LINGUISTIC ERRORS OF INDONESIAN AIR TRANSPORTATION HOAX

The research was to find out linguistics’ indicators of errors found in the body content of false news discourse mingling around air transportation issues that appeared 2-years before the presidential election in 2019. The qualitative research used hoax discourse from the turnbackhoax.com collection. Errors found in discourse were later generated according to linguistics’ domain in Dulay et al., as well as additional spelling errors and surface strategy taxonomy. Toward false news (n=44) found in the data source, errors majorly emerge in newly ungrammatical word formation with omission process equitably distributed in all linguistics’ domains. Errors in phonology portray localization in several dialects in Indonesia, while errors in morphology and syntax appear for language economy purposes. Primarily affected, lexicon and semantics is the domain with the most variant of surface structure errors (n=4). This is relevant to the hoax spread’s intended purpose to deceive the audience by modifying words in originally true news to give different content meanings. Omission resulting in word shortening in the form of blending and clipping with 98% occurrences is popularly used by millennials with a medium level of media literacy to communicate in the digital society environment. To enhance relatedness in readers’ thoughts and feelings, explain errors that deliberately happened in pragmatics and spelling. Being familiar with these indicators can alert the public to question a post’s credibility, thus encouraging the urge for clarification.


INTRODUCTION
The mounting of social media content production delivered by multiplatform drives new challenges in news-sharing habits. Networked news stories accessible online spread faster and wider in coverage through social networking sites (SNSs) (Al-Rawi, 2017). The daily newsroom is available online nowadays to respond to customers' needs and interests regarding up-to-date information. This service, however, requires significant restructuring in technology, professionalism, and business (Ureta & Fernández, 2017). Planning, coverage, and reporting performed by the journalist continued with the process of revision, editing, and scrutinizing by the editor, considered as regular key processes of presenting news to the audience (Hakim, 2021) requires online skills readiness of technology and ethical online publishing (Njuguna, 2021). Clarification and verification of news sourcing from elite and non-elite actors maintain online content newsworthy (Van Leuven et al., 2018).
However, the detrimental effect of online news reporting in market-driven journalism enhances the social problem ramification of digital society in the form of online falsehood. Truckloads of media content spread online at hand are various in quality for being created by non-domain experts without proper verification or gatekeepers (Pal & Banerjee, 2019). Considering its spreading tools, false news is collected in turnbackhoax.id utilizes three main social media platforms, Twitter, YouTube, and Whatsapp, which act as excellent ways for someone to post and/ or consume news with no structure to decipher true from false. Taking advantage of the impetuosity of unreliable or unprofessional domains in sharing viral news, fabricated news spreads with deceptive motives to control public behavior (Azzimonti & Fernandes, 2022;Gunther, Beck, & Nisbet, 2019;Salam, 2018).
Mingling around social media platforms, false news appears in the form of discourse (with additional pictures/video, which is still considered as language in symbol). The fact that the media used to spread is considered low-reliable, and the tendency to find errors in discourse is significant. The error analysis (EA) framework investigates error patterns to identify and rectify difficulties in producing standardized sentences for publication.
The various finding appears from previous studies on identifying parts of the sentence that contains errors produced by non-native, second language, or foreign language speakers, such as in complex nominal groups or article and plural -s in English written manuscript among Japanese (McDowell & Liardét, 2020) and in lack of native standards thus it has a low chance to be accepted in top academic publication (McKinley & Rose, 2018). In the case of learners from a morphologically rich language such as Russian, English grammar correction in writing is considered challenging. Empirically, it is performed better by a minimally supervised machine learning classification approach supported by reliable annotators than a machine translation (MT) system operated using a corpus (Rozovskaya & Roth, 2019).
For some bilingual regions, Indonesian is considered as the second language after the tribal mother language, which has the potential for interference and errors. Interference mostly occurs among early childhood learners yet also among university students in level of vocabulary (Andini, 2018;Rofii & Hasibuan, 2019;Susanti, 2021;Wikanengsih & Rostikawati, 2019).
Being part of applied linguistics to systematically collect errors in grammar and lexicon produced by second/foreign language learners (Hinkel, 2018), the EA technique has never been employed to give an in-depth look at fake news discourse. Considering its spreading media, which is in informal social networking circumstances, errors in Indonesian hoax news can be deliberately implemented in the form of informal language. Thus, the research bridges EA in providing the practical and grounded foundation for language teaching and offering linguistics indicators which are useful yet simple ways to suspect hoaxes in informal news binged daily in SNSs among the Indonesian community.
Without proper editing, hoax news freely circulating social media contains errors that are intentional and misleading. The research tries to categorize the type of linguistics errors in hoax discourse based on Dulay's taxonomy in the domain of (1) phonology, (2) morphology and syntax, (3) lexicon and semantics, and (4) pragmatics. Those errors are later generated into the category of taxonomy in the surface structure, including (1) omission, (2) addition, (3) misinformation, and (4) misordering (Dulay, Burt, & Stephen, 1982). This is a qualitative study to give a descriptive analysis of the error phenomenon in hoax discourse collected from turnbackhoax.id (a site dedicated to fighting hoaxes in Indonesia). In this case, paying attention to errors found in the body of the story can provide education for the public on visible signs of hoax news.

METHODS
The qualitative research utilizes all discourse on air transport issues collected online from turnbackhoax.id, including title, picture/video, and news body from 2017-2019. Transportation, as one of the major subjects of hoaxes, is proven to be a reflection of real social problems in the Indonesian community (Pinem, 2021). Regardless of being part of the academic society of air transport, a population of one of the aerospace schools including students, staff, teachers, and instructors, has the potential to intentionally click the hoax's headline news to read it for various purposes (Pinem & Effendi, 2021).
The range of the data period is determined by considering the 2-year campaign phase before the presidential election in 2019. The whole collection of hoaxes on turnbackhoax.com is in Indonesian. Even though some of them are reposted from other countries, it comes with an Indonesian version; thus, error analysis is done by comparing it to standard Indonesian.
Using error analysis techniques toward the text, data is acquired from discourse. Further, errors are categorized and analyzed linguistically based on their indicators to be interpreted in the discussion. The researcher observes overall discourse news as an object spot some language error predictions of Dulay taxonomy in linguistics, including phonology, morphology and syntax, lexicon and semantics, and pragmatics and taxonomy in surface strategy, including omission, addition, misinformation, and misordering (Dulay, Burt, & Stephen, 1982).

RESULTS AND DISCUSSIONS
Overall, there are 44 files recorded on turnbackhoax.id collection from 2017 to 2019, categorized as hoaxes in the aviation industry. The subject of the news mentioning airlines, airports, aircraft, tickets, and immigration is directly related to the process of a flight. One indirectly related topic mentions PT Dirgantara Indonesia (PTDI) is one of the state-owned aircraft industries. A significant gap in numbers is found between news in 2017 and two newer years, 2018 and 2019, as seen in Figure 1.
Hoax news is usually shared individually, in informal and favorable circumstances. Thus, it possibly contains some errors in the content (text, picture, or video) for the sake of misleading/falsification or sensationalism purposes. The journalistic intention is carefully planned in those errors as the instrument to mislead its audience. In the research, generating linguistic devices of false news discourse will show the language choice of a specific social identity. This can be helpful in tracking back the origin of the news (Li & Su, 2020). Information in the news is formed in a text, whether written in printed or online media or read by an anchor in a broadcast. The influence of an anchor in delivering false news has been proven significant for respondents with different levels of authority and awareness of false news (Jost, Pünder, & Schulze-Lohoff, 2020). Considering this fact, knowing some characteristics of the anchor or author who spread the news is possible to detect its origin. Even though hoax spread in social media is mostly written, some habits in speaking, such as dialectal varieties, are likely to appear in informal writing to point out these characteristics. Table 1 shows the erroneous phonology in writing hoaxes.
Replacing a word in a written text with its speech repertoire is considered unstandardized. In sample (1) in Table 1, the Indonesian particle 'si', which is pronounced [si], is written 'ci' or 'c' (English letter) since they are similar in sound. In this case, foreignization occurs as a translation process of borrowing (Ulfah, 2019) to make the work foreign. Interestingly, characteristics of sociophonetics (Thomas, 2018) can also be interpreted by these errors, including omission (sample 2) of the weak vowel [ə] (schwa) sound (slamat, slesai, and negri), which is a type of Indonesian dialect typically influenced by the Javanese phonological process in Javanese speakers (Subiyanto, 2010). Another localizing characteristic is found in closer sound substitution as in pisik (Sample 3) as one of the most common Indonesian phonological errors (Tarigan & Sulistyaningsih, 1997), which is known as characteristic of Sundanese, and in contraction substitution of the diphthong [au] to vowel [o] as in ko (Sample 3a), which is popular in dialect among East Indonesian people.
In case of an error in sample 4 dikasi, twostep corrections are applied. Firstly, the word dikasi undergoes [h] omission in the final. However, the correct version of dikasih is considered a dialect. Thus, it must be changed to the exact word with the closest semantic meaning in standardized language. Through the diction choice, the final correct word appears in diijinkan (passive voice: being allowed). As the last indicator, elongated vowels are found to intensify adjectives (Sample 5), as in panjaaaang as products of Javanese transformational generative phonology (Kusuma & Anam, 2019), in adverb such as in akhirnyaaa (finally), in a particle such as in aamiin (amen), and extra consonants as in MassyaAllah (whatever God will) to emphasize the word.
Since it is usually spread informally using the chain message technique on social media, an error in a sentence is the most recognizable indicator of hoax news. Formal information should be written in the official language to pass review by an editorial board, which includes a grammar check. On the contrary, looking at the type of spreading media into consideration, hoax news spread is possibly limited by words such as in Twitter (n=280 characters). Therefore, information normally is written in shortened form or acronyms. However, these acronyms are not common in news item text, and in fact, many are not officially accepted. Table 2 shows the erroneous in morphology and syntax. Morphological errors of omission, addition, altering, misordering, and misinformation are commonly found among second or foreign language (ESL/EFL) learners (Indriani, 2019;Shin, Cortes, & Yoo, 2018). Being considered as a parameter of students' language mastery, grammar error is still problematic, for example, in English writing tasks for master's degree students (Hapsari, Ginting, & Bram, 2019), which is mainly influenced by the interference of students' first language, affection toward and existing knowledge of the language as well as teacher's role (Spahiu & Kryeziu, 2021;Zhang et al., 2021). However, in spreading hoax news online by a native of Indonesia, these errors in the form of word shortening (Sample 6) in Table 2 are made to keep computing characters less under word limitation, such as in Twitter. Letter omission results in shortening and omission, along with the usage of a number as a substitution in a word, are some of the indicators found in morphological errors. The number used as a substitution of the omitted word is to sign plural nouns, such as in (Sample 6b) detik2 and orang2 to substitute detik-detik and orang-orang or simply replace written text mentioning numbers, such in (Sample 6c) salah 1 (1 = satu = one) or (Sample 6d) 01 to refer Indonesian president which is called by his sign code RI-1/ Republik Indonesia 1. Surprisingly, shortening forms are readable since they are already widely known by the social community. Finally, the last errors relate to the sentence structure. Instead of omitting letter/s, the whole passage is missing a subject, such as in Sample 7. Missing a subject means an incorrect sentence. Besides missing the subject, the sentence tends to become too big, making it potential for errors in compound-complex sentence structure; thus, it requires conjunction. A big sentence lacks break marks, such as a comma or full stop to separate ideas. It is problematic to determine the topic from this type of sentence. Therefore, correction is done by rearranging the order of the sentence or splitting it into several sentences.
To purposely mislead the public perspective, hoax news deliberately plans errors in the meaning. Selecting a completely different lexicon with inconsistency in semantic features or simply omitting it drives the reader to the author's intended belief. So convincing, narration on hoaxes supported by pictures or video looks accurate and impossible to distinguish without rechecking other more reliable sources of information. In a different view, hoax news's existence reflects real social issues in the Indonesian community (Pinem, 2021). Thus, it is possible to mess up real news with fabricated ones by modifying its lexicon. Table 3 shows the erroneous in lexicon and semantics. In semantics, replacing diction (entirely or partly) drives to different references, especially when placed among a real news story. Misleading references in low-credibility posts are purposely done to deceive people and lead opinion. Falsely matching certain pictures from one story to another with different news references (Sample 8) is implemented toward the victim of the crashed aircraft of JT 610 and the shipwreck survivor of MV Lestari Maju (Sasmito, 2018a). Another replacement is done in the case of labeling a real written verse about an aircraft accident with a famous public figure (Sri Mulyani, an active Indonesian finance minister) as the author, which violates authorship and reference. The real author and the public figure are different people. Thus, replacing the proper noun is strongly questionable (Helsyanto, 2018). The next story involves a flight ticket mentioning one of the state's most wanted people (Sample 8b). The departure and arrival airport and passenger's names are easily connected to the hottest political news. Since information is later proven to be a made-up story, those names are just coincidentally similar, yet referring to a different person (Febrylian, 2018).
Referring a proper name to a well-established company (Sample 9) gives a sense of credit and reliability. However, creating subtle modifications by mixing it partly with an unreliable source can be mischievous. Potential crime lingers in a partially different email domain or address, especially since every character in the online domain is counted precisely. In short, different characters mean different references. When it comes to job vacancies in the aviation industry, Angkasa Pura 2, Angkasa Pura 1, and Garuda Indonesia are some names to look forward to. However, closely paying attention to every character of the email address or the official website is a priority. A state-owned company will always keep its online media official (Ramaputra, 2017).
Modification in the lexicon is also found in the word omission. With or without a particular word gives a completely different meaning or reference. Sample 10, Detik-detik pesawat gagal landing, is missing the word 'simulasi', which contrasts the real dangerous situation of failed landing attempts to a make-up situation that commonly occurs in training or simulation. In delivering information in the news, no single word must be omitted since it will change the meaning (Adi, 2019).
One of the characteristics of a language is dynamic in terms of vividness, which conceives of a dynamic form-meaning relation as proposed by Vygotskij (1987) in (Bertau, 2017). As long as sharing context takes place between the speaker and the listener, newly invented terms can surface. Even though officially, no reference can be found in The Great Dictionary of Indonesian Language, making these errors have no correct forms, online references can still track the meaning of Cinkolim (Sample 11) based on its historical context (Wirayuda, 2019) and Aseng which are sarcastically referred to a Chinese. Cinkolim occurs by blending the slang form of Cina-Kolonialis-Imperialis (Colonialist Imperialist China). This word was invented for political context by the anti-Soekarno press in 1965 after the tragedy of the 30-September Movement. Meanwhile, Aseng is a language variation form for a foreigner, 'asing' with a homonym of the common Chinese-given nickname of A-Seng (Khairil, 2017).
A different verb is different in meaning and references. Unless it is categorized in a group of synonyms, replacing a verb with another one with slightly or wholesome different semantic features will depict an altered context. The word 'mengganti' (Sample 12) means to replace the object mentioned in the sentence with a new one. The sentence 'Mengganti prasasti Bandara Lombok' relates to replacing the old stele of the airport name along with the official inauguration in the form of the previous president's signature with the new stele, which is misleading. Instead of replacing (destroying) it, a new stele with a newly assigned name is added as part of the process of substituting the airport name, which has been officially approved (Sasmito, 2018b).
Looking at every indicator mentioned in phonology, morphology and syntax, and lexicon and semantic, similarity in illocution is depicted as deceiving and directing readers to the author's planned purpose based on a news context. Implicit purpose in utterances can only be investigated through a pragmatic perspective, namely illocutionary acts. These errors shown in the next pragmatics discussion also appear in other linguistic domains. Table 4 shows the errors in pragmatics. Understanding errors from the pragmatics perspective are as if listening to sound from a transcribed utterance or comprehending the main idea of a passage. The readers of both forms can grasp implicit meaning from what is signaled by the errors.
Spelling errors shown in misplaced capital letters in the first letter of a non-proper noun such as Mewah, Aset, and Rezim (Sample 13), an overly written capital letter in the whole word such as PECAT and BEKERJA NYATA (Sample 13a), and bold written text such as (Sample 13b) 'Masihkah kita hanya duduk termenung' can be interpreted in pragmatic terms as the pointed topic or subject idea. In phonology, it is expected to be utter in a higher tone, while in syntax, it is considered the main/subject idea or functions as a theme in discourse.
Intensifying subject ideas are also pictured in exaggerated elongated vowels in a word, such as panjaaaaang (long) and yuhuuuuu akhirnyaaa Another category of error in pragmatics can also be seen in the foreignization of Sample 15, viralkan (to make viral), which derives from the word 'viral' in English with the additional suffix -kan to make it a compact command word. Different versions of this category are found in Sample 15a, bagikan..ke grup kalian (share to your group) and komen aamiin (leave a comment by typing amen). The purpose of this command is an invitation to forward the message to a larger audience by using media links. Collecting ratings for popularity is discreetly done for profitable business or political purposes.
Apart from errors in linguistics, some errors can also be found at the spelling level, considering that social networking sites are used in an informal situation. Indonesian spelling guidelines (1972), used to rule Indonesian standards in writing, are renewed to General Guidelines for Indonesian Spelling (PUEBI, 2015). It organizes the writing of letters, words, punctuation, and loan words. Thus, the errors are analyzed by comparing them to online PUEBI (Pedoman Umum Ejaan Bahasa Indonesia). Table 5 shows the erroneous in spelling. The first spelling error relates to the preposition of place, which is pictured by a space between a preposition and a word (name of a place). Since in Indonesian, the preposition of place -di is similar to the suffix -di used to transform a verb to a passive voice, unnecessary extra space appears in this passive voice verb. By giving extra space after -di in a passive voice, a sentence such as Sample 17 is misplaced with -di as a preposition of place. The emerging error suggests (Sample 17) as two different words (prep of place + word) because extra space indicates different words.
This extra unnecessary space between words also appears in Indonesian compound words with prefixes and suffixes (Sample 17a) and (Sample 17b).
Argued for an error occurring in the news, missing capital letters by social networking media users should not have happened considering the range of appropriate age to have a social media account. Capitalizing proper nouns such as in (Sample 18) susi (name of a person), garuda (name of a company), samsudinnoor (name of an airport), banjarmasin and palu (name of cities in Indonesia), or the initial of the first word in a sentence such as in (Sample 18a) seolah-olah and ini are basic writing skill which is taught in the earliest education. Another error in this category is improper or overly written capital letters, such as in Samples 18b and 18c. Other than the initial letter of the first word in a sentence, no root should be written in capital letters. Either forget or mistype, causing these missing capital letters results in spelling errors.
Bolding in Indonesian is used only to (1) highlight words with italics and (2) part section in writing (Pedoman Umum Ejaan Bahasa Indonesia). Writing a sentence in bold in Sample 19 has no matching to either highlighting an italic word or parting writing section. Therefore, this is also considered as a spelling error that can be explained pragmatically.
Well-known with English as the international aviation language, airports and airlines are familiar with English terminology for their operation. Incompatible foreignization for standardized loan words such as in Sample 20, recruitment, professional, Check, branded, out, and Fee occurs as a result of applying those words in Indonesian writing. Errors appear by keeping the word in the original or adding the prefix -di as a passive sign in digrounded (Sample 20a) to the original word grounded (Sasmito, 2018c).
This type of erroneous word is known as slang. Considering the object of this research which is segmented into hoax collection circulating air transportation subjects, the type of slang for these words is known as jargon. Regardless of being ungrammatical in various forms in Malay (Izazi & Tengku-Sepora, 2020), slang is considered understandable and accepted for communicating. Among millennials in Indonesia, communication using slang words on social networking sites is to address, initiate relaxed conversation, express impressions, and show intimacy (Rezeki & Sagala, 2019). With enough words to formulate a dataset, slang and formal words contribute to a pertained topic in a dyadic conversation, especially about food (Tho, Setiawan, & Chowanda, 2018). The type of linguistic errors in hoaxes emerges in a variant of word formation at the surface structure level. Repeatedly, (a) omission phenomena which appear in more than one linguistic domain (phonology, morphology and lexicon) are relevant to other error finding in EFL students' writing activity caused by  intra-lingual transfer (Aziz, Fitriani, & Amalina, 2020). Another category of errors that also occur repeatedly (2 domains) is (b) foreignization (4 domains), (c) elongation, (d) capitalization, and (e) bolding. It can be drawn from the findings that significant errors occur in the lexicon and semantics domain. This domain has five forms of surface structure variants categorized as omissions process by replacing the original word (proper noun partially or in whole or verb) with the new false one, deleting one keyword from the original statement, and blending to create a new term. In fact, intentionally deceiving public perspective can be fortified by modifying originally true news. Partial word modification shifts discourse function and meaning, which is common in false news. Hence, thorough investigation and cross-referencing to other reliable sources is the only way to clarify news to be true or false.
By looking at indicators derived from hoax discourse, some error categories can be classified into Dulay's taxonomy in surface strategy, including (1) omission, (2) addition, (3) misinformation, and (4) misordering. Besides these four major types, various appearances are found as a result of the modification process. Table 7 shows the variant of surface structure in linguistic erroneous category.
Through the modification process, word formation results in new forms in informal Indonesian. Most omissions found in hoax error categories are in the word formation process of shortening (clipping and blending). Blending and clipping are the types of shortening in surface structure with 98% of occurrences. Mentioned as extra-grammatical morphology forms, errors in taxonomy cognitive are found in the form of abbreviations, blends, clipping, and reduplications (Mattiello, 2013) and types of language variation of slang in colloquium, jargon, and argot, which are widely used and growing nowadays (Nikonova, 2020). Looking at online media as contexts and domains which some are limited by the number of letters, the principle of this acronym, blending, and clipping is for the linguistic economy (Zaim, 2015), which provides favorable short and simpler communication yet highly accessible marked choices to the community of speakers (Harared, 2018). It can be seen in Table 8.
Defined as the language for conversation, the process of making the word short in hoax depicts authors' speech in the form of written text. Unsurprisingly this phenomenon of utilizing social media is popular among millennials for media literacy with the level of critical understanding, which is considered a medium (Briandana & Dwityas, 2019). Regardless of its newly ungrammatical shortened word form, this social group is able to maintain communication skills in social network relationships.
Speech in writing pictures in casual circumstances thus aims to strengthen the bond between the author and the reader. The readers' feeling related to certain news frames is proven to shape their opinion and responses, thus developing bonding using language enhances the journalistic' planned scenario (Bastick, 2021;Panwar & Arora, 2019;Pinem, 2018;Tsfati et al., 2020).

CONCLUSIONS
Recommending ways to avoid and eliminate false news spreading is considered an effort to educate a strongly engaged yet risky community on social media platforms about hoaxes. Looking at hoaxes spreading media which is considered informal, linguistics errors are unavoidable. Error analysis technique as part of the language learning evaluation process is applicable to give a descriptive insight of factors attached to errors found in fake news. Considered ungrammatical, errors in phonology, morphology and syntax, lexicon and semantic, pragmatic as well as spelling emerge in new word formation and are considered as intended misleading instruments by the source. The word formation error category is dominated by lexicon and semantic domain with 5 (five) surface structure variants. This phenomenon is consistent with hoaxes' purpose of deceiving the public through partial modification of words and meaning in originally true news. Omission  pada, sudah, karena, lagi, jadi, tidak, dapat, selamat, sama, dari, di, ke, padahal kpd, pd, sdh, krn, lg, jd, tdk, dpt, slmt, sm, dr, d, k, pdhll b Utk, dtdtngni, spt, akhr ny, pdhll, gak, hgga, syk, jkt repeatedly surfaced results in word shortening through the process of blending and clipping with 98% occurrences. Online media as context derives this finding yet is considered favorable among millennials to gain relatedness among the readers.