A language family is a group of languages related through descent from a common ancestral language or parental language, called the proto-language of that family. The term "family" reflects the tree model of language origination in historical linguistics, which makes use of a metaphor comparing languages to people in a biological family tree, or in a subsequent modification, to species in a phylogenetic tree of evolutionary taxonomy. Linguists therefore describe the daughter languages within a language family as being genetically related.^[1]

According to Ethnologue the 7,111 living human languages are distributed in 141 different language families.^[2] A "living language" is simply one that is currently used as the primary form of communication of a group of people. There are also many dead languages, or languages which have no native speakers living, and extinct languages, which have no native speakers and no descendant languages. Finally, there are some languages that are insufficiently studied to be classified, and probably some which are not even known to exist outside their respective speech communities.

Membership of languages in a language family is established by research in comparative linguistics. Sister languages are said to have a "genetic" or "genealogical" relationship. The latter term is older.^[3] Speakers of a language family belong to a common speech community. The divergence of a proto-language into daughter languages typically occurs through geographical separation, with the original speech community gradually evolving into distinct linguistic units. Individuals belonging to other speech communities may also adopt languages from a different language family through the language shift process.^[4]

Genealogically related languages present shared retentions; that is, features of the proto-language (or reflexes of such features) that cannot be explained by chance or borrowing (convergence). Membership in a branch or group within a language family is established by shared innovations; that is, common features of those languages that are not found in the common ancestor of the entire family. For example, Germanic languages are "Germanic" in that they share vocabulary and grammatical features that are not believed to have been present in the Proto-Indo-European language. These features are believed to be innovations that took place in Proto-Germanic, a descendant of Proto-Indo-European that was the source of all Germanic languages.

Language families can be divided into smaller phylogenetic units, conventionally referred to as branches of the family because the history of a language family is often represented as a tree diagram. A family is a monophyletic unit; all its members derive from a common ancestor, and all attested descendants of that ancestor are included in the family. (Thus, the term family is analogous to the biological term clade.)

Some taxonomists restrict the term family to a certain level, but there is little consensus in how to do so. Those who affix such labels also subdivide branches into groups, and groups into complexes. A top-level (i.e., the largest) family is often called a phylum or stock. The closer the branches are to each other, the closer the languages will be related. This means if a branch off of a proto-language is 4 branches down and there is also a sister language to that fourth branch, then the two sister languages are more closely related to each other than to that common ancestral proto-language.

The term macrofamily or superfamily is sometimes applied to proposed groupings of language families whose status as phylogenetic units is generally considered to be unsubstantiated by accepted historical linguistic methods. For example, the Celtic, Germanic, Slavic, Italic, and Indo-Iranian language families are branches of a larger Indo-European language family. There is a remarkably similar pattern shown by the linguistic tree and the genetic tree of human ancestry^[5] that was verified statistically.^[6] Languages interpreted in terms of the putative phylogenetic tree of human languages are transmitted to a great extent vertically (by ancestry) as opposed to horizontally (by spatial diffusion).^[7]

Some closely knit language families, and many branches within larger families, take the form of dialect continua in which there are no clear-cut borders that make it possible to unequivocally identify, define, or count individual languages within the family. However, when the differences between the speech of different regions at the extremes of the continuum are so great that there is no mutual intelligibility between them, as occurs in Arabic, the continuum cannot meaningfully be seen as a single language.

A speech variety may also be considered either a language or a dialect depending on social or political considerations. Thus, different sources, especially over time, can give wildly different numbers of languages within a certain family. Classifications of the Japonic family, for example, range from one language (a language isolate with dialects) to nearly twenty—until the classification of Ryukyuan as separate languages within a Japonic language family rather than dialects of Japanese, the Japanese language itself was considered a language isolate and therefore the only language in its family.

Most of the world's languages are known to be related to others. Those that have no known relatives (or for which family relationships are only tentatively proposed) are called language isolates, essentially language families consisting of a single language. An example is Basque. In general, it is assumed that language isolates have relatives or had relatives at some point in their history but at a time depth too great for linguistic comparison to recover them.

A language isolated in its own branch within a family, such as Albanian and Armenian within Indo-European, is often also called an isolate, but the meaning of the word "isolate" in such cases is usually clarified with a modifier. For instance, Albanian and Armenian may be referred to as an "Indo-European isolate". By contrast, so far as is known, the Basque language is an absolute isolate: it has not been shown to be related to any other language despite numerous attempts. Another well-known isolate is Mapudungun, the Mapuche language from the Araucanían language family in Chile. A language may be said to be an isolate currently but not historically if related but now extinct relatives are attested. The Aquitanian language, spoken in Roman times, may have been an ancestor of Basque, but it could also have been a sister language to the ancestor of Basque. In the latter case, Basque and Aquitanian would form a small family together. (Ancestors are not considered to be distinct members of a family.)

A proto-language can be thought of as a mother language (not to be confused with a mother tongue, which is one that a specific person has been exposed to from birth^[8]), being the root which all languages in the family stem from. The common ancestor of a language family is seldom known directly since most languages have a relatively short recorded history. However, it is possible to recover many features of a proto-language by applying the comparative method, a reconstructive procedure worked out by 19th century linguist August Schleicher. This can demonstrate the validity of many of the proposed families in the list of language families. For example, the reconstructible common ancestor of the Indo-European language family is called Proto-Indo-European. Proto-Indo-European is not attested by written records and so is conjectured to have been spoken before the invention of writing.

Shared innovations, acquired by borrowing or other means, are not considered genetic and have no bearing with the language family concept. It has been asserted, for example, that many of the more striking features shared by Italic languages (Latin, Oscan, Umbrian, etc.) might well be "areal features". However, very similar-looking alterations in the systems of long vowels in the West Germanic languages greatly postdate any possible notion of a proto-language innovation (and cannot readily be regarded as "areal", either, since English and continental West Germanic were not a linguistic area). In a similar vein, there are many similar unique innovations in Germanic, Baltic and Slavic that are far more likely to be areal features than traceable to a common proto-language. But legitimate uncertainty about whether shared innovations are areal features, coincidence, or inheritance from a common ancestor, leads to disagreement over the proper subdivisions of any large language family.

A sprachbund is a geographic area having several languages that feature common linguistic structures. The similarities between those languages are caused by language contact, not by chance or common origin, and are not recognized as criteria that define a language family. An example of a sprachbund would be the Indian subcontinent.

The concept of language families is based on the historical observation that languages develop dialects, which over time may diverge into distinct languages. However, linguistic ancestry is less clear-cut than familiar biological ancestry, in which species do not crossbreed.^[9] It is more like the evolution of microbes, with extensive lateral gene transfer: Quite distantly related languages may affect each other through language contact, which in extreme cases may lead to languages with no single ancestor, whether they be creoles or mixed languages. In addition, a number of sign languages have developed in isolation and appear to have no relatives at all. Nonetheless, such cases are relatively rare and most well-attested languages can be unambiguously classified as belonging to one language family or another, even if this family's relation to other families is not known.