Pick Your Human 

Why do most requesters choose the US for their research? For many, it’s the easiest way to ensure their work is done by a fluent English speaker. But in doing this, they narrow their demographic and miss out on some quality workers in the Mturk community. They also skip right over the fact that being in the US is not actually a guarantee that the worker is a fluent English speaker.

The bot checks available today may answer the “are you human” question, but maybe there are checks that could also open the requester to countries beyond the US. This would diversify their response pool and give them more dynamic data.

By asking simple questions with unique English identifiers in them, you could weed out those that may not be the best suited for your study. Let us consider a few different ways this can be done.

A few examples are that a study could include identifying idioms, keeping in mind idioms will exclude those not good with figurative speech. One could try mixing up the subject, object, and verb order as many non-English languages are subject, verb, and object. “The red car” vs “The car red”

Many languages can skip the subject completely as it’s inferred from the verb conjugation. English does not allow for this as easily. As an example: Question: “What did you do this weekend?” Answer:” Went to the mall.” Well, Who went to the mall? Just you, you and a friend, your family? English does not specify and thus to be clear, the subject must be present. Asking what is wrong with or missing from a phrase like “Went to the mall” could help verify a native/fluent speaker.

A sentence could simply leave out the first few words that define a comment, called a left-edge ellipsis, and then ask the worker to fill it in. For example: “You want coffee or tea?” Most English speakers would understand the sentence, but a native/fluent speaker would know that if grammar is a concern, the sentence should be “[Do] you want coffee or tea?”

Omitting preceding auxiliaries from the main verb and asking if the sentence is correct is another option as many other languages do not use them. For example: “I going to the bank.” leaves out a preceding auxiliary, a native/fluent speaker would know the phrase should read “I [am] going to the bank”

So requesters, if you would gain from having data outside of the US, consider including the unique aspects of the English language in your attention checks and bring other countries into your data.

Published by admin

Turkopticon helps the people in the 'crowd' of crowdsourcing watch out for each other—because nobody else seems to be! Almost half of the Mechanical Turk workers who wrote their Bill of Rights demanded protection from employers who take their work without paying. Turkopticon lets you REPORT and AVOID shady employers.