Latin Genealogy Terms: A Working Glossary of the Words That Actually Appear in Parish Registers

Most Latin in old church records is formula, not prose — a few dozen recurring words doing the same jobs in the same order, entry after entry. This is a working glossary of the Latin genealogy terms you will actually meet in baptism, marriage, and burial registers, plus the abbreviation marks and case endings that make them hard to read. Learn the pattern and the page starts to open, even if you never learned the language.

A baptism, a marriage, or a burial entry recycles the same small vocabulary — filius, baptizatus est, legitimus, eodem die, sepultus, vidua — with only names, dates, and occasional details changing from line to line. Learn the skeleton of each entry type, learn the marks the clerk used to compress it, and learn that Latin nouns change their endings, and you can read most entries without knowing Latin. The words below are grouped by the job they do in the entry.

Why Latin is in your parish register at all

Since the Council of Trent (1545–1563), Catholic parish priests have been required to keep records of certain sacraments in bound register books — baptisms or christenings, marriages, deaths or burials, and confirmations, as BYU's Script Tutorials summarize. Session XXIV of the Council put the requirement plainly for marriages: the parish priest is to keep a book recording the names of those married, the witnesses, and the day. The first official Rituale Romanum followed under Paul V on 17 June 1614, which pushed registers further toward a shared shape — though it does not follow that every later register used one identical template. Local practice persisted everywhere.

That is why your entries look so alike. It is also why a small glossary goes a long way.

The chronology of when Latin gives way to the vernacular is stubbornly local. In England, The National Archives dates the Latin of official documents from 1086 to 1733, after which official records were written in English — a firm English milestone, not a European one. Catholic parish records in many countries were kept in Latin into the twentieth century, and in some places Lutheran and Anglican registers were kept in Latin too. In France, François I's edict of 1539 required parish registration of births and deaths, though a registration requirement is not by itself a language switch. There is no reliable country-by-country table for Germany, the Netherlands, Italy, Spain, Poland, or Hungary. Expect to establish the transition for your own parish from the volumes themselves.

Fix one distinction before going further, because it causes real confusion: Latin the language is not the same thing as Latin the script. Your German, French, or English entries are written in the Latin alphabet without being Latin. A Latin entry, meanwhile, can be written in any number of hands, from a clean italic to a cramped Gothic cursive. If you cannot read the page, the problem may be the hand rather than the language — a separate skill, and one worth diagnosing before you reach for a word list. The broader craft of transcribing parish registers, wills, and census returns for family research sits on top of that distinction.

Kinship and relationship terms

These carry the genealogy. Everything else in the entry is scaffolding.

• filius / filia — son / daughter. The word that carries the most weight on a baptism page.
• pater, mater — father, mother. parentes or genitores — parents.
• uxor — wife; uxoris — of the wife; commonly abbreviated ux.
• maritus, coniux (also written conjux) — husband, spouse.
• sponsus / sponsa — groom / bride; sponsi / sponsae in their declined forms.
• patrini — sponsors or godparents; matrina — godmother, where the register uses the term.
• testes — witnesses.
• viduus / vidua — widowed man / widow. relictus / relicta — the surviving widower / widow of a named person.
• caelebs — unmarried.

Note the paired -us / -a endings running through this list. That vowel is doing work: masculine and feminine. Filius legitimus is a legitimate son; filia legitima is a legitimate daughter. If your register abbreviates to fili(us/a) or writes the ending as a superscript, that final letter is the sex of the child, and it is worth reading twice.

Event formulae: the sentences the clerk repeated

Latin registers work by formula, and the formula is what you should learn rather than the individual words. A Belgian baptism worksheet from BYU's family history library gives the pattern in the bare form the clerk actually used: [Eodem/ejusdem/eiusdem] die ______ baptizat(us/a) est. fili(us/a) legitim(us/a) — "on the same day … was baptized … legitimate son/daughter."

Learn that shape and each baptism entry becomes a fill-in-the-blank exercise.

Baptism

natus / nata — born. baptizatus / baptizata est — was baptized; frequently shortened to baptiz., bapt., or bapta. legitimus / legitima — legitimate; illegitimus / illegitima its negation. spurius / spuria and nothus / notha also appear as local terms for illegitimacy, but treat them as local usage to confirm on your page rather than universal labels.

Marriage

matrimonium (often matr.), nuptiae — marriage, nuptials. Marriage formulae vary more than baptism formulae. Read the verb your parish used off the page rather than assuming it from a word list.

Burial

sepultus / sepulta — buried; sepult(us/a) in cœmeterio — buried in the cemetery. defunctus / defuncta — deceased. obiit — he or she died.

BYU's guide to Latin burial records notes that a burial entry may carry the burial date, the deceased's names, marital status, spouse, age, occupation, parents' residences or birthplaces, masses, a will, witnesses, and the officiating priest — but it warns explicitly that which fields appear depends on the period, and that the order varies. Causes of death and occupations in particular need a local sub-glossary built from your own parish's volumes. No general word list will cover them.

Dates and framing words

• eodem die — on the same day. eodem anno — in the same year. ejusdem / eiusdem — of the same.
• die — day. mensis — month. Anno Domini, abbreviated A.D. — in the year of the Lord.

Eodem die is the workhorse and a small trap: it points back to the previous entry, so an entry containing it has no date of its own. You must read upward until you find one. Ordinal day numbers, month names in Latin, and feast-day or Sunday dating formulas all need a dated, region-appropriate calendar to convert — a separate research task, not a glossary lookup, and the movable feasts require care.

The abbreviations: marks, not words

This is where most readers stall, and it is not a vocabulary problem. As BYU's abbreviation guide puts it, the common use of abbreviations is one of the most complicating factors in reading old handwritten records. Clerks abbreviated to save space, materials, and time, and they did it by a small set of mechanisms:

• Suspension — the end of the word is cut off. Baptiz. for baptizatus.
• Contraction — first and last letters written, middle removed. dn̄s for dominus; hđe for herede.
• Superposition / superscript — a raised letter carries the ending. Mã for Maria; bapta for baptizata; Joes for Joannes.
• Conventional signs — symbols standing for whole words or syllables, such as q̄q̄ for quoque.
• The nasal stroke — a bar over a vowel. TEI's guidance on levels of transcription notes that this superscript stroke has a specific graphemic reference: it stands for m or n. So añ is ann-, and a bar over the o in a name may be restoring a missing n.

The habit that matters: an abbreviation does not have one context-free answer. The same short form can resolve differently depending on the formula it sits in and the conventions of that particular clerk. Build the resolution from the surrounding entry and from neighbouring pages in the same hand — the register itself is your best evidence. For the fuller mechanics across languages, our guide to manuscript abbreviations and ligatures works through the same repertoire as it appears in vernacular hands, since these marks travel out of Latin and into English, French, and German documents unchanged.

For deeper reference, Cappelli's Elements of Abbreviation in Medieval Latin Paleography remains the standard treatment of suspension, contraction, and the titulus; an English presentation of his prefatory treatise is freely available. Bear in mind that Cappelli's classifications are grounded in medieval manuscript practice, and how well they cover early-modern parish hands is a question to settle against your own register.

Names: declension is not a second person

The most common error in genealogical Latin is not mistranslation. It is inventing a person.

Latin nouns change endings by grammatical case, and names are nouns. Filius Iohannis means a son whose father is named Iohannes — John. Iohannis is the genitive, "of John." The proper term is genitive, and it is the same man as Iohannes two lines above. Mariae is Maria in an oblique case, not a woman named Mariae. Record the dictionary form in your notes, and record what the page says alongside it.

The second trap is that Latinized given names are not literal translations, and one Latin form can map to several vernacular names. FamilySearch's list of Latin versions of English given names gives Gulielmus = William, Guido = Guy, Hadrianus = Adrian, and Helena = Ellen, Helen, or Eleanor. That last one matters: the choice among the three is a judgment about the family, not a fact about the Latin. Surnames add a further layer, since practice on Latinising them varied by region and clerk. Where you must choose, note the choice as a choice.

For general vocabulary beyond the register formulae, FamilySearch's Latin Genealogical Word List covers grammar orientation, date phrases, and a working Latin-to-English list. The National Archives' Stage 1 Latin tutorial teaches the language from real documents, which is the right way to learn it. Neither replaces a dictionary — Lewis & Short for classical senses, Du Cange and the DMLBS for medieval and ecclesiastical ones — and none of them replaces the image of your own page.

Two layers, kept apart

The method that survives scrutiny is layered, and the layers must not be collapsed.

Layer one: transcribe what is on the page

Keep the abbreviation as written. Keep the bar, the superscript, the case ending, the archaic spelling. This is the diplomatic layer; TEI's guidance is that in a strict diplomatic transcription, abbreviations are left unexpanded and apparent slips of the pen stay uncorrected.

Layer two: expand, resolve, and translate — visibly

Mark expansions so they can be told apart from what the clerk wrote, for instance baptizat[us]. Resolve inflected names to their dictionary form in a separate note. Translate last, and keep the translation separate from the transcription.

Kept in that order, every reading remains recoverable and every judgment stays auditable a decade later, when you have forgotten why you decided Helena was Eleanor. Collapsed into one step, you end up with a smooth English paraphrase and no way back to the page. That discipline is the substance of faithful genealogy record transcription, and it is what makes a family tree defensible rather than merely plausible.

It is also why machine translation deserves a specific kind of caution here. Latin machine translation has genuinely improved — Volk's 2024 evaluation reports LLM-based Latin translation clearly superior to earlier approaches, and Fischer's 2025 work describes state-of-the-art results on historical Latin. But those studies evaluate normalized, already-transcribed Latin, not handwritten parish Latin thick with local suspensions and superscripts, and a 2025 review of machine translation in the LLM era records the persistent failure pattern: fluency prioritized over faithfulness, with dropped content and hallucinations. A model that produces confident English from an abbreviated entry it could not actually read is the worst result available, because the output looks finished. Use a translation as a candidate reading to compare against your own. Never as the evidentiary text.

Where a transcription tool fits — and where it doesn't

If you are working through a handful of entries, a word list and patience are enough. If you have photographed two hundred pages of a parish register and the wall is the handwriting rather than the Latin, the constraint changes: you need a first-pass reading of every page before you can even see which entries matter.

That is the stage Leo is built for. Its transcription model, ATR-1, reads Latin-script material — any language written in the Latin alphabet, which includes Latin itself but equally the English, French, German, Dutch, Spanish, and Italian your later registers switch into. It is trained to transcribe what is on the page rather than to smooth it: the abbreviation stays an abbreviation, the superscript ending is not silently resolved, Iohannis is not quietly regularized to Iohannes. That is the property worth insisting on, because a machine that expands baptizat(us/a) for you has made a genealogical decision on your behalf without telling you. Errors of the recoverable kind — a wrong character you catch against the image displayed beside the text — are a different order of problem from a fluent paraphrase you cannot audit. Translation, when you want it, is a separate step that writes to its own tab, leaving the base transcription untouched. Transcription and translation stay two jobs, as they should.

Two honest limits. Registers laid out as dense pre-printed tabular forms — printed column headers with handwriting crammed between them — are the weakest case for machine reading, and paragraph-format entries fare better. And no transcription, machine or human, removes the need to check names, dates, and relationship words against the image yourself. Those are the fields that carry your tree.

Building the glossary that matters: your parish's

No published list will match your register. The one that will is the one you build from it: a page of your own, recording each abbreviation as the clerk wrote it, your expansion, the entry it came from, and how confident you are. Add the local cause-of-death and occupation terms as you meet them. Add the way this particular hand writes baptizatus. Within a few dozen entries, the formula will be doing the reading for you, and you will only be checking the names — which is where your attention belonged all along.

Frequently Asked Questions

What are the most common Latin genealogy terms in parish registers?

The recurring Latin genealogy terms are few enough to learn in an afternoon. Kinship: filius / filia (son / daughter), pater and mater, uxor (wife, often abbreviated ux.), maritus and coniux, sponsus / sponsa (groom / bride), patrini (godparents), testes (witnesses), viduus / vidua (widower / widow), relictus / relicta (surviving spouse of a named person). Events: natus / nata (born), baptizatus est (was baptized), legitimus / legitima, matrimonium and nuptiae, sepultus / sepulta (buried), defunctus, obiit (died). Framing: eodem die, eodem anno, die, mensis, Anno Domini.

Why is my ancestor's parish register written in Latin?

Because the Council of Trent (1545–1563) required Catholic parish priests to keep bound registers of baptisms, marriages, deaths or burials, and confirmations, and the first official Rituale Romanum under Paul V in 1614 pushed those books toward a shared shape. Latin was the language of that record-keeping in many Catholic parishes into the twentieth century, and in some places Lutheran and Anglican registers used it too. When Latin gives way to the vernacular is stubbornly local — in England, official documents were in Latin until 1733 — so expect to establish the transition for your own parish from the volumes themselves.

What does filius Iohannis mean, and why do Latin names keep changing spelling?

Filius Iohannis means "son of John" — Iohannis is the genitive case of Iohannes, not a different man. Latin nouns change their endings by grammatical case, and names are nouns, so the same person appears as Iohannes in one line and Iohannis in the next. Mariae is Maria in an oblique case, not a woman called Mariae. The most common error in genealogical Latin is not mistranslation but inventing a person out of a case ending. Record the dictionary form in your notes, and record what the page actually says alongside it.

What does the bar over a letter in an old Latin record mean?

A bar over a vowel is a nasal stroke, and it stands for a missing m or n — so añ is ann-, and a bar over the o in a name may be restoring an n. It is one of several compression mechanisms clerks used: suspension (the word's end cut off, baptiz. for baptizatus), contraction (first and last letters kept, dn̄s for dominus), superscript endings (bapta for baptizata, Joes for Joannes), and conventional signs for whole words. No abbreviation has one context-free answer; resolve it from the surrounding formula and from neighbouring pages in the same hand.

Can I use machine translation to read a Latin baptism entry?

Treat a machine translation as a candidate reading, never as your evidentiary text. Latin machine translation has genuinely improved, but published evaluations test normalized, already-transcribed Latin — not handwritten parish Latin thick with local suspensions and superscript endings. The documented failure pattern is fluency prioritized over faithfulness, with dropped content and hallucinated detail. A model that produces confident English from an entry it could not actually read is the worst outcome, because the output looks finished. Work in two layers instead: transcribe what is on the page first, keeping abbreviations and case endings intact, then expand and translate as a visibly separate step.