When a paid-for application becomes free, there are two possible reasons behind: (a) it has sold so well and the developer has made so much money that they have decided to make it free as a gesture of goodwill to society; and (b) it has sold badly.
Grammarly, an application that detects errors in essays, has become free. I’m not going to speculate about the reason behind, but like the couple of other proofreading applications available on the market (including the free Microsoft WORD grammar checker), Grammarly is far from being reliable. In fact, it’s unlikely that a truly reliable grammar-checking application will be available in the next 100 years.
The reason is simple. Human language is infinitely generative. At any second, we may be producing utterances that no other person in the world has ever said before, like this very sentence. Now, although we might think computers are very smart and powerful, they still have to be taught what is a right sentence and what is wrong in a problem sentence. But given the infinitely generative nature of language, so that new sentences are popping up all the time, it is simply impossible to teach computers all possible problem sentences. Hence, no matter how powerful super-computers will become one day, this is an area that we humans can be forever proud of — that we are endowed with this innate capacity for language.
But what about feeding computers with all the grammar rules, and then letting them do the error detections? Well, that works with very simple ‘rules’, such as teaching the computer that if (a) “it’s” or (b) “its” is followed by ‘a/an …’ then (b) is wrong (but I’m sure you can still find some counter examples); or when ‘a’ is followed by a word that ends with ‘s’, this is probably an error (but even so, you will quickly see that there can be a problem, as some singular words end in the letter ‘s’, and then what if a new word enters the English lexicon!). But anything beyond that, the computer is at its wit’s end. We can feed the rules of syntax such as SVO, SVC, SV, etc., into a computer, but computers can’t even analyse a slightly longer sentence into Subject, Verb Phrase, Direct Object, Indirect Object, Complement, Adverbial, Relative Clause, etc. Now, a lot of the grammar errors made by ESL learners have to do with sentence structure (e.g., I very like chocolate), and it is exactly this important area where grammar-checking applications fail.
So, this should be good news for language teachers, because they will never be replaced by computers since they ARE needed to mark and correct students’ compositions. But there is a paradox here for language teachers – how they wish there was a fully reliable error correction programme that could mark all the student compositions for them, so that they don’t have to burn the midnight oil day in and day out, raving at the thousands of tedious mistakes that students make over and over again.