Our top three guesses for your English dialect:
1. American (Standard)
2. US Black Vernacular / Ebonics
3. Australian
Our top three guesses for your native (first) language:
1. English
2. Finnish
3. Italian
Meh. How does this even work? I thought they might ask me to speak or sth.
The use of prepositions and articles vary among dialects of English, and to some extent pronouns and verb tenses. Classic example: Americans wait "in line", whereas queued up Brits wait "on line." Ebonics tends to drop parts of verbs: "where you live?" versus "where do you live?" Scots is a quasi-independent Anglo-Danish language, with strong overlays from Norman French and English conquerors. It has different vocabulary and a somewhat different adaptation/simplification of Germanic grammar than English. I can say from personal experience that it is only partially intelligible to a speaker of American English. (Scots itself isn't included in the study.) Because of the history of immigration (and deportations) Anglo-Irish and Scottish constructions tend to be relatively well represented in Australia, New Zealand (northern Ireland), and the interior highlands of North America (e.g., hillbilly speech represented in American country music)--and, of course, in the English spoken in Scotland, Ireland and northern England.
People are also more likely to think English sentences correct which have structure similar to sentences in their native language, or ignore differences which are not marked in their native languages. For example, some Asian languages don't have different verb + article forms to mark agreement in number with the subject, and thus those native speakers find it harder to learn English conventions for subject-verb-object agreement in number. The cartoons at the beginning of the quiz seemed mostly directed at subject-verb-object ordering, which differ among languages. Somewhere on the site I thought I saw the claim that native English speakers are more tolerant of Scottishisms and Irish dialect than people who learned English as a second language.
The "guesses" use multivariate distance measures and clouds of answers for people from known linguistic groups. They calculate the distance between your answers and the various clusters. As the cluster-distance results aren't updated automatically, people from groups not well-represented in their initial/earlier sample (e.g., native Chinese speakers) will have shortest distances to other groups reported. The researchers will probably update their "Grammars of English" similarity-distance results and graphics after a planned number of subjects have participated in the study.