A dialect is a variety of a language tied to a place or social group, but there is no neutral rule for when a variety counts as a separate language. Mutual intelligibility is the usual test, though politics, history and identity also play a part.
Key takeaways
- Linguists often use mutual intelligibility to tell the two apart, but it comes in degrees and can run one way.
- Political and social factors frequently decide the label, as with Chinese, Arabic and Scandinavian.
- This map follows Glottolog, so Wu Chinese and Yue Chinese, with about about 81.4 million and about 73.1 million speakers (Ethnologue, 2019), are shown as separate languages.
- Other catalogues may split or join some languages, so counts differ.
How the line is drawn
There is no neutral rule for separating a language from a dialect. The most common linguistic test is mutual intelligibility: if speakers can readily understand each other without prior familiarity or special effort, the varieties are treated as dialects of one language [1]. The same source lists the problems. Understanding comes in degrees, close relatives can often manage if they try, and political and social factors often override the result [1].
Intelligibility can also be lopsided. In the Slovene and Croatian pair, one study found Slovenes understood Croats noticeably better than Croats understood Slovenes [1]. A test that gives different answers in each direction cannot draw a clean line.
That is why the usual summary is that the label depends on politics and prestige as well as speech. The phrase attributed to Max Weinreich, that a language is a dialect with an army and navy, makes the point, and the article on dialects says it makes any purely linguistic definition of "language" doubtful [2]. Institutional support such as dictionaries, schooling and official status is what typically marks out a standard variety [2].
Two patterns recur. A standard language can cover varieties that are hard to understand across the range, as with German [3]. Or neighbouring standards can be quite intelligible with each other and still be counted as separate languages, as with Scandinavian [3]. For the idea of a continuous chain of varieties, see dialect continuum.
An example from our data
The map follows Glottolog, which sets its own criteria for what counts as a language. Other catalogues may split or join some of these. The homepage says this too, in why areas overlap, and what the map cannot show.
Mandarin Chinese, Wu Chinese and Yue Chinese are listed as separate languages, with about about 918 million, about 81.4 million and about 73.1 million first-language speakers (Ethnologue, 2019 for each; approximate). Many speakers call all of these "Chinese" and see them as one language with regional varieties, and the Wikipedia article on mutual intelligibility notes that Chinese varieties are often called one language although distant speakers usually cannot understand each other [1]. Both ways of describing it are in use.
Danish and Swedish are the reverse case: separate languages with standards of their own, although spoken understanding between them is usually low [1].
The thing people get wrong
People often say "a dialect is just a language that is not written down, or not official". That is not how linguists use the word. Every variety is a dialect of something, and being a dialect says nothing about quality or status [2]. A language that has no army, no flag or no writing is no less a language.
The opposite mistake is to expect a clean test that settles every case. Even where the answer looks obvious, it can depend on whom you ask. This site does not take a side in disputed cases and follows its cited catalogue.
Frequently asked questions
Is a dialect a lesser form of a language?
No. Under the definition most linguists use, every variety of a language, including the standard one, is a dialect, so everybody speaks a dialect.
Are Danish and Swedish dialects of one language?
They are counted as separate languages, although they are related and speakers often understand part of each other. Spoken understanding is usually low, so the label rests on more than intelligibility.
Who decides what is a language?
No single body does. Catalogues such as Glottolog and ISO 639-3 make their own choices, and communities may see their speech differently from outside researchers.
Sources
- Mutual intelligibility, Wikipedia.
- Dialect, Wikipedia.
- Dialect continuum, Wikipedia.
- Every figure from our own map is worked out from the datasets on the data sources page (checked against Glottolog and Glottography, retrieved 13 October 2026).