Proto-Germanic Overview with Example Sentences

We know a lot about Proto-Germanic grammar, but the farther you get from “paradigms and sound laws” and into “style, word order, syntax nuances,” the more it turns into educated guessing.

I’ll break it into:

• What we know pretty solidly
• What’s “good reconstruction but still a model”
• What’s hand-wavy or basically unknown

---

1. Stuff we know very well

These are about as secure as historical linguistics ever gets.

Phonology & sound changes

• Consonant system (Grimm’s Law, Verner’s Law, later Germanic changes).
• Vowel system, quantity, most phonotactics.
• Regular correspondences with Proto-Indo-European (PIE).

That’s the backbone of all reconstruction; it’s about as solid as physics in this context.

Noun genders and most declensions

Yes, we’re very confident about:

• Three genders: masculine, feminine, neuter.
• Number: singular, plural (dual survives in pronouns only).
• Cases: nominative, accusative, genitive, dative (plus traces of instrumental/locative/etc. in fossilized forms).

You can reconstruct neat paradigms like (give or take notation):

• a-stems (masc/neut): **wulfaz, dagaz
• ō-stems (fem): *tungō
• i-stems, u-stems, n-stems (weak), etc.

We have these patterns in Gothic, Old Norse, Old English, Old High German, Old Saxon, etc., and they line up too cleanly to doubt the basic PGmc model.

Pronouns

We have a very good reconstruction for:

• 1st, 2nd, 3rd person
• Singular, dual, plural (dual in 1/2 at least)
• Case marking (nom, acc, gen, dat).

Something like:

• 1sg: *ek / *ik (nom), *mek / *mik (acc)
• 2sg: *þū, *þek / *þik
• 3sg masc: *hē, acc *hin(a)-, etc.

You can see the reflexes directly in Gothic ik, meina; Old Norse ek, mik; Old English ic, me, min; German ich, mich, mein.

Verbal morphology (core system)

We’re also in good shape on:

• Strong vs weak verbs.
• Ablaut classes I—VII (e.g. *bītaną, *bāt, *bitun, *bitanaz).
• The dental preterite (-d- / -t-), i.e. weak verbs.
• Basic personal endings for present and past.
• Infinitive, past participle, simple imperatives.

You can lay out a pretty complete paradigm for a “regular” strong verb and a few regular weak types and it’s not fantasy; it’s grounded in the old languages all pointing in the same direction.

---

2. Things we can reconstruct reasonably, but they’re still models

This is where it’s “highly plausible reconstruction,” not direct evidence.

Word order / basic syntax

We do not have Proto-Germanic texts. We infer syntax from:

• The oldest daughter languages (Gothic, early Norse runic, early West Germanic).
• PIE syntax and common inherited patterns.

The usual consensus:

• Proto-Germanic was probably verb-second (V2) in main clauses and more SOV in subordinate clauses, or some head-final tendency with heavy V2 drift already in place.

• It likely inherited an older PIE SOV / head-final bias, but already had:

  • clause-initial topics
  • finite-verb-second behavior in statements
  • fairly free constituent order due to rich case marking.

We can say:

• “It probably wasn’t rigid SVO like Modern English,”
• “It probably wasn’t rigid SOV either,”
• “It probably had V2-style main clauses.”

But exact syntactic details (scope of V2, placement of negation, adverbs, embedded clause structure) are much less secure than phonology and morphology.

Full verb paradigms

We’re solid on:

• Present vs preterite
• Indicative vs subjunctive (conjunctive)
• Person/number endings in general outline.

We’re less certain on:

• Allomorphy details (exact vowel qualities in every form).
• Distribution of strong vs weak for specific lexemes at the PGmc stage.
• Some edge categories (e.g. the exact shape and productivity of preterite-present verbs at the PGmc stage).

But you can absolutely write a reconstructed paradigm for “*beraną ‘to carry’” that would be recognized as “correct” in the field—just with small uncertainties at the fine details.

Derivational morphology

We have a decent handle on:

• Agent nouns (-eraz, -ariją etc.)
• Abstracts (-īnō, -iþō etc.)
• Adjectival suffixes (-iskaz “-ish”, etc.)

We know the key building blocks. What we don’t always know is:

• How productive each suffix was at the exact PGmc stage.
• Which derivatives existed in the proto-language vs only in a daughter branch.

Still, the set of suffixes and their prototypes is very well anchored.

---

3. The big gaps / speculative territory

This is what you asked explicitly: “Where are our biggest gaps?“

a) Actual usage-level syntax

We don’t know:

• How common certain clause types actually were.
• Exactly how topics, focus, and information structure were handled.
• Typical order of multiple objects, adverbs, particles in a clause.
• How common “scrambling” was vs more canonical orders.

We can guess from Gothic, Old Norse, Old English, but that’s at least several hundred years later and each branch has already gone its own way.

So anything like:

“Proto-Germanic normally placed negation between the subject and verb”

is going to be weak. We’re on safer ground with:

“It likely had a V2-like system and relatively free order because of case morphology.”

b) Lexical semantics and nuance

We can often reconstruct a form and a general meaning, but:

• Shades of meaning,
• Register (formal vs colloquial),
• Connotations, taboo levels,

are mostly lost. E.g. we might know there was a word **slut- “close, shut,” but how often it was used, in what style, and with what idioms? That’s mostly guessing.

c) Idioms, discourse markers, particles

Things like:

• Little filler particles (“well, you know, indeed, just…”),
• Fixed expressions and idioms,
• Formulaic greetings, curses, interjections,

are barely reconstructable. You can sometimes infer something if an idiom looks inherited across multiple branches, but that’s rare and usually arguable.

d) Detailed morphophonology

We sometimes don’t know:

• Exactly how certain sandhi/allophonic rules worked (e.g. real phonetic detail).
• Exact stress patterns in all contexts (though main stress is pretty well understood).
• The timing of certain vowel changes relative to others (fine chronological ordering).

The coarse structure is solid; the delicate chronology is often fuzzy.

e) Functional categories and clausal architecture

In more modern syntactic terms:

• Was there already a full CP-IP-VP-style clausal hierarchy?
• Where exactly did complementizers, negation, and modal particles sit?
• How grammaticalized were modals vs lexical verbs?

Those are theoretically interesting questions, but the evidence is thin; you mostly extrapolate backwards from Gothic + early Norse + early West Germanic + PIE. That turns into “best fit” modeling, not hard reconstruction.

---

4. How confident is “reconstructed Proto-Germanic” really?

Rough confidence levels, bluntly:

• Phoneme inventory + major sound laws: very high confidence.
• Basic noun and verb paradigms (endings, cases, persons): high.
• Pronouns (forms, cases, duals): high.
• Genders and major declension classes: high.
• Derivational morphology inventory: moderate—high.
• Basic syntax (V2 tendency, finite verb placement, free word order): moderate, but not precise.
• Actual word order frequencies & stylistics: low.
• Idioms, discourse, pragmatics: extremely low to nonexistent.
• Exact lexicon size and boundaries: impossible; we know only a fraction.

If someone hands you a “fully written” Proto-Germanic sentence, it’s at best:

• Phonologically + morphologically well-grounded,
• Syntactically and lexically plausible,
• But absolutely not something we know was ever actually uttered in exactly that form.

---

5 Simple Example Sentences in Proto-Germanic

Here are five short Proto-Germanic sentences that stay in the “safe zone”: simple verbs, simple noun phrases, uncontroversial morphology, and structures that show up cleanly in Gothic, Old Norse, and Old English.

For each, I give:

• Proto-Germanic (reconstructed)
• English gloss
• Gothic
• Old Norse
• Old English
• Why the PGmc reconstruction is defensible

These are about as “70%+ safe” as you can get. Anything syntactic beyond this starts sliding into speculation.

1. *Sunnō brennidē.

“The sun burns.”

Gothic: sauil brannida Old Norse: sólin brennr Old English: sēo sunne byrneþ

Explanation:

• *sunnō “sun” (fem ō-stem) is rock-solid; reflected nearly identically everywhere.
• Verb **brennijaną / brennaną (“burn”) shows up as Gothic brannjan, ON brenna, OE brennan.
• Third-person singular present with ending -idē / -iþ is straightforward.

Confidence: very high.

2. *Ik sehwa þuk.

“I see you.”

Gothic: ik saihwa þuk Old Norse: ek sé þik Old English: ic sēo þē

Explanation:

• Pronouns *ik and *þuk (accusative 2sg) are extremely well supported; they reflect directly in Gothic ik, þuk and ON ek, þik.
• Verb *sehwaną (“to see”) → Gothic saihwan, OE seon, ON sjá (via syncope + later sound changes).
• PGmc present ending -ō / -a / -u depending on class; *sehwa is a conservative, generic stand-in following Gothic closely.

Confidence: very high for the morphology, medium for precise verb ending.

3. *Þat lambą ist hweitą.

“The lamb is white.”

Gothic: þata lamb is hweits Old Norse: þat lamb es hvítt Old English: þæt lamb is hwīt

Explanation:

• Neuter noun *lambą clearly reconstructs from Gothic lamb, ON lamb, OE lamb.
• Adjective *hweitaz “white” → Gothic hweits, ON hvítr, OE hwīt.
• Copula *ist is certain.
• Neuter agreement (hweitą) is secure.

Confidence: very high.

4. *Dagaz ist langaz.

“The day is long.”

Gothic: dagis ist laggs Old Norse: dagr er langr Old English: dæg is lang

Explanation:

• *dagaz “day” is an a-stem masculine attested across the board.
• *langaz “long” → Gothic laggs, OE lang, ON langr.
• Basic copular clause; essentially impossible to mess up.

Confidence: extremely high.

5. *Wulfaz runnōþ at hūsą.

“The wolf runs to the house.”

Gothic: wulfs rann at razn Old Norse: ulfr rennr at húsinu Old English: wulf rynnþ tō hūse

Explanation:

• *wulfaz “wolf” is airtight; all branches have this.
• Verb *rinnaną “run” is well reconstructed; ON renna, OE rinnan, Gothic rinnan.
• Preposition *at has direct reflexes in ON at and Gothic at.
• *hūsą “house” is extremely solid (Gothic razn means “house” too but hūs appears elsewhere in WG/Norse).

Confidence: high, with the only “soft spot” being the exact PGmc form of “run” in the present (rinnōþ → safe enough following West/Norse analogues).

What makes these “safe”?

• All nouns and adjectives used here have regular, stable reflexes in Gothic, Old Norse, and Old English.
• All verbs are uncontroversial strong/weak classes with clear PGmc ancestry.
• All pronouns are attested in near-PGmc form in Gothic and early West/Norse Germanic.
• Word order is kept minimal: straight predicate sentences or straightforward verb—object order, which avoids debates about PGmc V2/SOV structures.

5 Complex Example Sentences in Proto-Germanic

Alright, here’s a set of five more complex Proto-Germanic sentences, with parallel translations and a quick sanity check on what’s solid vs. guessed.

Big picture caveat:

• Lexemes, cases, and most pronouns are on pretty firm ground. (Wikipedia)
• Exact finite verb forms (esp. for anything except well-documented models like beraną or sprekaną) and some clause markers (þan “when”, saei “who”) are modeled on Gothic/PGmc paradigms and are partially speculative. (Wiktionary)

1. “When he sees the wolf, he runs to the house.”

Proto-Germanic (reconstructed)

Þan iz wulfą sehwiþ, iz at hūsą rinnidi.

English

When he sees the wolf, he runs to the house.

Gothic (approx.)

þan is saiƕiþ wulf, rinniþ du raznai.

Old Norse

Þá er hann sér úlfinn, hleypur hann til hússins.

Old English

Þonne hē wulf geseoþ, hē rynþ tō hūse.

Why the PGmc is (mostly) plausible

• iz “he” is the regular 3sg personal pronoun. (Wikipedia)

• wulfą is acc.sg of wulfaz “wolf”, a textbook a-stem.

• hūsą “house” is a standard neuter a-stem.

• Temporal þan is modeled on Gothic þan (then/when); using it as a subordinator “when” here is reasonable but not guaranteed. (lrc.la.utexas.edu)

• Verbs:

  • sehwiþ “(he) sees” is an analogical 3sg pres. of sehwaną “see”, shaped on patterns like nimidi (from nemaną) and sprikidi (from sprekaną). (Wikipedia)
  • rinnidi “(he) runs” is likewise modeled as a class-3/4 style present for rinnaną “to run”; the stem and general shape are plausible, the exact vowels/endings are more guessy.

I’d call this sentence high confidence on lexemes and cases, medium on the finite verb shapes, and low-to-medium on the exact subordinator choice.

2. “The man who gives a sword to his brother is good.”

Proto-Germanic

Sa mannaz, saei swerdą brōþrai seinammai gibidi, ist gōdaz.

English

The man who gives a sword to his brother is good.

Gothic (approx.)

sa manna, saei sverd gibaþ brōþr seinamma, ist gōþs.

Old Norse

Sá maðr, er gefr bróður sínum sverð, er góður.

Old English

Sē mann, þe sweord brēðre hisse giefþ, is gōd.

PGmc reasoning

• sa mannaz “the man” uses the demonstrative sa plus mannaz “man” (well supported). (Wikipedia)
• saei as a relative “who” is modeled directly on Gothic saei; PGmc is often reconstructed with something like sa-i, so this is plausible but somewhat Gothic-colored. (Project Gutenberg)
• swerdą “sword” is standard; cf. OE sweord, ON sverð.
• brōþrai dat.sg of brōþēr “brother” follows the usual a-stem/dat.sg pattern by analogy; the exact vocalism in PGmc is a little murky.
• seinammai “to his (own)” = dat.sg of seinaz “his (reflexive)” formatted after the dative of sa / þammai in pronoun tables — structurally sound but the exact shape is somewhat extrapolated. (Wikipedia)
• gibidi 3sg pres. of gebaną “give” is modeled on the class-5 strong conjugation like beraną / nimaną (biridi / nimidi), and Wiktionary-style reconstructions of gebaną. (Wiktionary)
• gōdaz “good” is a bog-standard adjective with solid reflexes everywhere.

So: very solid lexically and morphologically, with the main soft spots being the dative forms and the relative particle.

3. “We heard that the king spoke to his people.”

Proto-Germanic

Wīz hausidum þat sa kuningaz sprak þiudim seinim.

English

We heard that the king spoke to his people.

Gothic (approx.)

weis hausidēdum þatei sa kuniggs sprak du þiudam seinam.

Old Norse

Vér heyrðum at konungrinn talaði til lýðs síns.

Old English

Wē gehīerdon þæt se cyning spræc tō his folce.

PGmc reasoning

• Wīz “we” is the regular 1pl nominative pronoun. (Wikipedia)

• hausjaną “to hear” is a class-1 weak verb; Gothic hausjan is directly attested. (germanic.ge)

  • Preterite 1pl hausidum: stem hausid- + 1pl -um is exactly what you expect for weak I preterite morphology.

• þat as a complementizer “that” is straightforwardly based on the neuter demonstrative þat used this way in Gothic and later Germanic. (Wikipedia)

• sa kuningaz “the king” is the well-supported kuningaz “king” with demonstrative. (Wiktionary)

• sprak is the sg preterite of sprekaną “speak”, directly matching the reconstructed strong-5 pattern (sprekaną / sprak / sprēk- etc.). (Kaikki)

• þiudim “to (his) people” is meant as a dat.pl of þiudō “people, nation”, modeled after typical dat.pl -im / -um endings; that exact shape is somewhat speculative.

• seinim dat.pl (agreeing with þiudim) is analogical to pronoun/adjective dat.pl endings; again, structure fine, exact form fuzzy.

Here, core lexemes and the weak-verb preterite are solid; the plural dative forms and the exact complementizer behavior are more reconstructive “best guesses.”

4. “You (pl.), if you hear my words, (will) live.”

Proto-Germanic

Jūz, jabai wōrdō meinō hausjiþ, libiþ.

English

You (pl.), if you hear my words, live. (i.e. “you will live” — PGmc is using present for future here.)

Gothic (approx.)

jūs, jabai waurda meina hausjiþ, libiþ.

Old Norse

Þér, ef þér heyrið orð mín, lifið.

Old English

Gē, gif gē mēne word gehȳraþ, libbaþ.

PGmc reasoning

• Jūz “you (pl. nom.)” and meinō “my (gen.sg.)” come straight from personal-pronoun paradigms plus possessive formation. (Wikipedia)
• jabai “if” is directly inherited from Gothic jabai and commonly reconstructed for PGmc. (lrc.la.utexas.edu)
• wōrdą “word” → nom/acc pl wōrdō is the usual neuter plural pattern.
• hausjiþ is a 2pl present of hausjaną: stem hausj- + -iþ from PGmc present endings. That’s exactly how weak I presents are reconstructed. (germanic.ge)
• libiþ is 2pl present of a weak verb libjaną “to live” (cf. Goth. liban, ON lifa, OE libban); the stem shape is fine, the exact PGmc weak class and ending here (-iþ) is analogical to hausjiþ.

Semantics (present used with conditional for future meaning) is directly paralleled in Gothic and early Germanic, so the structure of the sentence is quite plausible even if the precise vowel qualities in the verbs are approximate.

5. “I want us to carry the child home.”

Proto-Germanic

Ik wiljō þat wīz barną heimą beraim.

English

I want us to carry the child home.

Gothic (approx.)

ik wiljau þatei weis barn du haimai bairam.

Old Norse

Ek vil at vér berim barnið heim.

Old English

Ic wille þæt wē bern þone hām beron.

PGmc reasoning

• Ik “I” and wīz “we” are straight from the PGmc pronoun system. (Wikipedia)

• wiljō “I want” is 1sg present of wiljaną “to want, to will”; this verb is well-reconstructed and underlies Gothic wiljan, OE willan, ON vilja.

• þat as a complementizer again mirrors Gothic usage of þatei “that” < þat + ei. (lrc.la.utexas.edu)

• barną “child” is the standard neuter noun (cf. OE bearn, ON barn).

• heimą “home(wards)” is a fairly secure locative/goal noun/adverb; it has stable reflexes across Germanic.

• beraim: this is deliberately modeled on the 1pl present subjunctive of beraną “to bear, carry”, from published reconstructions (1pl pres. subj. beraim). (Wiktionary)

  • Using a subjunctive here after a volitional verb “want” fits both PIE and early Germanic syntax reasonably well.

This one leans a bit heavier on reconstructed verbal mood (subjunctive) and on the syntax of “want + that-clause,” but all of the pieces (lexemes, endings, and structure) are supported by comparative PGmc grammars.

How “real” are these?

If you treat these as:

• phonology + morphology demos based on reconstructed paradigms and attested daughter forms: they’re in the safe zone.
• actual sentences that a Proto-Germanic speaker definitely uttered: obviously not; these are stitched together from best-fit pieces.

If you want to crank this one notch more “serious”, next step would be: pick one of these sentences and we can:

• cross-check every single morpheme against a PGmc grammar/lexicon,
• and mark each piece “very likely / plausible / speculative” more granularly.