic in human interaction over time, Fisher (2001: 13) [1] even states that changes in recipes are “inextricably tangled with man’s history and assumed progress.” People do not only write about food in recipes, letters, and travel accounts, but also talk over food at gatherings and family reunions. This focus on food as a discourse topic influences our language use as Lakoff (2006: 165) [2] observes, stating that “cuisine has in many ways affected our language, both our vocabularies and the way we construct discourse around food.” As online media and communication channels are increasingly used on a daily basis, it is not surprising that this popular conversation topic has also become established in an online environment. Some types of food discourse, such as recipes, remain rather stable in the online context, while others have evolved and adapted.

Traditionally, much research in the area of food discourse has focused on the structure and language of recipes (for example, Norrick 1983;  [3] Cotter 1997;  [4] Bex 1996;  [5] Riley-Köhn 1999;  [6] Wharton 2010 [7]). Many researchers have investigated the contextual and sociolinguistic variation that recipes are subject to. Bex (1996: 169)  [8] and Wharton (2010: 68) [9] state that a specific audience is reflected in genre variation, as recipes are adapted to their intended readership. Lakoff (2006: 160/164) [10] asserts that recipe writers “construct” their readers by selecting a particular set of features adapted to them. Cotter (1997: 54) [11] similarly stresses the potential for an unlimited range of sociolinguistic variables in recipes.

There is also a lot of variation within the genre of recipes that can be ascribed to authors’ individual style choices and personal preferences (cf. also Lin et al. 2008 [12] on style variation in cooking recipes). Nevertheless, there are some features that seem to occur in most modern recipes and can thus be considered “typical.” They have been clearly defined by many researchers, for example Norrick (1983), [13] Cotter (1997), [14] Wharton (2010), [15] and Diemer (2013): [16] Recipes usually follow a bipartite structure, starting with a list of ingredients, followed by step-by-step instructions on preparing the food in (mostly) chronological order. Other recipe components like indications of required tools or additional advice are facultative and thus subject to further variation (Riley-Köhn 1999: 195). [17] The syntax of recipes is rather restricted and marked by an ellipsis of subject and article and a prevalence of verb-initial imperatives.

In addition, presupposed cooking skills and knowledge of cooking-related language are an intrinsic aspect of recipes. These presuppositions complicate understanding for outsiders who do not belong to the subculture of cooking cognoscenti and consequently could run into problems when encountering genre-specific instructions, semantics, and technical vocabulary (Norrick 1983: 180; [18] Cotter 1997: 58 [19]).

The content of recipes is thus often characterized by complex instructions as well as an increasingly aesthetic (Hess 1996) [20] and evaluative content such as personal opinions and lifestyle comments (Diemer 2013). [21]

The transition from traditional print to online food discourse has been comparatively seamless. With the development of both computer hardware (for example faster connectivity or more powerful processors) and software (for instance platforms and applications allowing easy media integration, increased content accessibility and content sharing), and the evolution of the internet and the role it plays in globalized societies, there has been an increase of multimodal elements in online food discourse. We conceive of multimodality as a “mix of modes” which, although present in almost all communication, has seen a marked increase in the last decades (Renkema 2004: 76). [22] Kress (2011: 46) [23] regards computer-mediated linguistic modes, such as images, video instructions, and links to other sites, as “partial means of making meaning.” Consequently, online recipes differ from print recipes mainly in the increased integration of multimodal features, as can be seen in the BBC Food Recipes section (2014), [24] illustrated in Figure 1, with the inclusion of videos, pictures, links and sharing functions.

Figure 1

BBC Food Recipes: Title Page and Easy Chocolate Cake [25]

BBC Food Recipes: Title Page and Easy Chocolate Cake25

-> Voir la liste des figures

For this paper, we selected two examples of online food discourse that move beyond the domain of recipes and into lifestyle and cultural contexts: food blogs, as an example for written computer-mediated communication, and Skype conversations about food, as oral counterpart. The paper aims to give a broader understanding of food discourse (other than recipes) in an online environment. With our analysis of food blogs we illustrate how the medium influences established parameters and long-term trends observed in print recipes while our analysis of Skype conversations about food indicates that oral online food discourse leaves the restraints of the recipe genre completely, focussing instead on cultural and lifestyle issues rather than the specific procedures of meal preparation.

Food blogs are written, asynchronous online forums dedicated to the preparation and evaluation of food in a lifestyle context. Asynchronous in this context means that the communication does not take place simultaneously but consecutively (Herring 2001), [26] so that reactions to blog posts occur time-delayed. As Diemer and Frobenius (2013) [27] remark, food blogs unite elements of printed recipes, diary-like texts, and web-specific features, such as hyperlinks (e.g. to other blogs), search windows, tags, online comments and advertising banners, as illustrated in Figure 2 showing the main page of Pastry Affair.

Figure 2

Pastry Affair: Mango Margarita [28]

Pastry Affair: Mango Margarita28

-> Voir la liste des figures

Skype conversations from the Corpus of Academic Spoken English (CASE)  [29] about food fall into a different category. In an oral, synchronous (i.e. simultaneous), dialogic computer-mediated communication setting, participants talk about a wide variety of topics ranging from food over culture to education. Figure 3 shows a typical configuration.

Figure 3

Skype Conversation in CASE [30]

Skype Conversation in CASE30

-> Voir la liste des figures

The Data: FBC and CASE

In order to analyse the language of food blogs, a corpus of blog texts related to food was compiled at Saarland University, Saarbrücken, Germany, in August 2011: the Food Blog Corpus (FBC, Diemer and Frobenius 2011 [31]). It consists of the 10 most recent consecutive blog entries from the first 10 blogs listed in The Times Online Ranking (100 blog posts in total) and comprises 826,073 words. All blogs were written in English, although not necessarily by native speakers. Two versions of the corpus were compiled: a “raw” version preserving metadata and multimodal features such as layout, images, references, and navigation elements, and a part-of-speech tagged text-only version (POS-tagging in CLAWS, Rayson et al. 2014), [32] which allows the analysis of grammatical units. The corpus contains 207,938 tokens and 8,039 types. The standardized type-token ratio (TTR), an established measure of lexical variation, is 69.53. Compared with a standard, non-specific text type, this is quite high, indicating considerable lexical variation and diversity, as well as a specialized register. The corpus analysis was performed with WordSmith. [33] All tokens with a share higher than 0.01 percent, which means a minimum occurrence of 20, were analysed. Nouns and verbs were analysed separately and listed according to frequency. The items were then manually grouped according to meaning (ingredients, tools, methods, and measures) or function (for example qualifiers or modifiers), followed by a qualitative discussion.

CASE (Diemer et al. forthcoming), [34] the Corpus of Academic Spoken English, also compiled at Saarland University, Saarbrücken, Germany, provides the data for the analysis of food conversations via Skype. The corpus consists of 200 Skype conversations (more than 150 hours) between English as a Foreign Language (EFL) speakers from Germany, Bulgaria, Spain and Italy. More than three quarters of the talks include both video and audio (the rest only audio). Participants were given different topic prompts to facilitate a natural development of conversations, such as “University life,” “The role of online education,” and “Food.” An average conversation lasted between 30 and 60 minutes. CASE allows research in fields such as discourse analysis, computer-mediated communication, and multimodality. It provides detailed sociolinguistic background data on participants with nominative information removed. For the analysis of food talks, a subcorpus of all conversations concerning the topic “Food” (seven in total as of February 2014) was compiled from CASE data, including 14 German and Italian speakers, over 40,000 words, and 330 minutes of conversation. The subcorpus was then transcribed according to CASE transcription conventions, wordlists and concordances were compiled and analysed following the same procedure as established during the analysis of the Food Blog Corpus.

Food Discourse in Blogs

Food blogs in the Food Blog Corpus are characterised by discursive, interactional, and contextual features, a register mix, and lexical characteristics, all of which will be discussed below.

Blogs in general are a written, asynchronous genre of computer-mediated communication (see Herring et al. 2004, [35] Beisswenger & Storrer 2008 [36]), of which food blogs are a subgenre that follows the general structure and distinct layout of the genre. A typical food blog in the Food Blog Corpus is organized in a vertical format, with the name of the blog, often as part of a title image or banner, displayed at the top of the page (see Figure 2). A central window often contains the blogger’s food-related text, which may be combined with pictures or hyperlinks. The recipes are usually displayed below that. At the bottom of the page we often find tags (indexical items which facilitate the search for particular recipes) and comments by the readers. The structure of the comments section is open-ended, with the newest entries appearing at the top of the list and the older ones moving down with each new addition. Comments are sometimes addressed in the main text, which may prompt further comments, creating an asynchronous dialogue between blogger and readers, although the delay in responding may actually vary from a few seconds to much longer time spans. The remaining page may contain more pictures, advertisements, as well as links to archived posts and other food blogs. The multimodal character is in evidence through images, links, and occasionally video content. Figure 4, from Diemer and Frobenius (2013), [37] shows the typical layout of a food blog in the Food Blog Corpus:

Figure 4

Food blog structure in the Food Blog Corpus, [38] reproduced with permission from the authors

Food blog structure in the Food Blog Corpus,38 reproduced with permission from the authors

-> Voir la liste des figures


Food blogs in the Food Blog Corpus create regional and temporal context for the food presented and discussed through lexical means, often in an associative or contrastive manner. Not surprisingly, the most common regional references are French and Italian (as in examples (1) and (2)), the main influences on European cuisine. Other food-related regions are less represented in the corpus. There is also frequent mention of individual food-related contexts, with home, work, wedding, and party as the most common lexical items (example (3)). The temporal context is established through mention of seasons (spring,summer), time of day (morning,afternoon), and mealtimes (snack, breakfast, lunch, dinner), as illustrated in examples (2) and (3).

These regional and temporal references are frequently associated with cultural aspects such as clichés (example (1)) that may also be used contrastively, as in example (4).


The bloggers in the Food Blog Corpus describe a wide variety of food, and use regional food names and names in the original language, such as bienenstich, gelato, or tartines, as illustrated in example (5). Tools (see also example (9)) and ingredients are described in great detail, and bloggers often make use of verbs addressing and involving their readers such as tell, ask, or feel (example (6)). The use of local food names seems to prompt the usage of lexical items or even short phrases in the same language, resulting in code-switching, as in example (7).

As previously mentioned, Bex (1996), [39] Wharton (2010), [40] and Lakoff (2006) [41] state that recipes are adjusted to their intended readership, and that recipe authors “construct” their audience according to their expectations. As a consequence the perceived level of knowledge of recipe readers can vary considerably, resulting in a classification of non-expert, as identified by Lakoff (2006: 164) [42] as “someone who needs precise guidance in the kitchen in order to get the job done,” advanced, that is “someone who looks for exacting detail in order to achieve Mastery,” and expert, “someone who is already professional in all but literal truth, who simply needs a little advice from a professional colleague.” With regards to these different levels of expertise, presupposed knowledge of cooking skills and cooking-related language may complicate understanding for a non-expert audience (Norrick 1983: 180; [43] Cotter 1997: 58 [44]). A clear example of the reduction of presuppositions to cater for a less experienced audience are children’s recipes, as described by Brunner (2013). [45] Waxman (2004)  [46] even observes a general tendency for recipes to be adapted to a less and less expert audience, becoming “increasingly more specific,” “longer and more explicit.”

These perceived levels of cooking expertise can also be applied to the food blogs we examined based on a close analysis of how the readership is designed in terms of linguistic features that are used. In the Food Blog Corpus, the audience seems to be constructed as less or non-expert. The level of explicit detail, in terms of ingredient lists, specific amounts and measures, and instructions as well as the integration of multimodal elements, such as the step-by-step pictures illustrated in Figure 5, are all indicative of the non-expert audience.

Figure 5

Step-by-step pictures in Homemade Marshmallow Recipe [47]

Step-by-step pictures in Homemade Marshmallow Recipe47

-> Voir la liste des figures

In contrast to recipes designed for an expert audience, ingredient lists in the Food Blog Corpus do not only include key components for the meal described, but often also basic staples like water or salt (example (8)) which would usually not be explicitly mentioned in ingredient lists designed for an expert audience based on the presupposition that they would be aware of the fact that such basic staples are a common ingredient in this kind of recipe and thus not worth mentioning (Norrick 1983). [48] The tools needed to prepare the food are often described in great detail, as in example (9). Similar features have also been identified by Brunner (2013) [49] in children’s recipes.

Personal Opinion and Lifestyle

Personal opinions and comments on lifestyle are prominent features of food blogs, which marks a distinct difference to print recipes. Fisher (1983) [50] characterises recipes as “rarely funny, and almost never witty” but she already observes a trend to include more personal comments in modern recipes and an increase of what she calls “gastronomical chitchat,” [51] resulting in recipes being “larded with asides.” [52] A diachronic analysis (Diemer 2013) [53] of selected recipes, ranging from the Middle English recipe collection of the Curye on Inglisch (15th century) [54] over the 1631 edition of Gervase Markham’s English Housewife[55] the British recipe collections of Hannah Glasse (1747) [56] and Isabella Beeton (1861), [57] and the American cookbooks by Amelia Simmons (1796) [58] and Braiden Rex-Johnson (1992), [59] shows that recipes did not as a rule contain opinions and lifestyle advice until the 19th century. It is only in the examples from the 20th century that these occur in the recipes themselves. This trend has also been commented on by Waxman (2004: 100), [60] who expects future recipes to contain even “more in the way of inspiration and guidance and less in the way of highly specific prescriptions.”

This development can also be observed in in the Food Blog Corpus where these personal comments seem to be essential for the character of the analysed food blogs, creating a less formal, more personal and interactive register. Food bloggers in the Food Blog Corpus frequently use verbs expressing opinion, like find and think, to place their statements in an evaluative context and to render the recipe more approachable, as in examples (10) and (11).

The evaluative context is also prevalent in food bloggers’ frequent use of adjectives that can be characterised as positively connoted, following Biber’s and Finegan’s (1989) [61] classification of lexical markers of positive and negative affect such as amazing, delightful, or incredible, and disgusting, annoying, or horrible. Examples for positively connoted markers like good, fresh, favourite, great, pretty, special, and beautiful are frequently used by food bloggers in our data, mostly referring to their own food (examples (12) and (13)). There is also frequent hyperbole, as in example (12). Negatively connoted terms are rather rare in the Food Blog Corpus.

In our data, there is a repeated blending of positively connoted and descriptive content, as well as a constant shift between evaluations of food and non-food issues, as in example (14), expressing personal views and a connection to individual lifestyle.

Quantifiers as a Means of Measuring

Food bloggers in the Food Blog Corpus often use adjectives and adverbs to quantify ingredients or measures. The general trend to include more and more precise measurements (for example observed by Waxman 2004 [62] and Fisher [63]) is also followed in the ingredient lists of the Food Blog Corpus. In contrast to that the quantifiers found in the description of preparation methods are less detailed, consisting of rather unspecific terms such as about, pretty, little, much, or large instead of exact measurements such as 3 oz or 100 ml. Example (15) uses about, example (17) good and big rather than precise measurements. In example (16) the use of pretty reduces the prescriptive strength of what is said to an extent that the instructions are less stringent, which suits the context of the food blog. It could be argued that the use of these less specific quantifiers at strategic places in the step-by-step instructions of the recipe might pre-empt complaints if the recipe does not turn out as suggested (see example (16)).


Another distinct lexical feature of food blogs in the Food Blog Corpus is hedging, that is the relativisation of a word or utterance (compare for instance Brown and Levinson 1987 [64]), which is realised both verbally and through indeterminate noun phrases in our data. Bloggers use verbal hedges such as try and could, adverbial phrases like kind of, just, almost, and indeterminate noun phrases like thing and stuff. The use of verbal hedges is often connected to advice based on bloggers’ own cooking experience (example (20)) and not set aside from the recipes themselves. In examples (18) and (19), the use of think, would, and could seems to phrase advice tentatively. Like the use of modifiers, this lessens the prescriptiveness of the instructions. Examples (18) and (20) show a similar use of almost and just.

Indeterminate noun phrases like thing and stuff are common in oral discourse, functioning mostly as hedges. As example (21) shows, they are also used in the Food Blog Corpus, seemingly to establish a more informal character, like many of the lexical features described above (for example personal opinions, less specific quantifiers, and hedging).

Food Blogs: Conclusion

In summary food blogs analysed in the Food Blog Corpus illustrate a distinct subgenre of blogs dedicated to food-related discourse addressed to a non-expert audience, including detailed recipes as well as non-food-related lifestyle content. The computer-mediated communication genre is evident through the integration of multimodal and interactive elements. The language of food blogs is characterised by spoken language features, such as hedges and unspecific quantifiers, the use of regional and temporal contextualisation, and evaluative elements.

Food blogs in the Food Blog Corpus often showcase accomplished cooks and their rather advanced, detailed recipes in an aesthetically pleasing, informal, and interactive environment that allows (asynchronous) interaction and recreation of the food presented. Food blogs thereby seem to involve their readers, thus appealing to a wide audience. However, this audience may actually be interested not so much in cooking, that is the recreation of the recipes presented in the blogs, but rather in the multimodal presentation of the hybrid, food- and lifestyle-related content. In this context, reading food blogs might actually be an indication that food blogs may not primarily constitute an instructional but rather an entertaining genre.

Food Discourse via Skype

Similar to food blogs, food discourse via Skype shows distinct discursive and interactional patterns, contextualisation features, and register characteristics, in this case influenced by the genre of synchronous oral computer-mediated conversation. An analysis of food related conversations from CASE is used to illustrate these features. German participants are identified by SB (= from Saarland University, Saarbrücken) and Italian participants by FL (= from Bologna University, Campus Forlì).

Discourse and Interaction

As far as discourse in CASE food talks is concerned, an analysis of topic development shows that food itself takes center stage, while there is less focus on preparation methods (in contrast to food blogs). However during the conversations, there are still some explanations and descriptions of specific or typical meals and recipes. This can be considered conversational recipe telling as illustrated by Norrick (2011), [65] although these recipe tellings are relatively rare, rather rudimentary, and superficial (see example (22)).

[Example 22: Video from CASE 07SB28FL36] [66]

One salient feature of interaction in CASE Skype conversations is the exchange and discussion of regional and national clichés in a humorous manner. As example (23) shows, there is a strong connection between talking about food and laughter. This favours the creation of rapport between conversation partners, as observed by Spencer-Oatey (2000). [67]


The conversation partners try to put food in a larger context and give their interlocutors an impression of their general food habits. They establish regional, temporal, and personal backgrounds for the food items discussed, focussing on contrasts, for example by comparing global (for example, Chinese, Mexican, American, …) with national (German vs. Italian) as well as regional (Bavarian) cuisines. Temporal references include mealtimes, holidays, and seasons (lunch, Christmas, summer), while personal context is created particularly through descriptions of family traditions (e.g. my grandma’s sauce). Example (24) illustrates regional, temporal, and personal context.


There is a connection between food and culture in the CASE food subcorpus, which, although not surprising in itself, emphasizes the major role that food plays in creating cultural clichés. Traditional and stereotypical food types can be found in terms like sausages, pasta, or currywurst, establishing German and Italian backgrounds, but other cultural contexts are also mentioned and discussed (empanadillas, moussaka). Cultural expectations are expressed by connecting food clichés to cultural clichés, like bratwurst and lederhosen, which tend to be associated with Bavarian culture (see example (25)). These cultural references are frequently realised by code-switching, particularly with dishes such as kaiserschmarrn or carbonara and with cultural contexts like Oktoberfest. It could be argued that these code-switches occur either because participants do not know the translation or because there is none, as illustrated in example (26), where the German participant SB49 uses the word knödel (German potato or bread dumplings) instead of an equivalent English term stating “I don't know if there is a word in English for that.” The use of code-switching in this context seems to prevent the stagnation of the conversational flow.

[Example 25: Video from CASE 07SB50FL34] [68]

Due to the orality of the discourse, the focus seems to lie on the description of dishes rather than on detailed explanations of preparation methods, resulting in the use of generic ingredients such as meat instead of pork or beef, as in example (27).

[Example 27: Video from CASE 07SB50FL34] [69]

Other unspecific terminology can be found in adverbial and nominal hedges such as actually and kind, as well as thing(s) and stuff which illustrate both the oral register and the non-expert background of participants (examples (28) and (29)).

Positively or negatively connoted terms occur frequently in descriptions of finished meals and food types, although there seems to be a tendency to prefer positively connoted terms. Participants qualify food as super-(un)healthy, great, or good and use verbs such as love, like, and enjoy (see example (30)).

The combination of different cultural clichés can lead to cultural misconceptions, for instance when the German concept of Spaghetti ice cream meets the Italian concepts of gelato and pasta, which results in the reaction in example (31).

[Example 31: Video from CASE 07SB48FL32] [70]

CASE Food Discourse: Conclusion

In contrast to food blogs from the Food Blog Corpus, food discourse in CASE Skype conversations does usually not contain recipes so that the focus lies more on personal opinions and lifestyle descriptions in a general food-related context. As recipes are rather rare in CASE, the unspecific quantifiers and hedges concerning measurements documented in the Food Blog Corpus do not play a major role in CASE data, although hedges in general are frequently used as part of the oral register. Food discourse in Skype conversations is further characterized by non-expert language features, such as vagueness in content and lexis. It establishes regional, temporal, and personal background, and makes extensive use of evaluative language. Cultural clichés and stereotypes, frequently realised through code-switching, can often be found in combination with non-seriousness and laughter, creating rapport between participants.

Conclusion: Online Food Discourse

In this paper we analysed two examples of online food discourse that move beyond the domain of recipes to show how the medium changes parameters established in print recipes. We used the Food Blog Corpus to illustrate written and Skype talks from CASE to illustrate oral computer-mediated discourse about food.

Both make use of the multimodal environment established by new technical developments. The two online discourse types mix evaluative and descriptive content as they shift between special-purpose (i.e. food) and lifestyle, as well as general content. While the food blogs put recipes into a larger lifestyle context flavoured with personal comments and opinions, CASE food talks mostly focus on general lifestyle content and personal views on food and culture, sometimes illustrating them with the help of superficial and rudimentary descriptions of food preparations and recipes.

Food blogs in the Food Blog Corpus seem to be constructed for a non-expert audience, which explains the tendency towards a decrease of presuppositions, realised for example through the use of detailed ingredient and tool lists, including basic staples, as well as detailed step-by-step instructions, illustrated by pictures. Where technical terms occur in food blogs, there seems to be a need to explain them to a non-expert audience. Non-expert language features, such as hedges and vagueness in content and lexis, also occur in CASE food talks, where they are mostly related to the oral register in general and less to the recipe genre. Technical terms appear to be virtually absent in CASE. Another feature specific to these conversations is the difficulty of transferring food terminology into English, which contributes to an increased use of unspecific food terms, code-switching, and vague descriptions.

A common characteristic of both discourse types is the strong reference to regional, temporal, and personal contexts. In the Food Blog Corpus, this may serve to involve the audience and increase the recipe’s appeal. In the international context of CASE, intercultural perspectives are integrated through cultural expectations and stereotypes. These extend to different food cultures, which are still prevalent despite the influence of globalisation. In this context food may serve as a medium to bridge the gap between different (food) cultures by creating rapport between individuals and thereby understanding of each other’s culture.

It can be argued that the continuing trend towards a non-expert context observed in this paper, not only in the recipe genre but also in a multimodal online food-related environment, is accompanied by a stronger focus on culture and lifestyle issues as well as personal views, and may reflect a general tendency towards a less specialised, more involved, and entertaining food experience.

Parties annexes

Parties annexes