Historical linguistics studies how languages change over time and how languages are related through common ancestry. The comparative method uses systematic sound correspondences across related languages to reconstruct proto-languages — ancestral languages not directly attested in written records. The Indo-European language family, reconstructed by 19th-century scholars, demonstrates that English, Latin, Sanskrit, and Persian all descend from a single ancestor. Language families are organized into genealogical trees, though sustained contact between languages also causes horizontal transfer through borrowing and creolization.
Study the classic sound correspondences between Latin, Greek, and Sanskrit for cognate sets. Apply the comparative method to a small constructed data set. Trace the history of familiar English words through etymologies to develop intuition for regular change.
You already know, from your study of phonological systems and language variation, that languages are not static — they shift over generations in pronunciation, morphology, and vocabulary. Historical linguistics takes that observation as its starting point and asks a more ambitious question: if languages change systematically, can we trace those changes backwards and reconstruct what earlier languages must have looked like, even when no written records survive? The answer is yes — through the comparative method, one of the most powerful inferential tools in the humanities.
The comparative method begins with cognates: words in different languages that appear similar in form and meaning because they descend from the same ancestral word, not because of borrowing. Latin *pater*, Greek *patēr*, Sanskrit *pitṛ*, English *father*, and Persian *pedar* all mean "father" and are strikingly similar. Crucially, the similarities are not random — they follow the same correspondences across dozens of other words. Latin *p* systematically corresponds to English *f* (Latin *piscis* / English *fish*, Latin *pes* / English *foot*). A single coincidence could be chance; a systematic pattern cannot be. When the same sound correspondence appears across a large vocabulary, the only explanation is shared ancestry from a common prototype word in an ancestor language.
From these correspondences, linguists reconstruct the proto-language — the ancestral form from which the attested descendants diverged. The linguist posits the simplest sound that, transformed by independently attested processes, could yield all the observed descendants. For the *father* words, the reconstructed Proto-Indo-European form is *\*ph₂tḗr*. The asterisk marks it as a reconstruction, not an attested form. This is the scientific core of the discipline: the reconstruction is not a guess but a hypothesis subject to revision as new evidence appears, in exactly the way that unobserved physical entities are inferred from observable effects. The result is the Indo-European language family — a tree of related languages descending from a single ancestral community that probably lived around 4,000–6,000 years ago.
Your knowledge of morpheme types becomes essential here because sound change and morphological change interact. A suffix that survives intact in Latin may be eroded in French, contracted in English, or repurposed in Russian. Tracking changes requires decomposing words into their morphological parts and examining each part's history separately. Your knowledge of writing systems matters too: written records extend the historical record but also mislead — spelling often lags centuries behind pronunciation, and literacy was restricted to narrow social groups, so written texts may not reflect spoken vernaculars. The language families we can reconstruct most confidently — Indo-European, Semitic, Dravidian — are those with long written records combined with many living descendants speaking related languages, providing multiple lines of convergent evidence.