Variations in Indonesian cooking instructions: A register analysis of spoken and written online user-generated content

. Cooking instructions, both written and spoken, are some of the most recognizable types of text. In this paper, the authors outline the characteristics of spoken and written cooking instructions as well as the differences between the two, focusing specifically on online user-generated content, which has proliferated in the last two decades. The study applies register analysis to spoken and written sample texts from TikTok and Cookpad. The study finds that the two types of texts share most of the situational characteristics but differ in the occurrences of certain linguistic features such as verb forms and pronouns. This preliminary study may serve as a foundation for further research on the topic of online language registers, especially on written and spoken cooking instructions.


Introduction
The intersection between culinary and linguistics is often overlooked by scholars specializing in either discipline.Especially regarding variations in cooking instructional text, one of the most distinctive types of text across languages.Previous studies have focused on examining and/or describing written recipes, whether in terms of its lexico-grammatical features, stylistic choices, or diachronic development [1]- [3].Yet little attention has been given to the spoken counterpart of written recipes and the distinguishing features between these two types of text, though one study examined the two types of texts through the lens of systemic functional grammar [4].
The aim of this study is twofold.First, to describe the situational, linguistic, and functional characteristics of online spoken and written Indonesian cooking instructions.Second, to outline the similarities and the differences between the two registers.For this study, a sample of user-generated texts was collected from the video-sharing app TikTok and the recipe-sharing site Cookpad.Online user-generated content was chosen because it is easily accessible and readily available in large quantity.The study employs a comparative register analysis with an emphasis on the variations in lexicogrammatical features, focusing specifically on the choice of verb forms and their co-occurrence with personal pronouns.

Register and register analysis
A register is a variation of language defined by its context-whether it is situational, occupational, or topical [5].It can be contrasted with dialectal and sociolinguistic variations in terms of geography, gender, class, and other social groupings.Biber & Conrad [6] consider register as one of the three perspectives one can use in analysing variations in texts, alongside genre and style.
This study refers to Biber & Conrad's [6] framework of register analysis.A register analysis seeks to describe a language variety in a text by elaborating the situational characteristics in which it is produced, its linguistics features, and how the two are related functionally.A good register analysis should be comparative; one can only describe something by describing what it is not.Besides, a register is defined not by the mere presence of a certain lexico-grammatical feature, but by the feature's prevalence in the register compared to that of other registers as well as their co-occurrence with other salient features.
While register, genre, and style are related to each other and sometimes used interchangeably, register analysis can be contrasted with a style analysis, which focuses on investigating aesthetic choices in text creation, or genre analysis, which focuses on the structural conventions of complete texts.Register analysis, unlike genre analysis, can be applied to excerpts.This flexibility makes it well-suited for analysing a wide range textual material, whether it is written or spoken.

Written and spoken cooking instructions
Written cooking recipes, which include instructions on how to prepare a dish, has long been recognized both as a genre and a specialized register cross-linguistically [2].In their discussion of online registers, Biber & Egbert [7] categorize recipe as a particularly distinctive "how-to" text, subdued below the more general register of "informational descriptions, explanations and procedures".As a genre, the structure of a written recipe is highly conventionalized, usually divided into two main parts: ingredients and instructions.As a register, written English recipe often has characteristics that are less pronounced in other registers of the language, such as the frequent use of imperative verb forms and the repetitive use of certain conjunctions, which limits the range of complex sentences that can be produced [1].
Literature on spoken cooking instruction is scarce, let alone discussion of it as a register (or a subregister), even though there are certainly distinctive register-like qualities in spoken texts containing cooking instructions.These texts range from the highly scripted TV cooking shows to the more interactive and spontaneous cooking class conversations.More recently, an explosion of usergenerated content on the internet has also provided researchers with an abundance of data that can be tapped for register analysis.

Verb forms and personal pronouns in Indonesian cooking instructions
As explained in the previous subsection, the prevalence of certain linguistic features makes it possible for us to identify recipes as a distinct register in English.However, the list of characteristic features of recipes may differ between languages.For example, while direct imperative verb forms are especially common in written English recipes, languages like Japanese and Korean (and, until recently, Indonesian) often use indirect passive verb forms in the instructional sentences [2].
'Cut the star fruits, wash the carrots clean, (and) cut them, and put everything into the juicer, so that we can get the juice for one glass.' The usage of passive verbs for instructions is also not unusual in spoken Indonesian (both Standard and Colloquial), as can be seen in the following example [8]: (2) Di-angkat, jangan di-seret!PASS-lift don't PASS-drag 'Lift it, don't drag it!'Shiohara [2] finds that despite the ubiquity of passive forms in old written Indonesian recipes from the 60s and 70s, there has been a noticeable shift toward using only bare active-imperatives in recent cookbooks, perhaps to imitate the conventions found in English-language recipes.However, this might not be the case for the spoken register, since many people seem to retain the use of passive verbs when giving direct cooking instructions to people in real-life conversations.
Another interesting aspect to be investigated is the usage of pronouns.According to Klenová [1], the occurrence of personal pronouns in cooking instructions humanizes the addressor and gives the addressee (in the case of cookbooks, the readers) the feeling of being interacted with.This is especially true for spoken instructions, in which the addressor produces the text as if they are talking to an unspecified addressee [4].Shiohara, for one, notes that in Indonesian TV cooking shows, the instructor often precedes active-declaratives with kita, the first-person plural inclusive pronoun, even though none of the audience directly interact with them [2]: (3) Kita masukin bawang bombe dulu 1pl.incl put.in onion first 'We put the onion first' or 'Let's put the onion first'.

Data source and methods of data collection
The data analysed in this study comprises two kinds of texts: spoken and written.To make a meaningful comparison, all the texts analysed in this study present the same recipe (with minor variations) of telur ceplok kecap, Indonesian-style sunny-side up with sweet and spicy soy sauce.
For the spoken text, the authors took a sample of 3 TikTok cooking instructional videos, 1 each from 3 different Indonesian content creators, namely Michele Alexander (@cheekykiddo, Text A), Bara Ilham Bakti Perkasa or more popularly known as Tanboy Kun (@tanboykun_asli, Text B), and Ibra Isman (@2beha10ribu, Text C).All three of them are known for their popular food-related content on TikTok.As for the written text, the sample includes the top 3 user-submitted posts in the Cookpad search results for "telur ceplok kecap" as of October 20th, 2022 (labelled Texts D, E, and F, respectively).Data source for the spoken and written texts are presented in Table 1.Since this study focuses on differences in linguistic features of instructions, only the instructional parts of the written texts are extracted for analysis, with the ingredients part being left out.
There are some limitations with the data used in this study.First, the sample is quite small; there are a total of 281 words from 3 spoken texts and 230 words from 3 written texts in the samples used for analysis.However, as noted by Biber & Conrad [6], reliable counts for common linguistic features (such as nouns, verbs, or pronouns) in register analysis can be done with few representative samples, including even 100-word excerpts.Secondly, the authors rely mostly on their own judgements as native speakers to decide whether the language of the texts selected is predominantly Indonesian.For the purpose of this study, "Indonesian" is defined broadly as a multidimensional spectrum between the prescriptive, formal Standard Indonesian, and the diverse colloquial varieties (including the widely used Jakartan Indonesian).Thus, the texts presented in this study may vary in terms of formality and politeness, as well as in the degree of code-mixing with regional dialects, English and/or other languages.

Methods of analysis
This study employs the framework of register analysis as devised by Biber & Conrad [6], which includes the qualitative discussion of situational characteristics of spoken and written cooking instructions, and quantitative comparison of lexico-grammatical features.
The situational characteristics of a register can be examined by describing seven elements: 1) participants, 2) relations among participants, 3) channel, 4) processing circumstances, 5) setting, 6) communicative purposes, and 7) topic.For the lexico-grammatical anaylsis, the authors decided to compare the occurrence of two main features, viz.verb forms and personal pronouns, in both registers.The instructional verb forms are further divided into three categories: 1) imperatives, which are bare verb stems without preceding subjects, 2) active declaratives, which are verbs preceded by subjects, and 3) passive declaratives, which are verbs with di-prefix.Meanwhile, the personal pronouns are categorized by their person (first or second) and number (singular or plural).
Since the size of sample is small, the quantification of the features was done manually without the assistance of computers.This is also to ensure the accuracy of the features categorization, as some forms found in the texts may not be readily recognized as belonging to the categories devised by the authors.

Situational characteristics of spoken and written cooking instructions
The situational characteristics of written and spoken texts examined in this study are summarized in the following Table2.First element, the participants.All the spoken and written texts examined here have a similar set of participants: a single addressor, plural and unenumerated addressees, as well as a lack of onlookers-defined by Biber & Conrad [6] as "participants who observe but are not the direct addressees of the register".The instructional nature of cooking videos and written recipes leaves no room for onlookers; the moment anyone starts to watch or read the instructions, they become an addressee and not an onlooker.
Second, the relations among participants, measured in terms of interactiveness, relative status, personal connection, and shared knowledge.In terms of interactiveness, both TikTok and Cookpad feature reaction buttons and comment sections that can be used to respond to posts in real time.This contrasts with traditional cookbooks, in which the interaction between the addressors (in this case the writers) and the addressees (readers) are almost non-existent.The relative status between the addressors and the addressees (whether in the spoken or written registers) are assumed to be equal as fellow users of the internet.And while there may be those among the addressees who have personal connection to the addressors, the anonymity of the internet makes it impossible for all participants in the texts to have known each other.Lastly, in regard to shared knowledge, all the cooking instructions assume at the very least basic knowledge of cooking and its vocabulary.
Next is an analysis of the channel used.There are two main modes used: speech for the spoken register and writing for the written register.Texts in both registers are permanent (that is, pre-recorded), although the specific medium used differs.The spoken texts are recorded as videos while the written texts are typed down to become articles and posted on the internet.The processing circumstances involve the text production and comprehension processes.In terms of production, the spoken texts may have been scripted prior to the recording and went through editing to remove unwanted parts before being posted to the respective creator's TikTok account.On the other hand, Cookpad allows true editing (that is, changing the content of a text instead of just removing some part of it) after the initial posting of a recipe.This means that the text can continue being produced by the addressor even long after the first time it was uploaded.Meanwhile, in terms of comprehension, as both the spoken and written texts are pre-recorded, they can be skimmed through if the addressees only need the key parts.
Since the texts are pre-recorded before being posted on the internet, the setting (time and place of text production) is not shared between the addressor and the addressees.The texts are all posted on accounts that can be viewed by random internet users, so they can be considered to take form in the public domain.All the texts are also roughly contemporary, with the oldest being posted in April 2021 and the newest one being posted in January 2023.The communicative purposes of the texts analysed here include the general purpose to present a how-to or a set of procedural steps, more specifically on how to prepare a telur ceplok kecap.Since these texts give instructions for a real-life dish, we can also consider it to be factual and not speculative nor imaginative.Finally, texts in both registers evoke the general topic of daily activities, with the more specific topic of cooking.
As we can see, there are a lot of similarities in situational characteristics of both the spoken and written registers of cooking instructions.However, there are also a few important distinguishing characteristics between the two, specifically in their channels (modes and media used) as well as processing circumstances.

Lexico-grammatical features: verb forms and pronouns
The following frequency of verb forms are observed in the texts, as seen in Table 3. Already from the subtotal rows of the two registers, we can notice the differences between the two.At first glance, Shiohara's [2] suggestion that the shift toward using imperatives is largely a written register phenomenon seems to be confirmed here.As we can see, all three written texts examined in this article exclusively use imperatives for instructions, while less than 50% of the instructional verbs in the spoken register use the imperative form.However, when we examine the data closely, we can see how imperatives remain the most popular choice of verb form even in the spoken register.
The following examples show the usage of imperative verbs (bolded) both in the spoken and written sample texts: The passive form, which Shiohara presupposes to have remained the norm in the spoken conversational register, ranks last in this analysis.Even Text C, which uses the passive form more frequently than the other two spoken texts, still uses the imperative form more than the passive form.
On the other hand, there is a surprisingly large number of active-declarative sentences in two of the spoken texts (A & B).As mentioned in the previous section, activedeclarative sentences differ from imperative ones in that they add subjects before the verb stems.In the following examples, the addressors refer to themselves, or to themselves plus the addressees by using first-person pronouns (singular gue 'I' and plural kita 'we') before the active verbs (bolded): As mentioned by Shiohara [2], this seems to be a common strategy used by presenters in TV cooking shows.The adoption of this strategy by the addressors in Text A and Text B shows that they attempt to imitate the presentation style of formal, professional TV shows' hosts.Meanwhile, the author of Text C, which does not employ this strategy, seems to present himself more as a friend who just happen to know how to cook some recipes.This is supported by Text C's usage of informal words and affixes such as antep-in [stay-APPL] 'let it sit' (instead of formal diam-kan) and udin 'already' (formal sudah), as well as humorous diction, such as calling his burnt sunnyside up mata dajjal 'Anti-Christ's eye' instead of mata sapi 'cow's eye', which is a commonly used synonym for ceplok 'sunny-side up'.
As for the occurrence of personal pronouns, the result is presented in Table 4.The chart makes it clear that the frequency of personal pronouns referring to the participants directly correlates to the frequency of active-declarative verbs, which is expected, because imperatives have null subjects, and the passive-declaratives allow the agent of a clause to be dropped.In fact, no single pronoun is used as an argument to complete the clause in Text C, which lacks activedeclarative verbs.The only use of pronouns in this text is to indicate possession, e.g., sutil gua 'my spatula' and irisan kalian 'your chopped bits (of ingredients)'.
The most frequently chosen pronoun to co-occur with the verbs is kita.It seems that the addressors add this pronoun to declarative verbs in order to avoid saying direct imperatives, which, in a conversational context, may seem intrusive and face-threatening to the addressees.At the same time, using only passive verbs without subject pronouns in speaking may feel too impersonal.Thus, the usage of active-declarative forms along with the addition of pronouns also indicates an attempt from the addressors to personalize themselves and put themselves on the same level with the addressees.

Conclusion
In this paper, we have examined a small sample of usergenerated cooking instructional texts from two different registers: spoken and written.The similarities and differences between the sampled texts are outlined through a comparative register analysis, which includes the discussion of situational characteristics, lexicogrammatical features, and their functions.The starkest difference between the spoken and written texts examined can be observed in the variations of verb forms.The written texts overwhelmingly use active-imperative verbs, while the spoken texts use three distinct verb forms.Differences can also be found in the occurrence of personal pronouns referring to the participants.Creators using active-declarative sentences are found to be more likely to use pronouns than those relying on imperative and passive verbs.A brief discussion on the reasons behind the choices touches upon the issue of formality and politeness.It is suggested that creators who use active declarative sentences are trying to emulate the formal, professional presentation of TV cooking shows and seek to avoid being perceived as face-threatening by the addressees.
Preliminary findings from this study can be used to guide further research in online registers, especially regarding spoken and written cooking instructions.The study also provides new insights on the reasons behind the occurrences of these lexico-grammatical features in both registers.A more comprehensive, corpus-based study of online user-generated cooking instructions is needed to identify the register categorization more accurately, as well as to point out the exact situational factors that influences the occurrence of certain features.Involvement from native speakers through experiments and surveys may also reveal patterns that have not surfaced before.

Table 1 .
Data source for the spoken and written texts.

Table 2 .
Situational characteristics of spoken and written texts.

Table 3 .
Frequency of verb forms.

Table 4 .
Frequency of personal pronouns.