The Effect of Latinization on Reading Time and Understanding: Greeklish in Communication and Social Media

The increasing prevalence of technology in society has an impact on young people’s language use and development. Greeklish is the writing of Greek texts using the Latin instead of the Greek alphabet, a practice known as Lati-nization, also employed for many non-latin alphabet languages. The primary aim of this research is to evaluate the effect of Greeklish on reading time. A sample of 732 young Greeks were asked about their habits when communicating through e-mail and social media with their friends and they then participated in an experiment in which they were asked to read and understand two short texts, one written in Greek and the other in Greeklish. The findings of the research show that nearly one third of the participants use Greeklish. The results of the experiment conducted reveal that understanding is not affected by the alphabet used but reading Greeklish is significantly more time consuming than reading Greek independently of the sex and the familiarity of the participants with Greeklish. The findings suggest that amending social and communication media with software utilities related to Latinization such as language identifiers and converters may reduce reading time and thus facilitate written communication among the users.


Social Networking
Greece's internet penetration rate stood at 82.2% of the total population [28] at the start of 2022 from 69.1% at the start of 2016 [29]. At the beginning of 2022, the number of social media users in Greece was 71.5 percent of the total population [28]. Koutsogiannis [30], who studied Greeklish in the context of a globalized environment, claimed that the Latin alphabet is used to a greater extent than the Greek, especially by children, who are familiar with digital environments and the English language.
The adoption of Greeklish in digital environments has attracted some attention as Koutsogiannis & Mitsikopoulou ([8], p.17) claimed "attitudes towards the use of Greeklish are deeply embedded in the Greek sociocultural context". However, research related to Greeklish is rather limited [30] and mainly focuses on software converters, transliteration practices and sociolinguistic issues [31] while research on the use and understanding of Greeklish is also quite limited. Androutsopoulos [32] referring to the users who employ Greeklish wondered "…but how many are they and exactly how they write? Nobody knows". To our knowledge phenomena like Greeklish, up to now have been approached from a linguistic point of view or from a philosophical perspective. The paper adopts a different approach and aims to address a number of questions related to their use by young people.
The aims of this research are to: • to measure the extent to which Greeklish is used by young people when communicating through internet and social media; • to evaluate the effect that Greeklish have on reading time of young Greek speakers; • to explore if understanding is affected by the alphabet (Greek vs Greeklish). This work studies the reading speed of computer users when reading Greek texts written in Greek and in Greeklish with the aim of understanding them. Text understanding is strongly related to reading speed [33]. The evaluation of the effect of Greeklish on reading time has both theoretical and practical implications: On the theoretical level, such an evaluation may 1) provide evidence that texts in Greeklish impose an extra conceptual load on the reader compared to texts in Greek and 2) quantify that conceptual load by estimating the increase in reading time; on the practical level, the results of the evaluation may suggest that software aids, such as language identifiers that identify Greeklish and language converters that automatically convert Greeklish to Greek, may be helpful for readers and their integration in social media may be beneficial for the users.
Although this research focuses on Greeklish, both its findings and approach may have a larger impact since Latinization (or Romanization) is not restricted to the Greek language. It is a practice used for many languages which use logograms or non-latin characters for reasons similar to those that gave birth to greeklish. As a result, Latinization is used for Chinese, Japanese, Russian, Arabic, Thai and many more languages. The common issues shared in all these cases include the existence of multiple systems for Latinization (despite the existence of an international transliteration standard), the development of transliteration DOI: 10.4236/sn.2023. 123005 70 Social Networking software and the societal concerns about the danger that such a practice represent for the traditional alphabets/logograms.
The paper is structured as follows: Section 2 discusses the emergence and evolution of Greeklish along with transliterating practices and related software.
In addition, studies on using, reading and understanding Greeklish are also presented. In Section 3, the methodology and the experiment that was conducted are presented, followed by the analysis of the results and discussion. Finally, conclusions, implications, limitations and possible future threads of research are discussed.

The Evolution of Greeklish
"Greeklish" or Latin-alphabet Greek is the "representation of the Greek language with the Latin script" [32]. Valassakis [34] and Moustaka et al. [35] argue that unholy, but also senseless, attempt to replace the Greek script in its own birthplace" and "a full-fledged attack against the classical Greek thinking" [41]. Professor of Linguistics Georgios Babiniotis, pointed out the risk of the "alienation" of Greeks from the image of the Greek words, due to the increasing use of Greeklish, and mentioned that "Greeklish is the best way of alienating from the image of the word. This is what young people will pay for it. We have Greek fonts and we can, using the Internet and the electronic media, use the Greek fonts that have the advantage of giving the image of the word, the visual idol, and to reconcile us with the spelling of the word and its meaning" [42]. On the other hand, Hatzisavvidis [43] as mentioned in [18] argues that "the Greek language is not going to be lost, because a language is lost when people who use it are lost".  [45]. They declared that they do not write in Greeklish and appears to project the identity of the "custodian" of the national spelling. On the other hand, the majority of students demonstrate that he is in tune with new writing practices in digital communication by stating that he uses Greeklish.
Koutsoutassiou [46] who investigated the graphemic policy of 106 Greek discussion forums in relation to Greeklish use, discovered, the vast majority of the forums have strict graphemic policies that forbid the use of Greek and implement stringent "preventive" measures to guarantee that all users will follow the rules without exception (use of a convertor that converts Latin characters to Greek ones, reprimanding and imposition of punishments, modification or deletion of messages in Greeklish without warning).

Transliterating Practices in Greeklish and Related Software
In linguistics, transliteration means "the system of conveying as nearly as possible by means of one set of letters or characters the pronunciation of the words in languages written and printed in a totally different script" ( [50]. As a result, it is not uncommon for Greek users to transliterate in their own [47] idiosyncratic way [7] depending on personal preferences taking into account the similarity between Greek and Latin letters' appearance, sound or keyboard layout [49]. Therefore, the use of Greeklish has adopted three main methods of transliteration: 1) Vocal: Based on sound resemblance, aims to represent phonetically the Greek text as accurately as possible while simplifying the historic Greek spelling, i.e., the Greek letter /θ/ yields /th/ and the diphthong /αι/ yields /e/.
2) Visual: Based on similarities between Greek and Latin letter shapes, using visually equivalent Latin characters or, in case of absence, numbers that optically resemble Greek letters. In contrast to the vocal trasliteration, it reproduces as much as possible the Greeklish spelling although it leads to some unorthodox solutions e.g. /8/ for the letter /θ/.
3) Locational: Based on the keyboard layout; it represents some Greek letters by the Latin letters that are placed on the same location on a qwerty keyboard.; it is similar to the visual transliteration but differs from it only in some letters e.g. the Latin letter /u/ is used for the Greek letter /θ/ and /c/ for the letter /ψ/ [7]  Taking into consideration all possible different types of Greeklish along with all their likely combinations, common Greek words can be transliterated into many alternative representations. As an example, Androutsopoulos [22] mentioned that the Greek word "διεύθυνση" ("address" in English) can be transliterated in twenty-three different Latin-alphabet versions.
Transliteration systems have been developed in an effort to automate the transliteration of Greek to/from Greeklish: deGreeklish [53], E-Chaos [54], Greeklish Converter v1.0 [55], Greek to Greeklish by Innoetics [56] are some of the transliteration systems developed. The majority of them are based on specific sets of rules that map directly each Greek character to a corresponding symbol of the Latin alphabet or use databases of Greek-Greeklish word pairs [19] [26] [47] [57] [58] [59] [60].
In 2012, Google started the Google Transliteration, an online service which converts Latin characters to phonetically equivalent characters to Greek and many other languages including Arabic, Hindi, Urdu, Punjabi, Persian, and other languages spoken in East Asia [57]. Androutsopoulos [32] investigated the usage and attitudes towards Greeklish in e-mail messages, using a questionnaire electronically distributed to Greek mail- considers it "ugly" but not a "threat" to the Greek language. Greek e-mail users seem to have a high degree of metalinguistic awareness regarding different transliteration conventions. Use of Greeklish poses more problems when used for reading rather than writing. Almost 50% of the responders agree that "reading Greeklish is a hard and tiring task". Laghos et al. [25] analyzed 1000 email messages, from 48 email accounts, exchanged between Greek/Cypriot academic and administrative staff at the public universities of Cyprus in 2011. They found that 1 out of 3 emails were written in Greeklish. The Greek language with Greek fonts is used in formal documents while Greeklish is highly used in informal e-discussions, mainly due to the speed and flexibility of using it. They concluded that using Greeklish appears to be a conscious decision suggested by convenience rather than technological constraints.

Using, Reading and Understanding Greeklish
Moustaka et al. [35] investigated the effect of Greeklish use on students' spelling. As a part of their study they recorded student habits regarding Greeklish use. A significant overall percentage of 77.4% use Greeklish. This percentage rises from 67.8% among middle school to 88.5% for high school students. It seems that students use Greeklish by habit and they feel they save time while a significant percentage mentioned that using Greeklish is a way to avoid spelling mistakes. Mouratoglou [18] investigated the Greeklish phenomenon from Greek language teachers' point of view. Increasing use of Greeklish was noticed as 49% of the responders reported that they use it while almost 34% did not answer the question. Extensive use of Greeklish was recorded in social media and mobile phones with the main reasons for using it being the speed of writing a message (68.60%) and time saving (52.89%). The majority of the sample (86.25%) considers Greeklish as a form of writing, mainly in informal forms of communication and recognizes its functionality as a tool for electronic communication (77.92%). A significant percentage of them (63.7%) stated that they use Greeklish on a daily basis.
Karagouni [27] studied both students' and Greek language teachers' views for Greeklish. The majority of students uses Greeklish primarily in social media as it provides speed and facilitates communication. Students do not believe that Greeklish has a negative impact on their writing. The use of Greeklish is reduced among the final-year Lyceum students, who noticed a negative effect of Greeklish in their writing and decided to discontinue their use. This finding seems to contradict the findings of Moustaka et al. [35] who reported that the use of Greeklish is reduced only among the students of the last grade of the general high schools and not among the students of the other types of schools. Regarding Greek language teachers who participated in the survey, the overwhelming majority do not use Greeklish; only a small minority admit Greeklish use, especially in social media environments. Greek language teachers believe that Greeklish, along with other factors, contribute negatively to students' correct writing.
As it is evident from the research mentioned above, Greeklish is only used for specific communication purposes [61]. In this vein, Laghos et al. [62] investigated the use of Greeklish in YouTube. They collected over 1000 user comments  [62] maybe due to the small number of participants. However, the findings give evidence that each participant sticks to their usual choice of alphabetical encoding in text-messaging [61].
Reading and understanding a long text in Greeklish is a demanding task, claimed Chalamandaris et al. [48] partly due to the fact that common Greek words can be transliterated into many alternative representations deriving from a combination of all types of Greeklish and it is rather rare for a user to consistently use only one type of transliteration. Tseliga [63] found that reading and understanding a sentence in Greeklish requires more effort as it is in average more time consuming by over 40% than reading and understanding the same sentence in Greek. Lees, Politis & Koutsogiannis [64] contend that, contrary to popular belief, the use of the Latin alphabet does not pose a threat to the Greek language, but rather contributes to digital communication among youngsters. Their findings also show a definite preference for the Greek script over the Latin script. When using the Latin alphabet, students frequently adopt orthographic transliteration to maintain traditional Greek spelling, notwithstanding gender-related variances. Thoma [65] found that students with a medium to high language profile read Greeklish more readily than students with a low language profile. Another crucial finding was that the use of Greeklish has no bearing on or connection to pupils' spelling abilities. Students with intermediate to advanced language profiles and students with poor language profiles both utilize Greeklish, and there was no statistically significant difference in the link between code usage and spelling performance. Finally, there was no connection between the use of Greeklish and the amount of free time students spent online on their language profiles.

Methodology
The following experiment was conducted to address the aims of this research: Participants (mainly young Greek internet users) were asked to read two short texts of equal length displayed on their computer screens. One of the texts was written in Greek while the other in Latin. After reading each text the participants were required to answer 5 multiple choice questions related to the text they had just read. "the theft") described how the authorities of the Louvre Museum realized that the famous panting was stolen, while the second text presented to the participants (referred to as "the thief") described how the theft was carried out and the arrest of the thief. The texts did not require any specific knowledge or terminology and based on standard readability indexes, they are both appropriate for university graduates (see next Table 1).
In order to ensure that text characteristics did not have an effect on reading time, half of the participants were firstly presented the text "the theft" in Greek alphabet and then the text "the thief" in Greeklish while the other half participants were presented firstly the text "the theft" in Greeklish and then the text "the thief" in Greek.
The experiment was implemented in Javascript using jspsych [66], a javascript library dedicated to create and execute psychological experiments. The experiment was hosted in heroku, a cloud application platform [67] and the responses of the participants were saved to a Heroku Postgress database. Statistical analysis of the results was carried out using R [68].

Demographics
A total of 732 respondents participated in the experiment. Most of the participants were females (63.1%) whereas 36.9% were males.  (Table 2) as well as the education level of the participants ( Table 3).
The vast majority (97.95%) of the participants were high school graduates (at least) while at the same time 35.25% of them hold a higher education or a postgraduate degree.
Finally, although two out of three participants (66.53%) do not use Greeklish there is a significant percentage (33.47%) that does use Greeklish. Figure 1 presents the percentages of Greeklish use in regard to the gender and education level of the participants.

Reading Speed
In this section, several factors possibly affecting the reading speed of the two texts written in Greeklish and Greek are examined.
Firstly, statistical properties of reading time recorded for both Greek and Greeklish texts are presented. Initially, the users were informed that the aim of the experiment was to measure their reading speed in Greek and Greeklish texts.
It was explained to the users that they would have to read two short texts of equal size, one written in Greek and the other in Greeklish and it was pointed that after reading each text they would answer five multiple-choice questions based on the text they just read. To initiate and complete the reading session of each text, the users were prompted to press a designated "start reading" and "end reading" button. The text was not available to the users after they pressed the "end reading" button. The reading time was measured in miliseconds by utilizing jspsych's automated time recording feature and equaled the time elapsed between the "star reading" and "end reading" buttons press. There are not any missing values in the sample. Furthermore, the measures of central tendency (mean, median) and dispersion (standard deviation, interquartile range), and especially the low difference between the mean and the median, suggest minor discrepancy from symmetric distributions, which is confirmed by the histograms plotted in Figure 2.
Based on the results shown at Table 4, the mean value of the reading time is much higher for the Greeklish text than for the Greek one, a fact which gives rise to the question whether the mean reading time of the text written in Greeklish statistically differs from the mean reading time of the text written in Greek. The results of the appropriate paired sample t test are presented in Table 5, where it is shown that p value is lower than the common significance level 0.05 and therefore the null hypothesis of the equality of the means is rejected. The mean of the differences (11.47 sec) and the corresponding 95% confidence interval of the difference of the means between the Greeklish and Greek text, which is (10.27, 12.68), indicate that the mean reading speed of the Greek text is signifi-Social Networking cantly higher than the Greeklish text, which is rather expected since all the participants are Greek native speakers.
Next, factors that affect reading time were tested. Multi-way ANOVA was conducted on the influence of 6 independent variables (gender, age, education level, frequency of connecting to internet, alphabet of the first text, use of greeklish) on reading time. The assumptions of the multi-factor ANOVA are satisfied.  • Is there a statistically significant interaction effect between the above mentioned factors? More specifically, is there a statistically significant impact between the levels of the three factors under investigation?

Multi
The results are presented in Table 6. It is important to highlight the fact that there are no interactions effects between the factors examined. The two tables indicate that gender and use of Greeklish are both statistically significant factors affecting reading time for both Greeklish and Greek texts. Based on the results presented in Figure 3, reading time for women is higher than that for men and is also higher for non-Greeeklish users compared to Greeklish users. The first important conclusion is that men were faster than women in reading both texts possibly because females read the text more carefully than males [68]. The second important conclusion is that Greeklish users achieve higher reading speed than those who do not use Greeklish for both texts; as it is expected, the participants who were familiar with the Latin alphabet read significantly faster the Greeklish text compared to those who do not use Greeklish; However, the results showed that respondents who use Greeklish read faster the Greek text, also. The specific result may imply that the use of different alphabets helps the readers to improve the reading process and to increase their reading capability.  A rather unexpected finding was the fact that reading time of the Greek text was affected by the alphabet of the text displayed first. More specifically, the reading time of the Greek text was increased when it was read after the Greeklish text for all but the male participants who use Greeklish. This is shown in Figure   4, where reading time for the Greek text when displayed first (denoted with the letter A) and when displayed second (denoted with the letter B) is shown. A possible explanation for this finding may be that participants used different reading strategies for the Greek and the Greeklish text, with the strategy used for the Greek text being more efficient. When the Greeklish text is read first, the participants (excluding males who use greeklish) delayed to change to the more efficient strategy they used for the Greek text. However, this is an issue that needs to be further investigated.

Understanding
In this section, the understanding of the two texts by the participants is examined. It is reminded that after reading each text (Greek and Greeklish) the participants had to answer 5 multiple choice questions related to the text. Figure   5 demonstrates the distribution of correct answers of the respondents for Greeklish and Greek texts. It is worthwhile to mention that the frequency distributions of the correct answers have exactly the same shape despite the fact that the reading times are statistically different between the Greeklish and Greek text as it was explained in the previous section.     (1,4), (2,3) and (2,4) are also statistically significant, at significance level 0.05, with similar interpretation. The grey color characterizes the non-significant tiles meaning there is no discrepancy from the independence for these tiles. The results are similar for the Greek text. The conclusion in both cases is that the understanding of the second text is significantly better than the first one regardless of the alphabet of the text displayed first. A possible explanation is that the participants after reading the first text and answering the corresponding questions are more prepared about the type and the difficulty of the task when they read the second text and they therefore achieve a better score.
In order to investigate the explanatory variables that affect the "perfect" understanding of the text a multiple logistic regression model is adopted (e.g. [70]).
In this model, the binary response variable is one (denoting the event of "suc-

Conclusions
In Nearly one-third of the participants stated that they use Greeklish for writing. This is a rather low percentage of young people who use Greeklish when compared to similar reported findings. According to Kappatou [71], for example, 69.4% of young people aged 13 to 25 use Greeklish when chatting while an even higher adoption of Greeklish is reported in Moustaka et al. [35]  Greek and the Greeklish text have been reported by Tseliga [51] although the conditions of the reported experiment were different since the participants had to read a set of simple sentences and state if each sentence was true or false.
Greeklish users are slightly faster readers than non-Greeklish users. More specifically, Greeklish users read the Greek (Greeklish) text 3% (4%) faster that the Greeklish non-users. Finally, as far as the reading speed is concerned, an issue that needs further investigation, is the finding that reading time of the Greek text was affected by the alphabet of the text displayed first.
Regarding the subject of text understanding, the participants achieved the same score when answering the multiple choice questions for the Greek and Greeklish text as shown by the corresponding (nearly identical) score distributions which have a mean of 3.42 and 3.34 respectively. The score is not differentiated by the gender and thus one could not support the hypothesis that the slower reading speed observed by female participants leads to a better performance. On the other hand, it should be noted that the limited size of text as well as the limited number of questions asked, may not provide a sufficient framework to uncover the possibility of a better and deeper understanding due to more careful reading.
The score achieved by the participants of the study in answering the multiple choice questions of the second text was better than the score achieved for the first text regardless of the alphabet of the text displayed first. This difference on the performance of the participants may be attributed to the fact that when they read the second text they were better prepared about the type and difficulty of the questions asked.
The study has some limitations that are mentioned here. Firstly, in terms of the experiment design, participants were asked to read two short texts of equal length displayed on their computer screens. Users reading times were identified based on mouse clicks. While this method provides valuable data, it would be