References
ProtoLingua does not invent its linguistics. Every structural question in the app traces back to published fieldwork, and the percentages beside the answers are read from these datasets at run time rather than written in by hand. This page says exactly whose work that is.
Datasets
Five databases are embedded in the application. Each entry gives the licence it is published under and states what was taken, since none of them ships here whole.
PHOIBLE asks that inventory data cite the original source corpus rather than only the aggregate. The inventories embedded here come from SPA, UPSID, AA, PH, GM, RA, SAPHON, UZ, EA and ER.
Works the app is built on
Not data, but the findings the questions are shaped around. Where the app tells you something is near-universal, this is where that claim comes from.
- Berlin, Brent & Kay, Paul (1969). Basic Color Terms: Their Universality and Evolution. Berkeley: University of California Press. The order colour terms enter a language, which the colour questions follow and which settles the green/blue split for a language below the five-term stage.
- Comrie, Bernard & Smith, Norval (1977). Lingua Descriptive Studies: Questionnaire. Lingua 42. The standard checklist for describing a language. Its section numbers are carried on the questions, and the Conformance page answers it from your language.
- Greenberg, Joseph H. (1963). Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements. In Universals of Language. MIT Press. The word-order correlations the typology report checks — universals 2, 4 and 24 by name.
- Hockett, Charles F. (1960). The Origin of Speech. Scientific American 203. The design features. What makes a system a language at all, and the first questions the app asks.
- Keenan, Edward L. & Comrie, Bernard (1977). Noun Phrase Accessibility and Universal Grammar. Linguistic Inquiry 8. The accessibility hierarchy: languages relativise an unbroken run from the top of it, which is why the app asks where your language stops rather than which positions it allows.
- Nichols, Johanna (1986). Head-Marking and Dependent-Marking Grammar. Language 62. Where a relationship is marked — on the head or on the dependent. One of the deepest divisions between language types.
- Zipf, George K. (1949), and Bentz, Christian & Ferrer-i-Cancho, Ramon (2016). The law of abbreviation: the things a language says most often, it says most briefly. Measured here across 1.46 million word forms rather than assumed, and applied when the app coins a word.
A note on what is measured
Where this app says a choice is rare, it means the figure was counted in one of the datasets above at the moment the page was drawn. Nothing is hard-coded, so regenerating a dataset moves the figures and cannot leave a stale number behind. Where nothing honest can be said, no figure is shown at all.