Can Google Translate be used for “Spoken English” App.?

A frequent question posed to me is why have a Learn Conversational English from Telugu software when Google Translate can do the job!

Here’s how Google Translate handled three sentences that I tried a couple or so days ago, with comments (slightly edited) on the translation from Telugu friends:

We plan to meet at the mandir today.

ఈనాడు మనం మందిర్ వద్ద కలిసే ప్రణాళిక.

Friend wrote: Sounds correct. A Telugu-speaking colleague confirmed that the last word is correct; I didn’t know it 🙂 But she also says that the whole thing [all three examples] is archaic/pure Telugu that isn’t spoken widely any more.

Ravi responded: I need current widely spoken Telugu. So this translation is not good enough for me.

Rama dropped in to meet us.

రామ మాకు కలిసే లో పడిపోయింది.

Friend wrote: This reads like nonsense. An English translation would be “Rama fell down like meeting us” 🙂 [Though the gender in the Telugu doesn’t match. Rama is male, and padipoyindi is female.]

Ravi responded: Complete failure here.

Another Friend wrote: My initial reaction to ‘Rama dropped in to meet us’ was that the sentence uses idiom unknown to the Telugu language, but then it even failed on ‘Rama came to see us’ & ‘Rama came to meet us’ – and failed badly.

The children enjoyed themselves thoroughly.

పిల్లలు బాగా తమని తాము ఆస్వాదించాను.

Friend wrote: Kind of okay, except that
a) it should be “aswadincharu” not “-anu”. The “-anu” ending is used for the first person, and “aru” for plural (or in a respectful sense). So “nenu bank ki veltanu” means “I will go to the bank” and “vallu bank ki veltaru” means “they will go to the bank”. Notice the “-anu” vs “-aru” ending. Though you probably know all this stuff anyway by now 🙂
b) “aswadincharu” is not appropriate in this context, according to my colleague, who says it’s used to denote thorough pleasure via one of the five senses (like taste, perhaps).

Ravi responded: I need it to be accurate as it is a ‘teacher’ software. So even this fails the test.

Friend wrote: Yea, quality-wise Google translate doesn’t cut it. I thought it’s widely known that this kind of automated translation software isn’t teaching-quality or even meant to be used for important stuff. There are a lot of jokes caused by machine translation on the internet.

Ravi conclusion: So, as of now at least, Google Translate is just not reliable or accurate enough to be an English/Telugu 2nd Language teacher.

Machine Translation Accuracy in General

The website of Prof. Harold Somers,, states that he is a Professor of Language Engineering (Emeritus) whose main interest has been Machine Translation (MT) and related technologies.
A presentation on his site, “Does Machine Translation have a role in language learning?”:, in its concluding slide states, “MT may be a nice toy – a novelty – but it’s not designed as a language-teaching tool, so you shouldn’t use it as one”; “If you want to haul hay, get a tractor”. However the latest date mentioned in the presentation is 2003. So I don’t know whether this conclusion stands even today.

Here is a March 2010 NYTimes Op-Ed article by David Bellos, Director of Princeton’s Program in Translation and Intercultural Communication ( It does not touch upon MT and 2nd language teaching but does comment on advantages/disadvantages of MT including Google Translate.

Google Translate API – Google May Turn It Off Anytime!

The wiki page on Google Translate:, states that Google announced in May 2011 that it will shutdown the API on December 1, 2011. Subsequently there was a lot of criticsm which led Google to cancel their termination plan and instead offer a paid service. See comments on to get an idea of the developer pain caused by the shutdown announcement.

I think it will be wise for to stay far away from Google Translate API. Anyway it is a paid API now and so we cannot use it.

But Google Translate website is still available free. Though, due to its quality not being great for English/Telugu translation I feel we should steer clear of it too.

Sentences Content & Categorization Are Essential

Even if some free machine translation software did accurate translation, there is the big task of choosing/composing sentences and categorizing them for the lessons to be followed by “Spoken English” students. Conversational language learning techniques seem to use scenarios (to get some idea see: So “Spoken English” software & content would be required anyway (even if MT were used).

This entry was posted in Spoken English App. Blog. Bookmark the permalink.