Tuesday, July 16, 2013

Language exchange websites and their interface flaws

I'm talking about Lang-8, verbling, italki. I'm going to compare how they permit you to identify languages you know/are studying, and how their system design affects their usability.

Lang-8 lets you list one native language, and then 2 (for non-premium members) languages you are studying. No indication of ability is available. It includes Latin, Gaelic, Mongolian, but not Ancient Greek.

verbling lets you list languages you are fluent in, and languages you are learning; they cannot be the same. It permits Latin, Gaelic, Mongolian, not Ancient Greek.

italki lets you list several languages, you can have multiple native languages, you may rank each language according to CEFR levels. It permits Latin, Mongolian, but not Gaelic or Ancient Greek. It includes an 'Other' category but you cannot specify what that language is, and may only have one.

1. Native language assumptions

Lang-8 forces you to choose only 1 native language. Too bad if (a) your bilingual from childhood, (b) your native language is a minority language not listed. The default assumption on Lang-8 is that native speakers, and really only native speakers, are the best for correcting the work of others in that language, no real concession is given to fluent L2 speakers of that language.

2. Rating languages

Lang-8 and verbling both operate with a simple binary opposition: you either know a language or you don't. Anyone who has studied a language and been asked, "So do you know X?" realises how difficult a question that is to answer. Unless you are a C2-level speaker you are likely to hedge your answer. verbling makes it more difficult to work around this issue, because you can't be fluent and a learner at the same time. So the ability of people to find and interact with more advanced learners is hindered. Why would someone want to do that? Quite simply, sometimes you don't need a fluent/native speaker, you would do fine practising with someone else who is learning the language. And in the case of minority languages, you are much more likely to find fellow learners than fluent speakers.

3. Minority language options

I have a fairly specific set of minority languages in view, but the problems I encounter are even worse for more endangered languages. Firstly, all three sites offer 'Latin', presumably because default lists of languages on the internet tend to list Latin. However, if you actually select Latin and see who has chosen Latin, you are often met with a lot of non-Latin learners, and sometimes Spanish speakers. Secondly, the inclusion of Mongolian is non-remarkable, it's a large enough language. Thirdly, I don't know why italki doesn't list Scottish Gaelic, it's an anomaly. At least 60,000 speakers. If you were a speaker or a learner of a more endangered language, say Blackfoot (estimated 5,000 speakers) you have no real options, you have to go to 'Other'. But 'Other' is an incredibly unhelpful category. Unless you can specify what 'Other' is, you have to trawl through every listing of 'Other' searching for another Blackfoot speaker.

As for Ancient Greek, I feel like if Latin can get an entry, Ancient Greek ought to have one.


Anyway, what am I getting at? These sites, like most of the internet language resources, favour majority languages. But that's not surprising. What is frustrating is that their system design makes them harder to use and ultimately less useful if you're interested in a minority language.

Simple improvements would change this:

1. Stop using simply binary distinctions between speaker/learner. Adopt the ability to rate every language according to its proficiency, like italki currently does.
2. If you allow error-correction of text entries, allow people to correct in languages they are not-fluent in. Either implement a vote system to allow for quality control, or limit it to CEFR listed C1/C2 speakers.
3. Allow multiple 'Other' options.
4. Allow user-specified 'Other' options, and let these be searchable.
5. Rewrite the list of languages offered according to ISO codes. The list would become huge, but you could simply separate it into two categories: major languages and minor languages, split by speaker population, and then anyone going for a majority language will easily find it, and people interested in minority languages will be able to go searching for their particular language.

[Edit: I will add in here Livemocha for completion. I think the functionality of Livemocha has declined since its recent takeover, but it actually does better in the language options than all the others: it allows multiple languages on both 'I speak' and 'I am learning', at multiple levels, and you can list the same languages in both groups, it also includes every language I am interested in except Ancient Greek, this is the best set-up of all the language sites I discuss]

No comments: