An automatic Greeklish to Greek transliteration system Aimilios Chalamandaris, Athanassios Protopapas, Pirros Tsiakoulis, Spyros Raptis Institute for Language and Speech Processing. Using the lEM software program developed by J.K. Vermunt [15] at the University of Tilburg. The saturated. Greek gibberish to Greek converter- convert extended ASCII Greek characters into normal Greek Convert gibberish greek characters, greeklish converter, convert Greek, Greeklish conversion This tool has moved to this page.
Method
Participants
Some 44 native Greek speakers with extensive reading and writing exposure to Greeklish completed this experiment (see Table 1 for a full description of the participants’ use of Greeklish).Table 1Mean values of exposure and use of Greeklish per week, as calculated by the self-ratingsof the participants of Experiments 1 and 2
Experiment 1 | Experiment 2 | ||
---|---|---|---|
Greeklish reading | Hours/week | 13.1 | 13.4 |
Greeklish writing | Hours/week | 11.2 | 9.2 |
E-mails in Greeklish | % received | 44 | 54 |
% sent | 42 | 53 | |
Online messaging in Greeklish | Hours/week | 9.5 | 12 |
% of total time | 79 | 87 | |
Online forums in Greeklish | Hours/week | 7.7 | 5.9 |
% of total time | 66 | 90 |
Materials
A total of 200 five- and six-letter Greek target words were taken from the GreekLex database (Ktori, van Heuven, & Pitchford, 2008), and their Greeklish correspondences were obtained. These Greek–Greeklish pairs were transcriptions (not transliterations), given their extensive graphemic and phonemic overlap, measured as the number of visually close or undistinguishable graphemes and the number of shared phonemes. To identify the overlapping graphemes, 20 Spanish college students without any knowledge of the Greek alphabet were asked to indicate which of the Greek lowercase letters could be perceived as known Roman letters. We considered as overlapping those graphemes that were indicated to be visually similar by at least 70% of the sample (i.e., α–a, ε–e, ι–i, ο–o, τ–t, κ–k, χ–x, ω–w). The selected Greeklish–Greek repetitions had 70% and 92% overlapping graphemes and phonemes, respectively, as measured at the fine-grained level of individual characters. We further corroborated that the strings were graphemically similar at the coarse-grained level by asking another 32 Spanish readers to rate the Greeklish–Greek pairs on a 1-to-5 scale (5 referring to highly similar). The mean similarity score for the pairs was 4.2. The position of the graphemic overlap varied across the prime and target strings. Greek targets (e.g., σοκάκι [alley]) were preceded by masked primes that were (1) the same as their Greek target (e.g., σοκάκι), (2) the Greeklish transcription of the target (e.g., sokaki), (3) an unrelated Greek word (e.g., δάπεδο [floor]), or (4) the Greeklish version of the unrelated word (e.g., dapedo). Greek base words used in the unrelated (Greek and Greeklish) conditions were matched as closely as possible to the targets (see Table 2). Furthermore, in order to confirm the extensive use of the Greeklish version of the Greek words, we computed the number of times each precise Greeklish word form had been used in a Greeklish-to-Greek online translator (Chalamandaris, Protopapas, Tsiakoulis, & Raptis, 2006). Greeklish transcriptions had been used on average more than 300 times, suggesting that they were uniformly accepted as valid Greeklish items. Moreover, ratings on whether each Greeklish transcription was considered the preferred Greeklish version of the target were collected by 20 Greeklish users who did not participate in the experiment. On a 1-to-7 scale (7 representing the “best” Greeklish transcription), the Greeklish transcriptions were rated at 6.9. A set of 200 pronounceable Greek nonwords (e.g., παδέμο) was also created. These nonwords were preceded by Greek or Greeklish repetition or unrelated nonword primes. Four lists were constructed so that each target appeared only once in each list, each time in a different priming condition. Different participants were randomly assigned to each list.Table 2
Examples of the word materials with their lexical properties, as well as mean lexical decision times (RTs, in milliseconds) and error rates (Err%) obtained in Experiment 1
Target | Greek | Greeklish | ||||
---|---|---|---|---|---|---|
σοκάκι (alley) | Primes | Effect | Primes | Effect | ||
Repetition (R): σοκάκι | Unrelated (UR): δάπεδο (floor) | UR – R | Transcription (T): sokaki | Unrelated (UR): dapedo | UR – T | |
Frequency | 22 | 22 | 21 | 22 | 21 | |
Length | 5.7 | 5.7 | 5.7 | 5.7 | 5.7 | |
Graph. overlap | – | 5.7 (100%) | 0.0(0%) | 4.0 (70%) | 0.0 (0%) | |
Phon. overlap | – | 5.7 (100%) | 0.0 (0%) | 5.2 (92%) | 0.0 (0%) | |
RTs | 659 (16.1) | 699 (15.6) | 40 (5.3) | 677 (15.3) | 692 (16.5) | 15 (4.5) |
Err% | 3.8 (0.7) | 4.3 (0.6) | 0.5 (0.7) | 4.0 (0.5) | 4.0 (0.6) | 0.0 (0.7) |
Graphemic overlap and phonemic overlap are given as the mean number of position-specific common graphemes or phonemes between primes and targets. Mean percentages of overlap or reaction time and error rate standard errors are presented within parentheses. The mean reaction times and error rates (within parentheses) for nonwords were 775 ms (3.5%) and 783 ms (3.8%), in the Greek repetition and unrelated conditions, and 786 ms (4.3%) and 782(3.7%), in the Greeklish repetition and unrelated conditions. Graph., Graphemic; Phon., Phonemic
Procedure
Participants were individually tested in a well-lit soundproof room. The presentation of the stimuli and recording of the responses was carried out using DMDX (Forster & Forster, 2003). On each trial, a forward mask (i.e., ######) was presented for 500 ms. Next, the prime was presented for 50 ms, immediately followed by the target, which remained on the screen for a maximum of 2,500 ms. Primes were presented centered in lowercase 10-pt Courier New font (character width: 0.12 in.), and targets were in lowercase 12-pt Courier New font (character width: 0.16 in.), in order to avoid overlapping pixels. Participants were instructed to press, as quickly and accurately as possible, one of two buttons on the keyboard to indicate whether the target was a legitimate Greek word or not. They were not informed of the presence of the primes, and none of them reported conscious knowledge of the primes’ existence. Trial presentation was randomized across participants. Each participant received a total of 12 practice trials (6 words and 6 nonwords). The experimental session lasted approximately 15 min.
Results and discussion
Incorrect responses and reaction times shorter than 250 ms or greater than 1,500 ms (less than 2.5% of the word data) were excluded from the analysis. Mean latencies for correct responses and mean error rates are presented in Table 2. ANOVAs on reaction times and error rates by participants and items were conducted based on a 2 (type of relationship: repetition/unrelated) × 2 (script: Greek/Greeklish) × 4 (list: 1/2/3/4) design.
The ANOVAs on reaction times revealed a main effect of type of relationship: Targets were responded to faster (28 ms) when preceded by related primes than when preceded by unrelated primes, F1(1, 40) = 106.08, p < .001; F2(1, 196) = 54.37, p < .001. The main effect of script was not significant (a 6-ms difference; both ps > .13). Critically, the interaction between the two factors was significant, F1(1, 40) = 15.37, p < .001; F2(1, 196) = 10.98, p < .01. Subsequent pairwise comparisons showed faster responses (by 40 ms) to targets primed by their exact Greek repetitions, as opposed to by unrelated Greek primes, F1(1, 40) = 77.41, p < .001; F2(1, 196) = 48.73, p < .001. Targets were also responded to faster (by 15 ms) when primed by their Greeklish transcriptions than by unrelated Greeklish primes, F1(1,40) = 18.55, p < .001; F2(1, 196) = 9.37, p < .01, even though the magnitude of the identity priming effect was significantly larger than that of the Greeklish priming effect (a 25-ms difference).
ANOVAs on the arcsine-transformed1 error rates did not reveal any significant effects (all ps > .11).
The nonword data did not reveal any significant effects (all ps > .13).
Experiment 1 showed a significant masked identity priming effect with Greek primes, as well as a Greeklish masked priming effect with Greeklish transcriptions, suggesting that for experienced users of Greeklish, the Greeklish-to-Greek conversion takes place in a ballistic and unconscious manner. However, the significant difference in the magnitude of the Greeklish–Greek and Greek–Greek priming effects suggests that Greeklish transcriptions activate their Greek correspondences less effectively than do Greek repetition primes. We interpret this difference as reflecting a distinction at the graphemic level of processing, since the graphemic overlap was not total (70%), while the phonemic overlap was almost complete (92%). A strong graphemic component in the Greeklish masked priming effect would predict that under similar circumstances, Greeklish transcriptions should lead to larger Greeklish masked priming effects than would Greeklish transliterations, due to the increased graphemic overlap. Experiment 2 was designed to address this issue by examining Greeklish masked priming effects produced by Greeklish transcriptions and transliterations.
From Wikipedia, the free encyclopedia
Greeklish, a portmanteau of the words Greek and English, also known as Grenglish, Latinoellinika/Λατινοελληνικά or ASCII Greek, is Greek language written with the Latin alphabet. Unlike standardized systems of Romanization of Greek, as used internationally for purposes such as rendering Greek proper names or place names, or for bibliographic purposes, the term Greeklish mainly refers to informal, ad-hoc practices of writing Greek text in environments where the use of the Greek alphabet is technically impossible or cumbersome, especially in electronic media. Greeklish is commonly used on the Internet when Greek people communicate by e-mail, IRC, instant messaging and occasionally on SMS.
Sometimes, the term Greeklish is also used informally for a non-standard language variety used by bilingual speakers of English and Greek, i.e. Greek with heavy admixture of English words or vice versa.
|
History
Some older traditions of using the Latin alphabet for Greek existed in earlier centuries. The term frankolevantinika properly refers to the use of the Latin script to write Greek in the cultural ambit of Catholicism. ('Frankos' is the Greek and Levantine term for Western European, and by extension Roman Catholic.) This usage was part of the broader tendency in the region for script to follow creed (e.g. Greek script for Turkish Orthodox Christians -- 'karamanlidika', and the use of Greek and Arabic script in Albania), and was routine in the Venetian-ruled Aegean in the Early Modern era. Indeed, the autograph manuscripts of several Greek literary works of the Renaissance are in Latin script (e.g. the comedy Fortounatos by Markos Antonios Foskolos, 1655). This convention was also known as frankohiotika/φραγκοχιώτικα, 'Catholic Chiot', alluding to the significant presence of Catholic missionaries based on the island of Chios. Hearkening back to this established term, a common (but derogatory) term for Greeklish is frankovlahika/φραγκοβλάχικα -- 'hillbilly Western' (exploiting the negative cultural stereotype among ethnic Greeks of the Vlachs).
Orthographic and phonetic Greeklish
Greeklish may be orthographic or phonetic. In orthographic use, the intent is to reproduce Greek orthography closely: there is a one to one mapping between Greek and Latin letters, and digraphs are avoided, with occasional use of punctuation or numerals resembling Greek letters rather than Latin digraphs. While letters are in the first instance chosen for phonetic similarity, visual equivalence, and corresponding keyboard keys, are used when phonetically similar letters are exhausted. Thus, psi (ψ) may be written as ps or y; xi (ξ) as x or 3 ; and theta (θ) as th or 8.
In phonetic use, there is no concern to reproduce Greek orthography, and the Greeklish is a phonetic transcription (usually with English phonetic norms, sometimes with other languages' like German) of Greek words --- although often there is a mixture of the two. In particular, iotacism is preserved: the various letters and digraphs now pronounced as /i/ are transcribed as i, and not differentiated as they are in an orthographic scheme (e.g. h, i, u, ei, oi for η ι υ ει οι). In a phonetic scheme, xi is usually x or ks or 3; ks or 3 is used if x has been chosen, following orthographic norms, for chi (χ). Psi and theta will usually be the digraphs ps and th. 3 is often used to represent xi (ξ) because of the similar shape of the number 3 to the original letter albeit mirrored.
An example of orthographic Greeklish could be the word 'plateia', which in Greek means 'square' and using the Greek alphabet is spelled 'πλατεία'. The word 'plateia' derives from the exact replacement of each Greek letter with its Latin respective: π=p, λ=l, α=a, τ=t, ε=e, ι=i, α=a.
An example of phonetic Greeklish could be the same word, 'square', written like this: 'platia'. The reason the same word is, in this occasion, written without the letter 'e', is the fact that, phonetically, the word 'square' in Greek soundsexactly like this: 'platia' (since -'εί'- is now pronounced /i/, as an instance of iotacism).
The most extreme case of orthographic Greeklish, which achieves the greater optical resemblance to the Greek prototypes, is perhaps the so-called 'byzantine' or 'arabesque' or 'calligraphic/artistic' Greeklish introduced in the Hellas mailing list by the mathematician George Baloglou. Main characteristics of Baloglou's 'byzantine' is the distinction of σ and s (σ=c ς=s), the distinction οf lower and upper letters, such as π=n, Π=TT or 5, θ=8, Θ=0 or Q, ψ=y, Ψ=4, and the unusual, but with great resemblance with the Greek prototype, transliterations σ=c, π=n ρ=p Ρ=P.
Books written in Greeklish
Giannis Androutsopoulos (see References) talks about Exegesis, a book in Greeklish that was published by Oxy Publications in 2000. The Greeklish transliteration was based on the Greek translation of the original book written by Astro Teller. A novel about Artificial Intelligence, it describes a computer program that has acquired a 'mind' of its own. The original book was written entirely in the form of e-mail messages, something that prompted Mr. Androutsopoulos and his collaborators to publish a version of it in Greeklish.
Web sites written in Greeklish
Most Greek personal or informal web sites were written in Greeklish in the past. Today this is not the case, as the use of Greeklish on a web site is considered inappropriate. It has been considered by many as an act of vandalism of the Greek language. However, there are still many Greek web sites which utilize Greeklish.
Greek companies which use Greeklish
Some Internet Service Providers in Greece use both Greek and Greeklish in their emails. For example, the corporate announcements sent to users via email are usually written in English, Greek, and Greeklish.
Use in advertisements
As of 2008, business advertisements using greeklish have appeared in Attiko Metro and other areas. Companies that used greeklish in some of their advertisements include Pizza Hut and Forthnet.
Use in business communication
Use of Greeklish for business purposes or business communication is considered as a lack of business ability or respect.[citation needed]
Current trends
Around 2004 a hostile movement against Greeklish appeared in many Greek online Web discussion boards (forums) where Greeklish was the primary 'language' of communication. Administrators threatened to ban users who continued to use Greeklish, thus making the use of Greek mandatory, but using Greeklish failed to become a serious reason to get banned. Examples include the Translatum Greek Translation Forum, the Athens Wireless Metropolitan Network Forum, the Venus Project Forum, the adslgr.com Forum, the e-steki.gr forum, the Greek Technological Forum and the e-foititis.gr greek student forum. The reason for this is the fact that text written in Greeklish is considerably less aesthetically pleasing, and also much harder to read, compared to text written in the Greek alphabet. A non-Greek speaker/reader can guess this by this example: 'δις ιζ χαρντ του ριντ' would be the way to write 'this is hard to read' in English but utilizing the Greek alphabet.
A counter argument used by forum users is that a lot of users live abroad, and write from computers they don't own (university or internet cafes). There, they don't have the ability to write in Greek (lack of fonts or proper locale), so Greeklish is the only option (because it's much simpler than it seems).
On Greek IRC and IM, most of the time only Greeklish is used.
On the Facebook social networking web site there is a group named 'Greeklish' which currently hosts more than 70 members who are pro the use of Greeklish.
Wide use for Greeklish in long texts is nowadays (2006) unusual. It is still used, however, among friends as an informal, alternative means of communication for short messages.
Another current trend in Greeklish is the introduction of Leet phrasing and vocabulary. Many Leet words or slang have been internalized within the Greek spoken language through Greek gamers online in games such as World of Warcraft.
Examples:
Greeklish | Explanation |
---|---|
Tsagia | 'Good bye', being a word meaning teas, but jokingly used as ciao in supposedly plural |
Re c | Pronounced 're sy' meaning roughly 'you' |
Kalimerez, Merez | Kalimeres (καλημέρες), meaning (Good) Mornings; note that the final z is inspired from byez |
Tpt | Tipota (τίποτα), meaning 'nothing' |
Dn | Den (δεν), meaning 'not' |
M | Mou (μου), meaning 'my' or 'mine' |
S | Sou (σου), meaning 'your' or 'yours' |
n | na (να), meaning 'to' or en (εν), meaning 'not' in cypriot dialect |
tr | tora (τώρα), meaning 'now' |
smr | simera (σήμερα), meaning 'today' |
klnxt | kalinixta (καληνύχτα), meaning 'goodnight' |
tlm | ta leme (τα λέμε), meaning 'we will talk again' |
sks | skase (σκάσε), meaning 'shut up' |
kn1 | kanena (κανένα), meaning 'no one' |
dld | diladi (δηλαδή), meaning 'so, therefore' |
vrm | variemai (βαριέμαι), meaning 'I am bored' |
Cypriot variant
Cypriot Greek has a distinct form of Greeklish, which reflects Cypriot phonology; for instance j can indicate the phone [dʒ], which is written in Cypriot Greek as τζι-, and corresponds to palatalised /k/ in standard Greek. For instance, Standard Greek και /ke/ [ce] 'and' = Greeklish kai/ke; Cypriot τζιαι [dʒe]̠ = Cypriot Greeklish tziai or je. Cypriot Leet/Instant Messaging use of Greeklish reflect this. For instance:
Cypriot Greeklish:
ego n 3ero re pe8kia.. skeftoume skeftoume omos tpt.. n mporo na me fantasto na asxoloume tin ipolipi m zoi me ena single prama.. kathe mera jini i idia i doulia. enna spaso. omos me tes epilogies p ekama .. tino pros iatrika j etsi.. (http://www.varkoume.com/phpBB2/viewtopic.php?p=558876&sid=07416b68bb274b6bb6954cba283449bb , 2006-03-25 )
Cypriot Greek:
Εγώ εν ξέρω ρε παιθκιά... σκέφτουμαι σκέφτουμαι όμως τίποτε... εν μπορώ να με φανταστώ να ασχολούμαι την υπόλοιπή μου ζωή με ένα sinɡle πράμα... κάθε μέρα τζιείνη η ίδια η δουλειά. Εννα σπάσω. Όμως με τες επιλογές που έκαμα... τείνω προς ιατρικά τζιαι έτσι...
IM-isms: n = en εν 'not' (in Standard Greek: d = den δεν); tpt = tipote τίποτε 'nothing' (tipota τίποτα in Standard Greek); j = je τζιαι 'and'
Examples
Καλημέρα, πώς είστε;
- Greeklish 1: kalimera, pos iste? (phonetic)
- Greeklish 2: kalhmera, pws eiste; (reconciling with spelling rules) (Baloglou's 'byzantine' variant: kalhmepa, nws eicte;)
- Typing as if the keyboard layout were set to Greek, when it is actually set to US English: Kalhméra, pvs eíste?
Θήτα (theta)
- Greeklish : thita
Greeklish-to-Greek conversion
Since the appearance of Greeklish there have been numerous attempts to develop applications for automatic conversion from Greeklish to Greek. Most of them can cope with only some of Greeklish transliteration patterns and can be found and downloaded in the Internet. The first complete system for automatic transcription of Greeklish into Greek, obtaining correct spelling is All Greek to Me! , developed and provided by Institute for Language and Speech Processing
The first open online application for the transcription of Greeklish to Greek, was developed by Artificial Intelligence Group at University of Patras, named deGREEKLISH.
An initiative has started to create a freely-available, open-source converter using user-supplied word transliteration: Greeklish OUT!. A similar open source Greeklish converter, written in the C# programming language, is available as a stand alone program.
See Also
References
Jannis Androutsopoulos (Γιάννης Ανδρουτσόπουλος), a linguist at King's College London, conducted extensive research on the history and sociolinguistics of Greeklish in 1998-2001; his publications, as well as publications in the media about the phenomenon, are available on a dedicated site.
External links
- Online Greeklish to Greek Converter by Qoolsoft.gr . API present to use it from any page or desktop.
- Online Greeklish to Greek and Greek to Greeklish Converter developed by site e-Greeklish.GR, Greece
- Online Greeklish to Greek Converter developed by George Galatas and Artificial Intelligence Group at Patras University, Greece
- Greek Conversions (GCNV32) v3.00 by CyLog Software ©1995-2002. Free program converting Greek-to-Greeklish and greelish-to-Greek, using customizable conversion schemes.
- TSIK Greeklish - Free Greek to Greeklish translation tool by Kyriakos Stavrou - TSIK
- Greeklish-to-Greek Conversion by Institute for Language and Speech Processing
- Greek-Greeklish Converter by ASDA
- Greek-Greeklish Converter by Translatum
- Transliteration Chart by Translatum
|
|