Para-Indo-European: positions taken and liberties stated
Para-Indo-European is a para-language: a speakable variety designed to sit inside the dialect continuum that Proto-Indo-European reconstructions describe, close enough that it would plausibly have been understood, with every question the scholarship leaves open answered by one stated, consistent choice. It is not a claim about what was actually spoken. This page lists those choices. Where linguists disagree, Para-Indo-European takes one side and keeps it everywhere; where it goes beyond what can be reconstructed, it says so here.
Every word in the lexicon carries one of three marks (the translator and the course show it on every word):
| mark | meaning |
|---|---|
| attested | a root reconstructed from cognates in the daughter languages |
| derived | an ordinary Proto-Indo-European derivation from such a root, not itself attested |
| coined | built for something its speakers had no word for, as a transparent compound of attested roots (car = "self-wagon") |
Sounds
| question | its position | the main alternative |
|---|---|---|
| How many laryngeals? | Three, *h₁ *h₂ *h₃ | One, or four |
| How were they pronounced? | h₁ as [h], h₂ as [χ], h₃ as [χʷ], written h x f (f for the labialised one, as q and v are the labialised stops; xw is h₂ + w). A choice, made so that everything can be spoken | h₁ a glottal stop; h₃ voiced |
| What do they do to vowels? | h₂ colours a neighbouring *e to *a, h₃ to *o; a vowel before a laryngeal and a consonant lengthens | (mainstream) |
| A laryngeal between consonants | Written and pronounced as its own syllable, with a short schwa: pxtéer [pəχˈteːr] for *ph₂tḗr, the vowel Greek patḗr, Latin pater and Sanskrit pitā́ continue in three different colours. Beside a syllabic r l m n it lengthens that syllable instead (xwlhnáx for *h₂wl̥h₁néh₂, Sanskrit ūrṇā). The same goes for a laryngeal before a consonant at the start of a word (xstéres, Greek astéres) | Not pronounced (rejected: the daughters' vowels show it was) |
| Voiceless aspirates | None of their own: the old *kʰ *tʰ are stop + h₂ (kakh₂- "laugh", kʷenth₂- "grief"), as laryngeal theory derives Sanskrit kh th | Separate voiceless aspirate phonemes |
| The stop system | Voiceless, voiced, and breathy-voiced (bh dh gh), spoken with a murmured release | The glottalic theory (ejectives instead of plain voiced) |
| Dorsal consonants | Three series: palatal c j jh, plain k g gh, labiovelar q v vh | Two series |
| The vowel *a | Mostly the product of h₂ colouring; kept in the few roots usually reconstructed with it (*magʰ- "can", *kap- in "head", *lakʷ- "lake", *ǵlakt- "milk") | No independent *a at all |
| The accent | Written with an acute on the accented vowel and spoken as stress | A pitch accent, as in Vedic and Greek |
| Syllabic r l m n | Unmarked: the syllable already says where they are, and the voice says them syllabic | Written with a ring below |
Grammar
| question | position | stated liberty? |
|---|---|---|
| Instrumental singular | -eh₁ on consonant stems (Sanskrit padā́ "with the foot"), -h₁ after a stem vowel | Bare -h₁ everywhere |
| Cases | Eight: nominative, accusative, genitive, dative, locative, instrumental, ablative, vocative; ablative and genitive share a form outside the o-stems | no (mainstream) |
| Number | Singular and plural only | Yes: the dual is dropped, for learnability |
| Neuters | Nominative and accusative are the bare stem (*ḱḗr "heart", *h₁nómn̥ "name", *móri "sea"); the plural is the collective in -h₂, and a neuter plural subject takes a singular verb | no (mainstream) |
| The nominative -s | Every masculine and feminine consonant stem takes -s in the nominative, m-stems and laryngeal stems included (earth dhéjhoms, house dóms, woman véns), where Proto-Indo-European had bare lengthened forms (*dʰéǵʰōm, *dṓm, *gʷénh₂) | Yes, for one learnable rule; s-stems (month méens) keep their bare form |
| Gender | Masculine, feminine, neuter | no |
| Ablaut and accent mobility | Most stems keep one grade and a fixed accent; strong and weak stems alternate in the consonant stems, and a few nouns keep a mobile accent | Yes: simplified, for learnability |
| Numerals | 1 to 10 attested; teens are unit + déḱm̥t (11 = óynodecmt); decades are unit + h₂ḱm̥t (30 = tríicmt); 20 is the old dual wícmtih; 100 cmtóm; 1000 jhéslom; larger numbers largest first (24 = wícmtih qetwóres) | Yes: 1 to 4 do not decline, and the ordering of large numbers is a stated choice |
| Verb aspect | Present, imperfect, aorist, perfect; no tense of its own | no |
| Moods | Indicative, subjunctive, optative, imperative; a mediopassive throughout | no |
| The future | The subjunctive ("will" in English) | Mainstream about where daughters got their futures; using it as THE future is a choice |
| Modal verbs | "Should, may" are the optative of the main verb; *magʰ- "can" and *welh₁- "want" are real verbs | Partly a choice |
| The infinitive | A verbal noun in -dʰyōy (as in Vedic -dhyai), used where English has "to" | Yes: Proto-Indo-European had no settled infinitive |
| Subject pronouns | Dropped unless they do work (contrast, coordination, gender in the third person): about 8% of clauses | Measured against Spanish, Italian and Latin translating the same sentences |
| Articles | None | no |
| Word order | Verb-final, questions included (a yes/no question is marked by intonation, not by fronting the verb, as in Vedic and Hittite); adpositions follow their noun | Yes, partly: the adpositions are fixed as postpositions; in Proto-Indo-European they were still freer adverbs |
| Possession | "X has Y" is "to X is Y": the possessor in the dative, *h₁es- agreeing with the possessed (Latin mihi est, Vedic, Hittite); *seǵʰ- means "hold, keep" | no (mainstream) |
| Causatives | "Make X do" is the *-éye- causative on the o-grade root (*bʰor-éye- "make carry", Vedic bhāráyati; *rowd-éye- "make weep", rodáyati), with the causee in the accusative | no |
| Comparison | The thing compared stands in the ablative, with no word for "than" | no |
| Pronoun ablatives | *med, *twed, *n̥smed, *usmed, *tosmōd, *toybʰyos (Vedic mát, tvát, asmát, yuṣmát, tásmāt, tébhyas) | no |
| "And", "or" | The enclitics *-kʷe and *-we, attached to the word they follow | no |
| Voice | Active and middle, with the middle also serving as the passive. Verbs that were middle-only in the daughters are middle-only here, in every mood, whatever the sentence asks for: follow (*sekʷ-, Skt sácate, Gk hépomai, Lat sequor), die (*mer-, Skt mriyáte, Lat morior), think (*men-, Skt mányate), seem (*bʰh₂-n-ye-, Gk phaínomai), fear (*bʰeyh₂-, Skt bháyate), wake (*bʰewdʰ-, Skt búdhyate), lie down (*legʰ-, Gk lékhetai), return home (*nes-, Skt násate, Gk néomai), rise (*h₃er-, Gk órnumai, Lat orior), enjoy (*bʰrewg-, Lat fruor), wear (*wes-, Skt váste, Hitt wešta), choose (*ǵews-, Skt juṣáte, Gk geúomai), pray (*h₁wegʷʰ-, Skt óhate, Gk eúkhomai), happen (*kʷel-, Gk pélomai), deny (*h₂er-n-, Gk arnéomai), divide (*deh₂-, Skt dáyate, Gk daíomai), forget (*mers-, Skt mṛṣyate), and their compounds. The middle subjunctive and optative take the middle endings on the mood stem (Vedic -ai, -īta); the 2sg imperative is *-so (Vedic -sva, Greek -so). The perfect keeps its own endings, which are the old middle set | Verbs whose witnesses split (say, speak, see, sit, sleep, want, turn, remember) stay active |
Orthography
The spelling is the author's own, from their rendition of Schleicher's fable, and stays regular so it can be read aloud without a key. It is phonemic: every form the engine makes is read back from its spelling into the sounds it was built from (mt/phonemic_check.py, over 30,886 forms). Two things it once left out are now written: a laryngeal between consonants, which was pronounced, and h₃, which is f so that xw can only be h₂ + w. A vowel before a laryngeal and a consonant is long and the laryngeal silent (viiwós), so that one is not written. A breathy stop before another breathy stop writes its h once (djhémoon): the cluster is released once.
Vocabulary beyond the reconstructions
Modern things are named with transparent compounds of attested roots, never borrowings, on one fixed set of heads and modifiers: vehicles on "wagon", buildings on "house", devices on the instrument suffix, everything electric on "power". A word the reconstructions do not supply is, in order of preference: an attested root, a regular derivation from one, a coinage on that recipe.
How the lexicon was checked
Every word marked attested, in the course and in the translator's common-word layer, was checked against the standard references (LIV², Mallory & Adams, NIL, and the etymological dictionaries of the daughter branches) for three things: that the root is real, that it means what Para-Indo-European uses it for, and that its form is right. Every proposed change was then checked again independently before it was applied; where the references disagree, the existing form stayed. Words whose attestation turned out thin were marked derived or coined instead of being kept as attested. Each noun's declension class and gender were checked the same way. Every change is recorded with its reason, and the course shows the reason in a word's popup.
The course sentences, adapted from the Proto-Germanic course, were checked for Germanic phrasing and corrected (possession, questions, causatives, comparison, bare cases, plural agreement).