Machine Translation in Saudi Arabia
Introduction
Technology has become an important part of human existence because technology is used in all spheres of life including translation. Globalization has created an increasing need for translation, and this has created a need for improving the technology and machines used for translation. Studies on translation machines have been going on from the 1950s, and translation machines have proved to be of great assistance to professional translators (Zbib, Malchiodi and Devlin 1-3). Studies on translation machines have yielded results that have led to the development of translation machines that have been used in several countries. Some of these translation machines have been developed for translating Arabic language in both Arabic countries like Saudi Arabia, and the West (Zbib, Malchiodi and Devlin 1-3).
Machine translation is a product of computer linguistics, which is a discipline developed by computer scientists and linguists. Computer technology has been employed in technical translation to enhance speed and make the process cost effective (Zbib, Malchiodi and Devlin 1-3). Scholars have observed that if researchers in machine translation come up with a perfect machines that can translate different languages, then people speaking different languages will be in a position to share information and ideas globally. The main objective of the translation process whether done by machines or humans is to preserve the meaning of the text (Zbib, Malchiodi and Devlin 1-3).
Studies on machine translation of the Arabic language have been prompted by the fact Arabic is language spoken by a considerable percentage of the human population in different countries (Zbib, Malchiodi and Devlin 1-3). This paper is going to explore different aspects of mechanical translation of the Arabic language. The first part of the paper will review descript machine translation of Arabic language in Saudi Arabia. The second part of the paper will review the advantages and disadvantages of machine translation with Arabic language. The third part of the paper will discuss the benefits of machine translation; giving example of some translation programs while comparing them. The last part of the paper while recommend that machine translation be included in the curriculum used to teach university students, and outline some of the benefits of this approach.
Descript Machine Translation of Arabic Language in Saudi Arabia
Most studies on machine translation that have been done in Saudi Arabia have mainly focused on the translation of the Arabic language to English and vice versa. The reason why most studies on mechanical translation in Saudi Arabia have focused on Arabic-English translation is to improve the communication between Saudi Arabia and English speaking countries (Bar, Choueka and Dershowitz 2-4). Most of the mechanical translation system used in Saudi Arabia are based on the transfer model. Some scholars have noted the challenges translators face when trying to translate English proverbs and idioms into Arabic. In the 1990s, Arabic scholars came up with English-Arabic mechanical transfer system, which was used to translate sentences obtained from political news originating from the Middle East (Bar, Choueka and Dershowitz 2-4).
Some scholars have examined the mechanical translation of nominal compounds in English into Arabic. These studies were influenced by the fact that nominal compound are used in reference and naming in all kinds of text. For instance, some Saudi scholars developed a mechanical translation system that specialized in transferring medical terms from English to Arabic (Bar, Choueka and Dershowitz 2-4). In addition, some Saudi scholars have discussed the importance of having a modular English-Arabic mechanical translation programs. Scholars have come up with mechanical English translation programs that translates subsets of knowledge, for instance, there are translation programs that uses artificial intelligence to translate the abstracts of research articles written in English into Arabic (K.Bar, Choueka and Dershowitz 2-4).
However, scholars have noted that little research has been done in creating Arabic-English mechanical translation systems. The few studies that have been done on Arabic-English mechanical translation have come up with methods for translating compound verbs (Bar, Choueka and Dershowitz 2-4). Other mechanical translation system that have been developed by Arabic scholars have focused on translating question statement from Arabic into English, while others have come up with translation engines that enable users to search the internet using the Arabic language. Some scholars in Saudi Arabia have come up chart parsers that are relatively efficient in translating Arabic sentences. Current studies on Arabic-English translation have focused on translating complex English nouns phrases into Arabic (Bar and Dershowitz 8-10).
Currently, scholars in Saudi are employing three basic techniques in coming up with programs they use to translate Arabic texts into English and vice versa. The mechanical translation systems used by these scholars differ in terms of sophistication and complexity. These techniques include the Interlingua approach, transfer-based approach, and direct transfer. However, most current studies on mechanical translation have adopted the transfer-based approach to mechanical translation (Bar and Dershowitz 8-10). There are several reasons that have made this approach attractive to researchers in mechanical translation. One of the reason is that most of the translation systems under the transfer-based approach are based on two languages, or used to translate limited concepts in a few languages (Bar and Dershowitz 8-10).
Second, scholars have favored the transfer-based approach because it is effective in situations where multilingual translations are required; under this approach it is possible to create a hub of languages where all translation can be executed (Bar
aqnd Dershowitz 8-10). The third reason scholars prefer this approach is that it enables the sharing of transfer modules, especially in scenarios where languages that are being translated are closely related. For instance, mechanical translation program used to translate English and Portuguese may have several common transformation with programs used to translate English and Spanish (Bar and Dershowitz 8-10).
Mechanical translation programs used to translate Arabic and English have three main components, which include a generation component, a transfer component, and an analyzer. One of the most commonly used programs in Arabic-English translation is Prolog. Translators in Saudi Arabia and other parts of the world prefer using Prolog because it has an efficient unification, a backtracking regime, and it is declarative (Aminian, Ghoneim and Diab 106).
The traditional approach of mechanical translation entails segmentation of the text being into words; these words are then looked up in the dictionary; the function of each word is then identified based on the applicable grammatical rules governing the language that is being targeted (Farwell 519-521). The process of translation entails morphological analysis of the text being translated; syntactic analysis of the text being translated; and how sentences are being generated in the language being targeted. However, scholars have noted that the bilingual mechanical translation used in Arab-English translations have some weaknesses because in most cases the electronic dictionaries used in these translation provide incomplete information that partially translates the text (Farwell 519-521).
Concerning semantic analysis, the traditional approach fails to spot the context and the hidden structure of a sentence that result in ambiguity during translation. To deal with ambiguities that result from using the traditional mechanical translation, translators are required to examine words based on the contextual meaning and establish the semantic connection in the words (Farwell 519-521). This is what researchers refers to as semantic analysis. Semantic analysis basically entails the following steps: knowing the predicate in a sentence which is crucial in establishing the meaning of a sentence; establishing the role and the internal connection between every argument; and lastly employing semantics to deal with the ambiguity (Farwell 519-521).
Some scholars have come up with rules that govern Arabic-English mechanical translation. Most mechanical Arabic-English translation programs follow specific rules from the time a sentence is fed into the program to the final output. For instance, some linguistic scholars in Saudi Arabia have developed mechanical translation programs that use tagger and parser based on some mechanical translation rules (Farwell 519-521). Some of these rules include: when the auxiliary is the same as will, then the future tense become indefinite. Second, when the auxiliary has the word “not”, then the sentence is in the negative form. Third when the tense of a sentence is in the future-indefinite and the sentence is in the negative active voice form, then the resulting Arabic sentence will start with verb subject instead of subject-verb (Farwell 519-521). The Benefits of Machine Translation
The technological developments that have been experienced in information technology have created an urgent need to do away with language barriers that prevent people from communicating. This has consequently led to an increase in the demand of translation programs because of an exponential increase in the amount of information that need to be translated in various fields (Gelbukh 304-306). In Saudi Arabia and other Arabic countries there is an increased urgency by people and organization to access new information technologies and discoveries. This is an indication that there is lots of information that need to be translated, and human translators cannot cope with the demand. Mechanical translation plays a significant role in feeling this gap (Gelbukh 304-306).
Mechanical translation will addressed the increased demand to translate texts, especially when what is needed is to translate the text as opposed to capturing the meaning of a text accurately. For example, mechanical translation can be effective in translating information online and website because it can be used to generate rough translations (Gelbukh 304-306). This means that one of the benefits of mechanical translation is that it can save human translators time they could have spent trying to translate repetitive texts and simple texts. In addition, some scholars have observed that one of the benefits of mechanical translations is that it will enhance the effort of availing new technologies to Arabs (Gelbukh 304-306).
The benefits of using mechanical translation in Saudi Arabia and other Arabic countries varies, but scholars have identified some key benefits. However, the potential benefits of mechanical translation in the Arab world seem to exist more in theory than in practice (Meftouh, Harrat and Jamoussi 30). One of the benefits is that mechanical translation will assist in addressing the demand in translations. A second benefit of mechanical translation to Saudi Arabia is that it helps in updating information in the fields of finance, economics, science, and technologies. Another benefits of mechanical translation is that it will assist in transferring technology and knowledge to the Arab world (Meftouh, Harrat and Jamoussi 30).
Moreover, some scholars have observed that mechanical translation will play a significant role in modernizing the Arabic language by adding new terms and concepts through developing new words and urbanization (Meftouh, Harrat and Jamoussi 30). The benefit of mechanical translation to the Arab world is also realized through making the process of retrieving and translating information easy to Arab users, and enable Arab users to interact with people in countries with advanced technological capacity (Meftouh, Harrat and Jamoussi 30).
Scholars that have identified the potential benefits of mechanical translation have also spotted some of the challenges speakers of the Arabic language face when using mechanical translation programs. The Arabic language is very rich, and it is belong to the Semitic languages that somehow contrast with Indo-European languages, and this creates difficulties during the translation process (Collins 1). Some of the features in the Arabic language that makes it to contrast with English and other European languages include: the writing in Arabic is from left to right; in the Arabic language there is no capitalization, and this makes it challenging in recognizing names of people and places; and the Arabic language lack punctuation rules, and the speakers are not properly trained in the few rules that are available (AlMutawa 21).
Other features of the Arabic language that pose a challenge to mechanical translators include: the word order in Arabic is flexible; the letters in Arabic change in shape depending on the position they occupy in a word; some Arabic letters have the a similar shape and they are only distinguished by assigning them certain special marks like dots; and sometimes Arabic sentences take a nominal form (AlMutawa 21). These unique features of the Arabic language often lead to several vague translations when machines are used. This has created the need of developing mechanical translation programs that take these unique features of the Arabic language into account (AlMutawa 21).
The Advantages and Disadvantages of Machine Translation with Arabic Language
Machine translation has several advantages, and one of these advantages in seen in productivity. In terms of productivity, machine translation enhances the productivity of human translators who normally tasked with the responsibility of undertaking pre-edit and post-edit translation or both (AlMutawa 27-28). The second advantage of machine translation is seen in terms of speed whereby mechanical translation programs are quicker than human translators. Cost is another advantage of mechanical translation because the only cost incurred in using mechanical translation is the initial cost of the program and its installation. The fourth advantage of machine translation is in confidentiality because people can use these programs to translate personal financial statements and emails (AlMutawa 27-28).
The advantage of mechanical translation is also seen in consistency, whereby translation program maintain the constituency of the text being translated, and this does away with the need to reexamine previous pages to determine how words were translated. Translation machines have also been preferred for their universality. In most cases, human translators normally specialize in certain types of domain (AlMutawa 27-28). On the contrary, mechanical translator have the capacity to translate texts into several domains if the users just turns on a corresponding setting. Another advantage of mechanical translation machines is in terms of availability because translation programs can be used around the clock (AlMutawa 27-28).
Another advantage of translation machines is brought by the fact that these machines enable the inclusion of the internal attributes of the texts being translated in various ways. For instance, most translation software have the capability to store language pairs, and other types of information referred to as attributes (AlMutawa 27-28). The attribute that is commonly stored by these programs include the date in which the document was created, the name of the individual who created the documents, the domain, identity of the project, and the client among others. Translators and other people using the program can benefit from the stored attributes by filtering the text for the attributes they find most significant (AlMutawa 27-28).
For instance, translators and other users of the program can look for texts that are similar to theirs by client and project. Another useful internal attribute of mechanical translation software is seen in the ability of these programs to store terminology databases (AlMutawa 27-28). Most of the mechanical translation software available in the market are equipped with a terminology database, which enable translators to fully benefit from all the features available in the mechanical translation software. The internal terminology databases found in these software enable translators to carryout fuzzy matching whereby a specific word or its use is suggested by the software (AlMutawa 27-28).
In the absence of a terminology database, translators using mechanical translation software will have to check for the meaning of new words in a separate electronic dictionary or use the traditional dictionary. Another advantage of mechanical translation software is seen in the ability of these software to carry out an analysis. For example, mechanical translation software have the ability to determine how long the translation process will take, which can be challenging task if translators do their job manually (AlMutawa 27-28). Moreover, mechanical translation software with good memory have the capacity to examine a document for repetitive texts and similar sentences. Mechanical translation software with good memory have the capacity to generate a raw word count ignoring features such as software code and HTML tags that could have an effect on the word count (AlMutawa 27-28).
Despite the advantages mentioned above, mechanical translation software have certain limitations or drawbacks. One of the disadvantages of mechanical translation software is that they lack in accuracy. In cases where an accurate translation of formal documents is required, then mechanical translation software cannot be utilized (AlMutawa 27-28). But in cases where the software is used, the translated text will require pre-editing and post-editing or both. The second disadvantage of translation software is that they generate mediocre translation quality of text with sentences and words that are vague. For a source language text to be accurately translated by a mechanical translator, the software used should be straightforward and vivid (AlMutawa 27-28).
Using translation software that includes vague words often leads to the generation of poor quality text. The third disadvantage of translation software is that these software are based on systematic and formal rules, and this lowers the capacity of these software to resolve the vagueness associated with these software by focusing on the context just like human translators (Mohamed 80-82). Another limitation of translation software is that these software cannot accurately translate certain types of words, for example, multiword terms that expresses ideas, proverbs, ellipses, opaque sentences, culture-specific words, colloquial phrases, vague words, and idiomatic phrases (Mohamed 80-82).
The disadvantage of translation software is also seen in their lack of the capacity to maintain certain features of the source culture text or include the culture of the targeted text. Translation software lack the capacity to retain a similar register as the source text, or develop translations for new developed technical terms. The disadvantage of mechanical translation software is also evident in the machine breakdown that the cost incurred in replacing these machines (Mohamed 80-82).
The disadvantage of mechanical translation programs is also seen in the tendency of these software to generate update at regular intervals. When the translation process is going on, translation software are likely to generate updates that can create major challenges for the translators, especially when the text being translated is large and the translator has made changes in the entire document (Mohamed 80-82). Translation software are also disadvantageous because they constantly generate revisions from the same text. Revisions in translation refer to new projects that make modifications to the texts that had been translated previously, and they capture the changes that have been made in the previous source text (Mohamed 80-82).
In most cases, after using a translation software, the translator is asked by the client to revise the text based on the translation manual that has not been released in the market yet, or a translation manual that is to be released shortly. The disadvantage of using mechanical translation software is also evident in the time a translator may spend in the translation process (Denkowski, Al-Haj and Lavie 68). Researchers have noted that despite the development mechanical translation software, the time translators spend in the translation process using these software may be the same as the time spent while using manual translation. On the contrary, some researchers have noted that in some cases, translators spend more time using translation software compared to manual translation (Farghaly 7-8).
Studies have shown that mechanical translation takes more time especially when translators are using new programs, which they have to master first. Some surveys have found that translators spend approximately one to two weeks before acquiring competency in using new programs (Farghaly 7-8). In addition, studies have shown that feeding new text to a mechanical translation program may require more time compared to manual translation process because the initial draft in mechanical translation is so important. Moreover, future mechanical translation normally depend on the initial information entered by a translator. Post-editing of a translated text is also determined by the quality of the initial draft (Farghaly 7-8).
Creating a terminology database is an exercise that is time consuming. Despite the advantages and disadvantages of mechanical translation mentioned above, evaluation of the effectiveness of different mechanical translation programs has remained a challenge. One of the challenges researchers in mechanical translation face is coming up with one good translation (Farghaly 7-8). On the contrary, there several perfect ways of translating any given text. These perfect translations may use different word, or word choices even if the exact same words are used in the translation. Nevertheless, human beings can vividly differentiate between a good and a bad translation (Farghaly 7-8).
Example of Some Translation Programs
The first mechanical translation program developed by an Arabic company was created over 20 years ago, and it was called the Sakhr system. This system avails an Arabic-English mechanical translation system that is bidirectional, and this means that the system is uses both statistical and rule-based processes to come up with accurate and fast translation. The Shakhr software is only mechanical translation system built in the Arab world that translates Arabic into English and English into Arabic (AlMutawa 30-34). The rest of the mechanical translation software used in the Arab world were developed in the west. According to the manufacturer of this software, it can be tailored for certain domains because of the extensive research that has been conducted in the field of natural language processing over the years (AlMutawa 30-34).
Some of the features found in the Sakhr software include media monitoring, news, speech-to-speech translation, online translation, professional translation services, and mobile translation among others (AlMutawa 30-34). The Sakhr software also has the ability to perform comprehensive analysis on the semantic, syntactic, lexical, and morphological levels to enable an extensive understanding of the text that is fed into the software. This enables the resolution of all types of vagueness before a text is translated. The Sakhr system is designed to be used by organization and companies that require translation, in addition to assisting in the translation of websites and documents. Later, researchers included a dictionary of medical terms into the system (AlMutawa 30-34).
A second mechanical translation system commonly used in the Arab world is the Tarjim software, which is a mechanical translation system availed freely online, and has the capability of translating Arabic into English and vice-versa (AlMutawa 30-34). This software translates web pages and texts, and it also includes dictionaries in the French, Arabic and English languages, spelling checker, and six glossaries. Compared to the Sakhr system, the Tarjim software provide simple translation services with few features that makes the software unsuitable for translating complex texts (AlMutawa 30-34).
Another mechanical translation system commonly used in the Arab world is the ATA software system, which is hosted in London, and specializes in translating business software written in Arabic. Al Mutarjim Al Arabey is another mechanical system commonly used in the Arab world, and it provides a bidirectional translation in Arabic and English in different types of texts, and the software can also be used by professional translators (AlMutawa 30-34). The Al Mutarijim Al Arabey software can be customized in many different ways in order to enhance the quality of the text being translated. The Al Mutarjim Al Arabey mechanical translation software has 46 special dictionaries in the fields of medicine, biology, veterinary, mathematics, physics, chemistry, geology and engineering. In addition, the software contain dictionary abbreviations, and it has the ability to translate proper names (AlMutawa 30-34).
The Al Mutarajim software also allow translators to add their personal dictionaries, and the software has the capability of translating over 10,000 words per minutes. Compared to the mechanical translation software mentioned above, the Al Mutarajim is advanced and suitable for translators specializing in scientific texts. Another mechanical translation software commonly used in the Arab world is the Al Wafi V. 4.0, which is an Arabic-English bidirectional software meant to be used by nonprofessional translators (AlMutawa 30-34). This translation software is easy to utilize, and suitable for people who require a quick translation of various texts. Al Wafi V. 4.0 has an English dictionary that contains over 200,000 words and an Arabic dictionary that has over 75,000 words. Moreover, the users of the software are able to customize their own dictionary (AlMutawa 30-34).
The Al Wafi mechanical translation software has the ability to translate proper nouns and a speed of translating up to a maximum of 1,000 words. The software contains and English spellchecker and a voice output that translators can use if they choose to do so. Golden Al Wafi V. 2.0 is another translation software commonly used in the Arabic world, and it is a bidirectional Arabic-English mechanical translation software (AlMutawa 30-34). The software is designed for professional users, and it contains a special dictionary that makes it effective for advanced translators. The dictionary that comes with the Golden Al Wafi V. 2.0 software has over 2 million entries in Arabic and English (AlMutawa 30-34).
In addition, the Golden Al Wafi translation software come with eight special dictionaries in sciences such as geology, engineering, chemistry, mathematics, physics, biology, veterinary science, and medicine. The software also allow users to create their own custom dictionaries. The Golden Al Wafi software has the capability of translating proper nouns and it has a translation speed of 1000 words per minute (AlMutawa 30-34). The software contains an English spellchecker and a voice output that translators can use if they so wish. The software has the ability to translate multiple documents, and display Arabic texts that have diacritics (AlMutawa 30-34).
MutarjimNet V. 1.0 is another mechanical translation software designed to be used in Arabic countries. This software is designed for companies and their employees. The inbuilt dictionary in this software has over 2.5 million words with more than 40 specialized dictionaries, which translators can only use to find the meaning of words (AlMutawa 30-34). The MutarjimNet V. 1.0 has the ability to translate PDF, email and word documents that are highlighted. The software has a translation speed of about 1,000 words every minute. Al Misbar is another English-Arabic mechanical translation software that is commonly used in the Arab world. It is unidirectional online mechanical translation software that provides translation for web pages and texts (AlMutawa 30-34).
Cimos systems is also a software that is used in translating international languages including Arabic, and the software is hosted in Paris. This software is mostly used by businesses, provides multilingual translation services that is effective in traditional translation. However, compared to the translation software discussed above, this software does not take into account special features in the Arabic language (AlMutawa 30-34). Most of the software discussed above consider special features in the Arabic language, but the difference is in the level of complexity which determines whether the mechanical translation software is suitable for professional translators, organizations, or lay translators (AlMutawa 30-34).
Al Nakel Al Arabi is a bidirectional Arabic-English mechanical translation system that has specialized dictionaries in areas such banking, computing, petroleum, gas, banking, medicine, aerospace, and finance among others. This system utilizes a knowledge database and a ruled-based approach (AlMutawa 30-34). Moreover, the software includes a parser that enable it to carry out a comprehensive syntactic analysis in addition to a limited semantic analysis to identify main idioms, phrasal verbs, and main verbs. The dictionary utilized by this program has 150,000 phrases and words, and has a speed of translating 60,000 words per hour. This is an internet-based product that is hosted in a server (AlMutawa 30-34).
Another web-based mechanical translation program is the multilingual translation system, and this product acts both as mechanism of assisting human translators, and as an automatic translation software that has the ability to provide rough translations. This software has the ability to translate approximately 100,000 words every hour (AlMutawa 30-34).
Another translation program commonly used in the Arab world is CAT tools or Computer-assisted translation. This translation program enable human translators to carry out the translation process using several translation software available in the market, and designed to assist in the translation process. Some of the translation tools available under CAT include translation memory tools, grammar checkers, terminology management tools, quality assurance tools, and spell checkers. There are other several mechanical translation software design in Europe and America that are used in the Arab world (AlMutawa 30-34).
Machine Translation Be Included in the Curriculum Used to Teach University Students and Some of the Benefits of This Approach
Literature on the question whether mechanical translation should be included in the curriculum used to teach university students has been growing, especially its inclusion in the general classroom. However, most studies have focused on how mechanical translation can be included in teaching computational linguistic, and what professional translator should be taught in regard to mechanical translation. In addition, some studies have investigated how mechanical translation can be used as a form of computer-aided language learning mechanism (Somers 1). In including machine translation in the curriculum, some universities have introduced Masters of Arts in Computer-Assisted Translation (Chan 240-241).
For example, the Chinese University of Hong Kong has a post-graduate program that focuses on both machine translation and computer-aided translation. In this university, computer translation is an elective introductory course that concentrates on knowledge transmission. Some scholars have identified four approaches of including computer translation in the curriculum (Chan 240-241). These approaches include teaching students about translation and computers, teaching trainee translators about translation software, teaching translation language, and teaching students about the use of translation software and machine. Scholars have recommended that the approach and curriculum used to teach university students about mechanical translation should be based on the first and fourth perspectives (Chan 240-241).
Experts have recommended that the mechanical translation courses taught to university students should include practical skills and theoretical concepts. Experts in mechanical translation have observed that computer-assisted translation is different from machine translation (Chan 240-241). In computer-assisted translation, the entire translation process is handled by computer, and the outcome of the translation process is determined by how machines systems are tailored and implemented. Hence, fundamental concept on the translation mechanism used by computers has to be taught to university students (Chan 240-241).
Students also need to be taught the mechanisms used to analyze the source language, and the process used to produce the output in the target language. Mechanical translation systems use different mechanisms and computer algorithms to generate translations, and university students should be taught how mechanical translations using these approaches (Chan 240-241). The mechanisms used in translation play an important role in determining the accuracy and quality of translation, and when students understand the different approaches used in translation they can identify the shortcoming and strengths of different translation approaches. Consequently, students can know the exact software they need to use to translate different types of texts, for example, scientific, non-academic and organizational texts (Chan 240-241).
Another benefit of teaching university students mechanical translation is that it enable students to carry out an evaluation of various mechanical translation systems based on their speed, the type of algorithms they use, and accuracy. In addition, based on the different developmental strategies employed by these systems, and the type of clients these systems target, the strength of these systems maybe sometimes be determined by the text being translated (Chan 240-241). The benefit of teaching university students mechanical translation is also seen in the fact that it enable students to understand crucial concepts in computer translation and the difference between computer translation and computer-assisted translation (Chan 240-241).
Some of the different mechanisms used in computer translation include: rule-based, corpus-based, example-based, memory-based, knowledge-based, statistical-approaches, pattern-based, and natural language text processing. Another benefit of teaching university student mechanical translation is that it provides them with a practical experience in using mechanical translation software, and this plays a significant role in expanding their knowledge based (Chan 240-241). In universities that teach students translation like the Chinese University of Hong Kong, students are allowed to work with advanced mechanical translation systems during class, and this has expanded the knowledge of these students in mechanical translation (Chan 240-241).
The importance of teaching mechanical translation to university students is also seen in the recent increasing in the demand of translators caused by an ever expanding globalization and internationalization of markets. It has also been noted that the increase in the demand for translators internationally has outgrown the number of human translators, and this confirms the importance of introducing mechanical translation causes at the university level (Dollerup 308).
AlMutawa, Faten. Machine Translation In Saudi Arabia. Thesis. Sharjah: American University of Sharjah, 2012. Print.
Aminian, Maryam, Mahmoud Ghoneim and Mona Diab. Handling OOV Words in Dialectal Arabic to English Machine Translation. Doha: Association for Computational Linguistics, 2014. Print.
Bar, K, Y.Choueka and N. Dershowitz. An Arabic to English example-based translation system. 26 June 2007. Web. 30 December 2015.
Bar, Kfir and Nachum Dershowitz. Using Semantic Equivalents for Arabic-to-English. Tel Aviv: School of Computer Science, 2011. Print.
Chan, Sin-Wai. Routledge Encyclopedia of Translation Technology. New York: Routeledge, 2015. Print .
Collins, Michael. Statistical Machine Translation: IBM Models 1 and 2. 20 October 2011. Web. 30 December 2015.
Denkowski, Michael, Hassan Al-Haj and Alon Lavie. Turker-Assisted Paraphrasing for English-Arabic Machine Translation. Los Angeles: Association for Computational Linguistics, 2010. Print .
Dollerup, Cay. Teaching translation and interpreting 2 : insights, aims, visions. Papers from the Second Language International Conference Elsinore, Denmark 4 6 june 1993. Amsterdam : Benjamin , 1994. Print.
Farghaly, Ali. “Arabic machine translation: A developmental perspective .” International Journal on Information and Communication Technologies, 3.3 (2010): 3-10. Print .
Farwell, David. Machine translation and the information soup Langhorne, PA, USA, October 1998. Berlin Heidelberg Springer, 2002. Print.
Gelbukh, Alexander. Computational linguistics and intelligent text processing : 14th international conference : proceedings. Berlin : Springer, 2013. Print.
Meftouh, Karima, Salima Harrat and Salma Jamoussi. Machine Translation Experiments on PADIC:A Parallel Arabic DIalect Corpus. Shanghai: Pacific Asia Conference on Language, Information and Computation , 2015. Printer.
Mohamed, Azza Abd El-Monjiem. Machine translation of noun phrases: From English to Arabic. Thesis. Cairo: Cairo University, 2002. Print.
Somers, Harold. Detecting Inappropriate Use of Free Online Machine Translation by Language Students – A Special Case of Plagiarism Detection. Manchester: School of Informatics University of Manchester, 2006. Print.
Zbib, Rabih, et al. Machine Translation of Arabic Dialects. 7 August 2013. Web. 30 December 2015.