Historical linguistics, also known as diachronic linguistics, is the scientific study of how languages change over time. It seeks to understand the nature and causes of linguistic change and to trace the evolution of languages. Historical linguistics involves several key areas of study, including the reconstruction of ancestral languages, the classification of languages into families, (comparative linguistics) and the analysis of the cultural and social influences on language development.

This field is grounded in the uniformitarian principle, which posits that the processes of language change observed today were also at work in the past, unless there is clear evidence to suggest otherwise.^{[not verified in body]} Historical linguists aim to describe and explain changes in individual languages, explore the history of speech communities, and study the origins and meanings of words (etymology).

Modern historical linguistics dates to the late 18th century, having originally grown out of the earlier discipline of philology, the study of ancient texts and documents dating back to antiquity. Initially, historical linguistics served as the cornerstone of comparative linguistics, primarily as a tool for linguistic reconstruction. Scholars were concerned chiefly with establishing language families and reconstructing unrecorded proto-languages, using the comparative method and internal reconstruction. The focus was initially on the well-known Indo-European languages, many of which had long written histories; scholars also studied the Uralic languages, another Eurasian language-family for which less early written material exists. Since then, there has been significant comparative linguistic work expanding outside of European languages as well, such as on the Austronesian languages and on various families of Native American languages, among many others. Comparative linguistics became only a part of a more broadly-conceived discipline of historical linguistics. For the Indo-European languages, comparative study is now a highly specialized field.

Some scholars have undertaken studies attempting to establish super-families, linking, for example, Indo-European, Uralic, and other families into Nostratic. These attempts have not met with wide acceptance. The information necessary to establish relatedness becomes less available as the time increases. The time-depth of linguistic methods is limited due to chance word resemblances and variations between language groups, but a limit of around 10,000 years is often assumed. Several methods are used to date proto-languages, but the process is generally difficult and its results are inherently approximate.

In linguistics, a synchronic analysis is one that views linguistic phenomena only at a given time, usually the present, but a synchronic analysis of a historical language form is also possible. It may be distinguished from diachronic, which regards a phenomenon in terms of developments through time. Diachronic analysis is the main concern of historical linguistics. However, most other branches of linguistics are concerned with some form of synchronic analysis. The study of language change offers a valuable insight into the state of linguistic representation, and because all synchronic forms are the result of historically evolving diachronic changes, the ability to explain linguistic constructions necessitates a focus on diachronic processes.

Initially, all of modern linguistics was historical in orientation. Even the study of modern dialects involved looking at their origins. Ferdinand de Saussure's distinction between synchronic and diachronic linguistics is fundamental to the present day organization of the discipline. Primacy is accorded to synchronic linguistics, and diachronic linguistics is defined as the study of successive synchronic stages. Saussure's clear demarcation, however, has had both defenders and critics.

In practice, a purely-synchronic linguistics is not possible for any period before the invention of the gramophone, as written records always lag behind speech in reflecting linguistic developments. Written records are difficult to date accurately before the development of the modern title page. Often, dating must rely on contextual historical evidence such as inscriptions, or modern technology, such as carbon dating, can be used to ascertain dates of varying accuracy. Also, the work of sociolinguists on linguistic variation has shown synchronic states are not uniform: the speech habits of older and younger speakers differ in ways that point to language change. Synchronic variation is linguistic change in progress.

Synchronic and diachronic approaches can reach quite different conclusions. For example, a Germanic strong verb (e.g. English sing ↔ sang ↔ sung) is irregular when it is viewed synchronically: the native speaker's brain processes them as learned forms, but the derived forms of regular verbs are processed quite differently, by the application of productive rules (for example, adding -ed to the basic form of a verb as in walk → walked). That is an insight of psycholinguistics, which is relevant also for language didactics, both of which are synchronic disciplines. However, a diachronic analysis shows that the strong verb is the remnant of a fully regular system of internal vowel changes, in this case the Indo-European ablaut; historical linguistics seldom uses the category "irregular verb".