When we study Error Analysis and the causes of errors and mistakes in language learning, we also study 'literal translation' as one of the major causes.

It is a kind of translation, done by people or machines, that represents the original text, but it does not attempt to convey its style, beauty and semanticity. Lexical choices are the most important and difficult area for the translators, and a machine cannot perform well in this particular area; it cannot convey the spirit of the source text. We can depend on machine translation when we need to translate some short sentences of everyday use, but it cannot be recommended for formal and technical translation.

In writing and translation work, the use of punctuation marks is also very important, and it can change the whole story and meaning, and a machine cannot do it at all. Look at the following example.

USE A COMMA AND SAVE LIVES

1- Let's eat kids.

2- Let's eat, kids.

Machine translation (MT) has undergone quite interesting developments. However, it still has some linguistic, pragmatic, and affective issues, which require human post-editing to shun any occurrence of “lost in translation” or misinterpretation that could lead to a disservice. For instance, the study by Taira et al. (2021) concluded that Google Translate is not ready yet for use in clinical settings because of inaccuracies that could result in harm to patients.

Nonetheless, neural machine translation (NMT) seems promising despite being expensive and having issues in translating vague expressions. NMT is powered by artificial intelligence, i.e., improvements are constantly made to ameliorate the quality of NMT output. Here is the full citation to the article by Taira et al. (2021).

Taira, B. R., Kreger, V., Orue, A., & Diamond, L. C. (2021). A pragmatic assessment of Google Translate for emergency department instructions. Journal of General Internal Medicine. https://doi.org/10.1007/s11606-021-06666-z

Good luck,

I am certainly not an expert in this kind of area, but it seems to me that machine translation will not be able to handle pragmatic aspects of communication until it operates not only on the basis of a fully developed linguistic knowledge, but also on the basis of a fully developed knowledge of the world. It may take some time before this state of affairs comes to be!

This is very important question. I think they working nowadays on bridging this gap, I mean the contextualization issue.

MT is approximative and needs human intervention for better pragmatic contextual accuracy.

I think people will be surprised how far MT has come - in concert with machine production of language. Well, in English at least. At lot of groundwork wad done by Helmreich in the 1980s - early 2000s. While MT right now cannot provide a lossless (or unsupervised) type of translation, it continually improves the proficiency of comprehending the source text. I suggest to look at work by Christopher Manning (a lot to be found here on RG) and also this: Cohn-Gordon, R. and Goodman, N., 2019. Lost in machine translation: A method to reduce meaning loss. arXiv preprint arXiv:1902.09514.

The short answer is no. because machine can’t translate emotions and emotions change in different context. Machine can give the prescriptive meaning, but it can’t give give the descriptive meaning. However machine translation often follow some certain corpus. Like if you are studying the effects of some L1 on some L2. There is a high chance of both L1 and L2 are popular languages they probably will have a corpus to consult. and might translate properly.

There is no hard and fast way of assessing how good MT is as a translator, but depending on the pair of languages subject to a translation there is an accuracy rate of between 60 and 80%.

There is no one such thing as "machine translation". There are different approaches (statistical vs rules-based, for example) and certain MT engines will combine those approaches since advantages of one can counteract the disadvantages of the other.

A very basic demonstration of how MT can indeed take into account context (i.e. pragmatics) can be achieved as follows. Open the website

https://www.deepl.com/translator

and choose a source and a target language you understand. In the left, source panel, start typing a sentence, but type fairly slowly - take a long pause after each word. In the right, target panel, you will see the suggested translation(s) changing as they adapt to the improved context provided by each additional word in the source sentence.

I think, all texts, in a way or another, are pragmatics-based since they are produced by specific authors to convey a specific message within a specific context and to address specific audience. So, such specificity is what constitutes texts' pragmatics. In non-machine translating process, Translators, as they are treating a text throughout the translation stages, pay attention to such pragmatics factors, at least for two reasons: First to get the exact meaning of the original text (ST), and second to produce a target text (TT) that has the same 'semantic/pragmatic effect' on the readers. To do so, translators use their accumulated knowledge of the two languages and cultures, their professional skills and understand well the purpose of the required accurate translation(s). I think, Machine Translation (MT) can not do this alone to produce such required accurate translation.

According to Claude (2003) Machine translation (MT) has been both praised and criticized since the 1930's when it was first introduced. Today, MT − much improved since then, is a vital tool for the human translator, although not without its problems. One important unresolved issue is the success of MT for different text types.