Northian grammar

Northian grammar is highly synthetic and fusional. This page aims to cover some of the more technical and historical points regarding Northian garmmar, specifically that of its oldest form, Early Galic Northian. The coverage will take a systemic, bird's eye view for the most part, relegating specific conjugational and declensional paradigms on appendical pages Northian nominals and Northian verbs.

Northian grammar, particularly in nouns, has been important to the reconstruction of Proto-Erani-Eracuran owing to its conservativeness. Though the Galic corpus is hardly large, its 12,000 or so words have been endorsed by historical linguists as a trove of relics that are either unique or corroborating forms for unique items elsewhere. As C. Cloverdale said, "Northian Gales are valued in this science for their fidelity in transmission and consistency in grammar." However, the outward conservativeness of Northian is attributed to the early date of its compositions, where archaic formations are expected, and its exceptional position in the field owes mainly to the fidelity of the transmission that has prevented the loss of relics.

Ablaut

Ablaut is a system of vowel apophony, altering the quality or quantity of vowels but not the meaning of the morpheme in which they are located, that is inherited from Proto-Erani-Eracuran. It affects most classes of words in Northian.

Though ablaut was a regular process closely associated with accent in the reconstructed proto-language, with some authorities proposing a direct correspondence between the accent and the full-grade *e vowel, by Galic times any precise corresondence had been lost (probably already by the final stage of the proto-language). Moreover, existing ablaut formulae have been disrupted by sound change and both general and sporadic analogical replacement. The result that surfaces in Galic Northian is a rich yet unpredictable plethora of alternate morphologies that confuse even later Hamruvunts, whose theses about correct grammar, when such a discipline arose, are sometimes woefully misguided by modern standards.

In nouns, there are four main ablaut patterns inherited and evolved, which are in scholarly discourse termed acrostatic, proterokinetic, amphikinetic, and hysterokinetic. The medieval Northian grammarians astutely observed that the position of the accent in the dative singular predicts the correct set of endings: where it was on the final syllable (oxytone or OX), full-grade endings (e.g. -ṓ, -aí) was used in the oblique cases, and where not on the final syllable (paroxytone or PX), the zero-grade set of endings (e.g. -i) were used. The former situation regularly developed from amphikinetic and hysterokinetic patterns, and the latter from the acrostatic and proterokinetic ones. The OX nouns were characterized by the nearly-universal genitive singular ending -ṓ, while the PX nouns had unpredictable endings there owing to the vagaries of sound change.

The medieval grammarians were not able to distinguish between the proterokinetic and acrostatic ablaut patterns because the latter were quite few and subject to the heaviest erosion in identifiable morphs and thus taught by rote. Surprisingly, Himinastainas observed that such "irregularities" arose mostly in body parts and the commonest animals and objects, so learning them by rote "is imperative". Amphikinetic and hysterokinetic nouns were not distinguished, on the other hand, because they differed principally in the nominative singular, which, synchronically, was largely irregular and must be learned by rote anyway.

Accent		Root	Suffix	Ending	Pre-form	Root	Suffix	Ending	Outcome	Galic
Acrostatic	nom	ó	Ø	Ø	*nókʷ-t-s	ó	Ø	Ø	noxṯ	Paroxytone
	acc				*nókʷ-t-m̥				noxtam
	gen	é	Ø	Ø	*nékʷ-t-s				noxš
Proterokinetic	nom	é	Ø	Ø	*mér-tis	é, Ø	Ø	Ø	mərətiš
	acc				*mér-tim				mərətim
	gen	Ø	é	Ø	*mr̥-téi̯-s	Ø	é	Ø	mərətaiš
Amphikinetic-I	nom	é	ō	Ø	*léy-mō	é	ō	Ø	láimō	Oxytone
	acc	é	o	Ø	*léy-mon-m̥	e	ṓ, ó, Ø	Ø	limṓnąm
	gen	Ø	Ø	é	*li-mn-és	Ø	Ø	ó	limnō
Amphikinetic-II	nom	é	Ø	Ø	*kré-tu-s	Ø	Ø	Ø	xarətuš
	acc	Ø	é	Ø	*kr̥-téw-m̥	Ø	Ø	Ø	xarətā̊
	gen	Ø	Ø	é	*kr̥-tu-és	Ø	Ø	ó	xarəźuuō
Hysterokinetic	nom	Ø	é	Ø	*ph₂-tḗr	Ø	é	Ø	ptō
	acc				*ph₂-tér-m̥				piterəm
	gen	Ø	Ø	é	*ph₂-tr-és	Ø	Ø	ó	fϑrō

In the most summary way, the acrostatic nouns had a persistent accent on the root syllable, while the proterokinetic ones shifted the accent one syllable to the right (namely to the suffix) in the oblique cases. The amphikinetic nouns were of two types, differing only in the nominative singular: one type, the more common, had a lengthened o-grade in the suffix, while the other one had the zero grade there. In both, the root was accented in the nominative singular, and the ending in the oblique cases; the accent of the accusative is disputed. The sources of both the unaccented o-grade and its length are also disputed. The hysterokinetic nouns had an accented suffix in the direct cases and accented ending in the oblique.

In their evolution to Galic Northian, the following changes have occurred. For amphikinetic nouns with an o-grade suffix, the nominative stem extended to the accusative; this must have been a fairly late alteration as the accusative suffix often has the long vowel of the nominative, showing that the long vowel was no longer analyzable as a full grade plus a lengthening element specific to the nominative, i.e. out of place in the accusative. There are also sporadic appearances of a short accusative suffix, which must be interpreted as the result of an earlier layer of levelling.

Amphikinetic nouns with the zero-grade suffix had irregular developments. Where the stem ended in -u or -i, the zero-grade suffix usually spread to accusative; perhaps this occurred under the influence of the proterokinetic, as with them these amphikinetics shared a zero-grade suffix in the nominative, cp. amphikinetic xrétuš "will" and proterokinetic xrétuš "powerful". The vocalized approximant was re-analyzed as a full-grade vowel, wherewith the zero-grade root was introduced, e.g. xarətuš < *kr̥tus. Accusatives ending in -ā̊ must be the outcome of the short suffix introduced from a full-grade nominative plus accusative ending -m, i.e. *-VR-m̥. It is impossible to distinguish the o-grade from the e-grade here, but where there was no o-grade in the paradigm, an original e-grade is assumed, cp. ziiā̊ "deum" from ziiaōš "deus".

Where the stem did not end in an approximant (-r, -s, -m, -n), the development was largely arbitrary, e.g. acc. sing. both more common xmąm (formed from the oblique stem xm- plus acc. ending -am) and (clearly) more archaic xā̊ < *dʰǵʰḗm < dʰǵʰem-m̥.

Proterokinetic nouns had their root syllables levelled nearly completely in favour of the zero grade, unless this produced an impermissible sequence of consonants, but the accent position is usually not altered and often discloses a former full grade. There are sporadic survivals of the full-grade root, but none in the productive suffixes of -ti- and -tu-, as accented zero-grade suffixes are universal there, perhaps also under the influence of the hysterokinetics. This produced a morphologically proterokinetic but accentually static pattern that became dominant for these suffixes in later Northian. The hysterokinetic declension survives particularly well in Northian, preserving a distinct full-grade in the accusative and a zero-grade in the oblique cases.

It is notable that, perhaps owing to a lack of ablaut in the root syllable, the hysterokinetic pattern was the most stable and productive (the acrostatic pattern is assumed to be vestigial even in the proto-language). The patterns with root ablaut, namely amphikinetic and proterokinetic, either lost productivity or were levelled to remove ablaut in the root. Many suffixes which have original amphikinetic patterns developed hysterokinetic compounds which later became productive, while the amphikinetic suffix lost productivity. Issinar asserted in 1940 that, taking the root and suffix together as a unit and allowing for the long-grade in the nominative, Northian nouns could have a maximum of two distinct stems; this rule appears to hold in many cases, though not without exception.

Ablaut rules developed differently in many suffix classes. For example, the mn-stems usually had the long suffix original to the nominative intruding the accusative, but the wn-stems often has a zero-grade suffix in the accusative; possibly this is because the combination C-wn- was vocalized as C-un- early, and the -u- there was thought of as an original, full-grade vowel, after the u-stems. Additionally, many words show compounds of suffixes, particularly derived feminines in -ī and -ū, and the accentual rules in these words sometimes depends on the character of the root, compulsorily it is verbal, and in other times becomes static on the suffix. Some classes of words apparently had vacillating accent or even no recorded accent, suggesting that these derivations were not common enough to have a widely-recognized pattern even if their meanings can be worked out.

Phonetics

Galic Northian is, by definition, the language exemplified by the Gales (composed c. 1800 – 1200 BCE) and also found in other corpora such as the Didaskalic Material. As such, it has been questioned whether certain features of it, such as the elision of final *-s and compensatory lengthening of a preceding short vowel, are sound changes in the underlying language or only poetic forms created by the conventions of Galic chanting. Similar behaviour has been observed in Syaran in the movable nu and, above all, in Tennite sandhi.

Nominals

See also: Northian nominals

The category of nominals in Northian encompasses nouns, adjectives, pronouns, demonstratives, reflexives, and certain adverbs. They are considered to belong to this class as they undertook similar grammatical processes and showed the same set of endings.

Endings

Athematic

The following chart recapitulates the ordinary endings of athematic nouns in Galic Northian. Because the ablative is syncretized with the genitive in the singular, with the dative and instrumental in the dual, and with the dative in the plural, it is usually not listed separately in grammatical tables for athematic nouns.

Forms are often unpredictable and variable under the influence of ablaut, laryngeals reflexes, analogy, vowel contraction, and compensatory lengthening for illegal consonant clusters in coda position. All endings are subject to modification according to the suffix. OX stands for the oxytone group of patterns, and PX for the paroxytone group. Certain neuter nouns take a collective ending; these nouns are not formally predictable. Because neuter nouns always have the same nominative and accusative forms, only their nominative endings will be listed, and in grey. Other than root nouns, there are virtually no neuter nouns that take the OX pattern; as such, their endings are listed together with the PX stems.

The cells listed in gree are typically paired with the full-grade noun stem, and the orange ones only sometimes; these do not apply for nouns with invariant stems.

	Athematic endings
	Singular			Dual			Plural		Collective
	OX	PX	Ntr.	OX	PX	Ntr.	M/F	Ntr.
Nominative	-ō, -ā̊, -s	-s, -š	-Ø	-ōi, -ā, -ō	-a, -ī, -ū, -ōy	-ī, -Ø, -ū	-aᵑhāʰ, -āʰ, -ōʰ	-a, -ō, -ā̊	-ō
Vocative	-Ø
Accusative	-m, -ąm, -əm, -ā̊						-ā̊, -āʰ, -ō, -ūš, -ī
Locative	-Ø, -i			-ō	-ū		-hū, -štū	-eC-Ø
Genitive	-ōʰ, -ā̊	-s, -š, -ō, -ā̆ṇġ, -ā̊, -Ø			-uš, -āʰ, -ō		-õm, -ą̄m
Ablative				-mō	-mō, -ma		-muš
Dative	-ə̄, -ay	-i, -aē
Instrumental	-ōy	-a, -Ø					-(m)βyōʰ, -(m)βiyōʰ, -βiš, -βīš

nom sing A general discussion of the athematic declension cannot omit the comment that, while many divergent forms are phonetically conditioned, there too are divergences resulting from divergent proto-forms. In no other place is this statement truer than in the nominative singular.

The marker of the nominative singular has been a tormented subject, in part also for the radical schism on the parent language's morphosyntactic alignment. By sole comparison, animate (= masculine and feminine) nouns may have been in the proto-language sigmatic, that is ending in *-s, or asigmatic, that is without final *-s and taking a long o-grade suffix; as root nouns had no suffix, they were thought to have been obligatorily marked by *-s.

The source of this long o-grade, in default of *-s, is disputed: some authorities regard it as the legacy of compensatory lengthening having dropped final *-s after a resonant, but others hold there was no *-s originally and attribute the long vowel to ablaut variation sensitive to the case. On the other hand, there are also nouns that have an exceptional zero-grade suffix, e.g. hanuš "jaw" and notoriously ϑā̀m "earth", and some of these could not have had *-s. There are also forms in related languages that show *-s added to the long-grade ending, in some root nouns and the present-aorist active participle suggesting *-ōnt-s.

In Northian, final *-s has been suffixed to animate nouns quite broadly but haphazardly in prehistory, so there is no obvious pattern to its distribution; many words have alternative forms differing by -s. We may distinguish three situations in Northian as to the nom. sing., stems ending in vowel, in resonant, and in non-resonants.

1. -s is always present and surfaces as -š after *i- and *u- in animate nouns, and its absence in these stems indicates neuter gender, both instance without regard to ablaut pattern.
2. Final *-s was absent in resonant-stems (-m, -n, -r, -l), whose nom. sing. was often signified by lengthened o-grade in OX and PX (though a few nouns have zero-grade). The long final syllable ending in a resonant was then opened, giving rise to -ā̊ and -ō.
3. After obstruents the distribution of *-s is not predictable: bā̊ "wife" and ϑanū "body" were asigmatic, but ą̄mβərətās "immortality" certainly had *-s.

In OX resonant stems, the lengthened o-grade is altered prehistorically by the opening of closed long syllables ending in a resonant.

voc sing The vocative consists of the bare strong stem in all cases. Where the nom. had *-s it is dropped, and where it did not the voc. is the full- or short-vowel-grade.

acc sing In the proto-language, the accusative ended in *-m and, as it contained no vowel, could theoretically not bear an original accent; this rule is violated by the semivowel (i-, u-) stems, where the vocalized vowel usually does bear and accent. In stems ending in non-resonants, the ending is vocalized as -m̥ > *-əm. *-m is preserved in u-stems, e.g. huyúm. If the stem ended in a long vowel, such as effected by Stang's law after *y, *w, and *m, the deletion of final nasals yielded -ā̊, e.g. zyā̊ < *dyēm.

loc sing The locative generally took the accusative stem and either added final -i or was endingless. Thus, for PX nouns, the locative and dative were often syncretized. For the effects of -i on the preceding vowel, see dat. sing. entry.

gen sing In OX the gen. singular always ended in -ōḫ < PNN *-os; its consistency led grammarians to consider it a feature of the OX declension.

In PX, the ending *-s when attached to the stem reflects a motley of forms, and this (compared to OX) irregularity in turn is the feature of the PX declension. The Northian evidence is important to the discourse on the phonetic process known as Szemerényi's law: by its regular operation, final *-s is dropped after resonants and lengthens the preceding vowel, but in Northian as in most languages, regular exceptions appear. In n-stems, *-s was either not dropped or was early on restored and became something like a glottal stop, as in fūváṇġ < *ph₂wén-s "of fire"; yet in the in- and un-stems, *-s was apparently not restored, resulting in gen. endings -ī and -ū, obtained by *-īn < *-in-s and *-ūn < *-un-s.

In liquid stems, final *-s is usually retroflexed, as in māϑrš < PEE *meh₂tr̥s. If the stem contained a long vowel, usually indicating a laryngeal, the result is -ā̊ < *-ās, e.g. jñiϑriyā̊. In s-stems, the ending generally disappears, e.g. mā̊ < *mn̥s-s.

In the semivowel stems (i- and u-) the ending *-s, obeying Szemerényi's law, disappeared and caused compensatory lengthening. But such long diphthongs in final position, as in other long syllables closed by resonants, usually lost the final glide, giving in the i-stems the ending *-ei̯-s > -ē and u-stems *-ou̯-s > -ō. For at least the u-stems, the intermediate form *-ōw must obtain, since the enclitic *-kʷe delabializes and becomes -kə after u-stem genitives.

abl sing For all athematic nouns, the ablative singular was syncretized wtih the genitive singular.

dat sing In OX the dat. sigular ending was originally *-ei̯. This ending susceptible to colouring by a preceding *h₂- or *h₃-, as well as the influence of i̯- and *u̯-, to become -ai and -oi respectively.

In PX, the ending was regularly *-i. But this ending was replaced by the OX ending in the i-stems early. For all nasal and laryngeal stems, the ending -i caused a preceding /e/ or /a/ to mutate to /i/ and /ai/ (written <aē>). For stems ending in -n, the -n sandwiched between i became /ñ/. In nouns of the type taēuuīḥ, the ending was full-grade even if the PX endings are otherwise employed, and there it appears after the suffix as -iiaē. In all cases the dat. singular ending following a vowel was a separate syllable. In u-stems, the ending is dropped just like final *-s of the genitive; the result is identical forms for the gen., dat., and loc. in the singular.

ins sing The OX ending -ōi̯ for the ins. singular originated as *-eh₁ in the proto-language. This ending is rarely problematic by phonological processes, but it is liable to be replaced in some stems, e.g. endings -ī and -ū in the i- and u-stems respectively, from the PX declension. The PX ending evolved from *-h₁. This ending was preserved only after plosives as -a. Following resonants, the preceding vowel was lengthened and opened. Following laryngeals, it disappeared.

nom-voc-acc du For animate nouns in plosives and resonant stems, the du. ending for all direct cases in OX was generally -ōi < *-ē, which is coloured in the usual ways to -ā and -ō, which do not mutate. After stems ending in laryngeals, there are concomitant spelling changes. In semivowel stems and all PX stems, the ending -a is visible after only after plosives, as it had the proto-form of *-h₁. After i- and u-stems stems, the ending was dropped causing the preceding vowel to lengthen, e.g. dorūḥ. After laryngeals, it disappeared.

nom-voc-acc du ntr For all neuter nouns, other than the u-stems, the ending was -ī.

loc du In OX the dual loc. ending was -ō < *-ou̯. In PX, the ending was -ū, which developed from original *-u lengthened in final position.

gen du The proto-form of the dual genitive is sometimes considered that of the locative with added *-s at the end, borrowed from the singular. Thus in OX the ending was usually -ō < *-ōw < *-ou̯-s, which was identical to the loc. form even in sandhi. But in some instances, the loc. form takes the strong grade stem, which provides a difference with the gen. In PX, the ending was -uš, which like the locative dissimilated to *-āḫ if there was a preceding u. In this case, the ending was -ōḫ. For the feminine nouns ending in *-eh₂, which are athematic in origin, the ending was a special -ō < *-eu̯s; see below.

The gen. du., unlike any of the other oblique cases outside the locative, was sometimes a strong case taking the full grade of the suffix. It has been argued the weak stem was replaced to disambiguate this form from the gen. sing. and that the strong grade was taken over from the collective; if the latter be true, the practice would probably be ancient. But neither explanation has received general acclaim because very few items are attested uniquely in the strong stem.

abl-dat-ins du These three forms were syncretized in Northian as -mō in OX and -ma in PX.

nom-voc pl The proto-form here was *-es, which regularly became *-ah, unless it followed *-u, in which case it bcame -ōḫ. An alternate Galic reflex of *-es is attested in compounds, where it appears as -iš or -iž in unaccented positions. The independent ending of *-ah > *-ā is rarely attested, in all likelihood because it became too similar to the athematic neuter ending -a and homophonous with the thematic ending -ā < *-eh₂. Instead, the form -ahā is seen, representing *-es-es. Such an innovation is independently attested in Xevednite and Nordic languages.

nom-voc pl ntr The ending prehistorically was *-h₂. After a stop, the ending became -ă. In the n- and s-stems, the laryngeal dropped and triggered compensatory lengthening of the full-grade suffix vowel. The resulting syllable was subsequently opened and became -ō in the n-stems (fnumō < *pnew-men-h₂) and -ā̊ for es-stems (neβā̊ < *nebʰ-es-h₂). In the i- and u-stems, the ending caused the zero-grade stem vowel to lengthen, resulting in endings -ī and -ū. After another laryngeal, the ending disappeared without a trace.

acc pl This ending was derived from *-m̥s following consonants or *-ms following vowels. In the case of semivowel stems, which occur in the weak grade in this form: for *-i-ms, the resulting ending was just -ī, except uniquely in the word for "three", where it remains as -īš (not *-īs!); for *-u-ns, the outcome was regularly -ūš, though -ū is also seen from time to time, probably an imitation of the i-stems. That the pre-form contained -ms is argued to indicate Northian was more archaic than most other daughter languages, which mostly show the reflex of *-ms > *-ns; in Northian, *-ms is diagnosed because at least *-ums seems to have a different reflex than *-uns, which occurs regularly in the wn-stems of nouns.

For consonant stems, the vocalization of *-n̥s (not distinguishable in this context from *-m̥s) is regular under Cloverdale's Law, where a syllabic resonant's surface quantity depends on the preceding syllable's (underlying) weight. Thus, where it was underlyingly heavy, the form *-ah > *-ā is created, and where it was light, *-āh > -ā̊ is used instead. Yet due to analogical replacement of the stem, the syllable on which the ending is based is not always present, and so the ending is not synchronically predictable; since the weak stem tends to replace the strong in this position, the combination of a heavy ending with a heavy stem is common. Additionally, a vocalized resonant that is superficially long under Cloverdale's Law still counts as a short vowel for the purposes of other instances of Cloverdale's Law.

The form of the acc pl was evidently a driving factor in the replacement of the simple nom pl ending, which had also become *-ah under the colouring influence of *-h, and it became reduplicated as *-ahah in most contexts, leaving *-ah as an irregular alternative. The form -ō is used in the laryngeal stems, though it is disputed whether this is merely an orthographical alteration to avoid contraction of like vowels or a genuine sound change.

gen pl The ending was consistently -õm, or -ą̄m after vowel stems.

abl-dat pl The ending was -muš.

ins pl The ending evidently consisted of the element *-bʰi̯- in the proto-language. It was usually added to *-os > -βiiōḫ, with Sievers's alteration to disyllabic -βiyōḫ following heavy syllables (long vowel or short and two consonants). The disyllabic form was noticeably more common. In demonstratives the equivalent sequence was -βīš or -βiš; it is not completely clear if this was simply an ablaut variant or reflects a different combination of morphemes.

Thematic

	Basic ā-stem endings			Basic o-stem endings
	sing	du	pl	sing		du		pl
nom	-ā	-aī	-aᵑhā̊	-ōʰ	-õm	-ō	-oī	-ā̊, -ā̊ᵑhāʰ	-ā
voc				-i
acc	-ā̊		-ā̊s	-õm				-ā̊s
gen		-auš	-aõm	-ōiia		-ōʷ		-aõm
loc	-aē	-aū	-āhū	-ōy				-ohū
dat		-āmiyā	-āmβiyō			-omyā		-omβyō
abl	-aoṯ			-ōṯ
ins	-ā		-āyš	-ō				-oyš, -ōyš

nom sg The ā-stems showed the expected ending -ā. M. and f. o-stems have -ōḫ < *-os, which scans short at the end of sentences and other pauses. N. o-stems have -õm.

voc sg The ā-stems have the same form as the nom. M. and f. o-stems have -i < *-e, while n. o-stems have the same form as the nom. In both cases, the accent is always retracted to the first syllable of the word.

acc sg for ā-stems is affected by Stang's law, which appears as -ā̊. The ending for m. and f. o-stems is the same as the n., -õm.

loc sg ā-stems have dysyllabic -ayi; o-stems have monosyllabic -oy.

gen sg ā-stems show -ā̊ for *-eh₂-s; o-stems have the compound suffix -ōyo, for *-osyo.

abl sg in ā-stems is dysyllabic aā̊ṯ; the quantity owes to dissimilation.

dat sg ā-stems

ins sg ā-stems

Noun stems

	-C	-t	-m	-n	-r	-s	-i	-u	-ī	-ū	-H	-r/n	-nt
OX-B	About 80 root nouns and suffixed nouns in -k, -l, -ns, -d, etc., fos = "foot", xō = "heart"	A few in -ōs, nepōs "child"; -it melit "honey"; -ut xaput "head"	About 20 feminine, in -ō or -am, xaii-ō "winter"	Many, in -ō and -ə̄, often binds m-, s-, t-, i-, u-, H-, θxām-ō "human"	haz-r "arm" and personal, feminizing suffix -ez-r	A handful in -ā̊, aōšt-ā̊ "dawn"; the perf. act. ptcpl. waēδuš "knowing"	Some in -ō and -iš, hak-ō "friend, ally", ouu-iš = "ewe"; neuter var. in -ai, oxθ-ai "finger"	A few dozen in -ōš or -uš, han-uš "jaw"	Some in -ī, wrk-ī "wolfess"	A handful in -ū, ϑen-ū "body"	fənδā̊ pāϑō, "path"	Derived coll. of heteroclitics, in -r ~ n-, ō-grade in nom. > -ā̊	A few in -āṯ, dāṯ = "tooth"
OX-A	No	Productive suffix -tāt-, āmβr-tās "immortality" and -tūt-, wəš-tūs "moistness"	No	Productive, in -ṓ and -ā̊	Productive through agentive -ter, duhiϑ-ṓ "daughter"	Productive, comparatives in -iiā̊- maj-iiā̊ "bigger", perf. act. ptcpl. in -uuā̊-, βiβiž-uuā̊ "trusting"	Verbal nouns, middle, in -doi, št-oi "an eating"	Certain denominal nouns, in -ṓ, piϑr-ṓ "uncle"	Non-ablauting feminine derivatives in -ī́	No	Certain nouns in -dōi, žβāž-dōi "libation-giver"	No	Productive pres. aor. act. ptcpl. of thematic verbs, in -ās hadáii-ās "sitting"
PX	No	No	No	Productive, neuter var. of OX n-stems, exhibiting same bound suffixes	yāϑr, "husband's sister"	Productive, neuter var. of OX s-stems, in -ō; neβ-ō "cloud", xrat-ō "power"	Productive, animate in -iš, action nouns -tiš, gomβ-iš "a step"; neuter in -i, mor-i "sea"	Productive, animate in -uš, wiš-t-úš "witness"; neuter in -ū, oii-ū "life"	Proliferate suffix deriving feminines in -ī; geniϑr-ī "genitrix"	Same as -ī but more common in adjectives, in -ū; hoxr-ū "mother-in-law"	bā̊ "wife"	Neuter, in -r ~ n- suffix, ya-ərə "year"; more in -tar, -mərə, -zərə	No
PP	About 30 root and suffixed nouns in obstruent stems	nox-ṯ "night", haš-ṯ "bed", xom-d "hand"	tā̊ "house"	Extended with -n-s, māʔō "moon"	māϑ-r "mother", βrāϑ-r "brother"	zraw-š "gore"	aŋhi "serpent"	Neuter nouns in -ū, gon-ū "knee", wəšt-ū "settlement"	Feminine pres. aor. act. ptcpl. with static accent, déδāṯīš "of the giving one"	No	A few, particularly neuter in -i, ošti "bone"	A few, in -r ~ n-, f-ō "shrine"	A few nouns; pres. aor. act. ptcpl. with static accent, déδāṯ "giving"

As appears from this schematic, most suffixes are associated with more than one accentual pattern. But even in Galic, the majority of suffixes have only one productive accentual pattern or separate productive patterns associated with masculine-feminine gender and neuter gender (the case of the n-stems and s-stems). Additionally, some suffixes are only productive through petrified compounds, which tended to be hysterokinetic and have invariant stems, such as the comparative in -iiā̊- and perfect active participle in -uuā̊-; otherwise, the s-stems in amphikinetic is non-productive.

Where there are multiple productive accentuation patterns, neuter nouns are almost always identified with the proterokinetic pattern, and masculine-feminine with the hysterokinetic or amphikinetic. The exception is for i-stems and u-stems, wherein proterokinetic accentuation is standard, and particularly productive through the compound with -t.

As is generally observed across all the Erani-Eracuran daughter languages, the suffixes ending in resonants and semivowels, i.e. n-, i-, and u-stems, tend to be productive over the suffixes in stops. There is some speculation why this is the case, one theory being the resonants were able to be vocalized as syllables and thus did not produce illegal clusters of consonants, given the tendency to delete vowels in unaccent positions. It is argued that the compound suffixes, which often mix one stop and one resonant are the results of transferrals from suffixes in stops to those in resonants; if so, these transferrals must have occurred quite early, as their older siblings in stops generally do not leave visible remnants in attested languages.

Unlike verbal formations, where suffixes often have specific meanings and exist as part of the standard morphology of verbs, the meanings of noun suffixes are barely if at all recoverable. Synchronically, there is no obvious meaning to most of the suffixes present in Galic, even the productive ones, above a rudimentary association with certain parts of speech from which a noun is derived. For example, the -tiš nouns most often are nomina actionis derived from verbs, -tuš having a similar function, -tāt and -tūt deriving states of being, but there is no comparable explanation for their common constituent -t-, which appears as an independent suffix in nouns like haš-t and nep-ot.

Adjective stems

Adjectives agree with the nouns they modify in gender, number, and case, within their lexical paradigms. Inasmuch as nouns have differing endings that convey the same number and case, so too do adjectives have lexical paradigms; adjectives do not agree with the paradigms of nouns that they modify.

	-k	-n	-s	-i	-u	-h₂	-r/n	-t	-nt	-ā	-o
OX	mera-xš "soft"	Masculine forms of adjectives with heteroclitic neuters, in -ṓ, fīuuō "fat, fertile"	Productive adjective in -ā̊, humen-ā̊ "well-intended", neut. -ā humen-ā; productive comparative in -iiā̊ -ištṓ, θáñ-iiā̊ "thinner"; productive perf. act. ptcpl. in -uuā̊ -uštṓ, tita-uuā̊ "having made", fem. -uśiiā, neut. -uš	A handful, in -iš, θraišt-iš	A handful, in -uš -uuṓ, meδ-uš "sweet"	máh-iš "big"	Adjective forms of neut. heteroclitics, masc. form in -ō and fem. in invariant -r-ī, faōuu-ərə "fat, fertile"	duš-ṯ "bad"	Productive derivative meaning "rich in, bearing of" in -uuā̊, β-uuā̊ "rich in power", from p- "power"; productive pres. act. ptcpl. of athematic verbs with mobile accent in -ṓs -ā̆tṓ, and of thematic verbs, in -ṓs -óṇδō	Masculine and feminine in -ō, some obligatory feminine in -ā, neuter in -õm
PX	No	No	No	Productive, adjectives in -uš, feminine in -ū or -uuī, neuter -ū, fərət-uš fərət-ūvī fərət-ū "flat"	Productive, adjectives in -iš, feminine in -yī, neuter -i, hámil-iš hámil-ayī hámil-i "similar"	Productive, feminine derivative of u-stem and i-stem adjectives, in -ī -iiā̊	No	No	No
PP	No	No	No	No	No	No	No	No	Present active participles of athematic verbs with root accent, in -ā̆t, -ā̆s

Adjectives need to agree with the nouns they modify not only in number and case but also in gender, but forms for each gender may not necessarily be distinct from each other. Synchronically, many adjectives have a single form for animate (both masculine and feminine) referents, and a handful have the same forms for all three grammatical genders. Whether an adjective has distinct forms for each gender is lexical, and there is no obvious semantic difference which appears to condition their presence or absence. The usual historical explanation is that the feminine gender was a late grammatical development and did not always correspond to semantics of biological gender, though the mechanisms of the grammaticalization of the feminine gender is uncertain.

For o-stem adjectives with a masculine nom. sing. terminating in -ōḫ, there is always a separate neuter form ending in -õ. Those which have a distinct, obligatory feminine form will have one ending in -ā. Thus these adjectives are called "three-ending" o-stem adjectives. Those without a distinct, obligatory feminine ending terminate in -ōḫ for both masculine and feminine referents. These are "two-ending" o-stem adjectives. It should be noted that feminine forms of o-stems add the -ā directly to the stem, not after the -o theme vowel; this is in contrast to the *-h₂ stems (see below) which is usually added following an existing suffix.

Adjectives terminating in -k, -n, -s, and -t generally do not have distinct forms for masculine and feminine referents, but a handful will have a -ī suffix following the existing suffix to create a distinct feminine form. The netuer form is distinguished from the animate form in one of two manners. It may be by ablaut, taking a short vowel grade when the animate has long grade or a zero grade when the animate has short. Or it may be by the absence of final -s in the nominative, where the animate nom. has -s.

Adjectives in -i and -u often have distinct feminine forms ending in -ī, but there are also adjectives which have just one form for animate refernets or even one form for referents of all genders. Furthermore, there are feminine forms which have a long vowel where the masculine has a short vowel, e.g. fem. nom. sing. -ī and -ū, contra masc. nom. sing. -iš and -uš. The long vowel is conditioned by final *-h₂, which is the same as in *-ih₂ > the usual feminizing suffix -ī.

Adjectives in -nt and -wos, mostly participles, create their feminine forms by adding -ī.

There is also a class of heteroclitic adjectives based on heteroclitic nouns, which are all neuter except hāuuərə "Sun", a feminine term. Those heteroclitic items which have PP inflection take the PX inflection for their primary adjectival forms, which are also neuter and have zero grade in the suffix. The masculine form is produced by adding OX n-stem endings, and the feminine by the -ī suffix to the neuter form. In the feminine, the heteroclitic suffix has full grade and the suffix zero grade in the strong cases, and vice versa in the oblique cases.

It has been noted that adjectives often show a different ablaut pattern compared to nouns derived from the same stems. That is to say, a noun with proparoxytone or paroxytone accent can often respectively form an adjective of similar meaning with a paroxytone or oxytone accent or oxytone accent. The source of this derivation is academically debated but remained visible and productive in Northian.

All Nordic languages have an extant distinction or some vestige of it between strong and weak declensions for the same adjective, where the "strong" represents the adjective's inherited declension and the "weak" its declension as an n-stem. This is not true of Northian, which lacks a weak declension. This would suggest in historical terms that the weak declension developed after the Nordic-Northian split in the Middle Bronze Age, and all the Northian adjectives are thus "strong" in Nordic terms.

Numerals

1 – 4

Cardinal numbers one through four are declinable as athematic adjectives of various declensional patterns, agreeing with the nouns (explicit or implicit) they modify in gender, case, and number. Of course, "one" is only inflected in the singular, "two" in the dual, and "three" and "four" in the plural. Numbers five and above are indeclinable.

	"one"			"two"			"three"			"four"
	m	n	f	m	n	f	m	n	f	m	n	f
nom	hā̊	hõm	hámīḥ	žuuōḥ	duyīḥ		ϑráiiāḫ	ϑrī́ḥ	ϑrižrāḫ	kośwóriš	kótur	kóśwṛšriš
voc	hõm
acc	hā̊		hámī				ϑrī́s		ϑrižrā̊	košwórā̊		kóśwṛzrā̊
loc	hám		hmiyaē	duoi		duvaō	ϑrištū		ϑrižṛžiiū	kóśwṛžyū		kóśwṛzṛzyū
dat	hmai			dumá		duvāma	ϑrimuš		ϑrižṛmuš	kóśwṛmuš		kóśwṛzṛmuš
abl	hmōḫ		hmiyā̊
gen				duōš		duvāvuš	ϑriiõm		ϑrižrõm	kóturõm		kóšwṛžrõm
ins	hmōi		hmiyā	dumβīḥ		duvāma	ϑriβyō		ϑrižṛβyō	kóśwṛβyō		kóśwṛzṛβyō

1 is a root noun with a stem ending in -m. As with other stems ending in -m, the accusative preform *sem-m̥ would by regular phonetic change become *sēm, i.e. the same as nominative *sēm, because the PEE ending *-m̥ regularly absorbs the previous resonant, hence also nom. ziiaōš < *dyeu-s but acc. ziiā̊ < *dyēm < *dyew-m̥; in the number, -əm is often but not always restored. In the oblique cases, the stem is in zero grade and appears as hm- < *sm-. The feminine forms are a regular development from the familiar feminizing -īḥ suffixation; the forms are a perfect match with Syaran μιᾶς = hmiyā̊, etc.

2 is only declined in the dual number. There are two stems in use: the monosyllabic žuuo- and dysyllabic duo-. It is not certain why the stem scans as two syllables in the neuter forms. Stringer says that the monosyllabic form reflects full-grade *dwoH-, and the dysyllabic has the stem in zero-grade *dwH-. But this theory does not explain the short /u/ found in dat. duma or the zero grade ending in place of an expected an full grade. Additionally, the initial consonant exceptionally does not become a fricative. It has been argued by others that the declension of "two" reflects an early attempt to introduce ablaut to a stem that had no original ablaut, upon the influence of other numbers 1, 3, and 4, which all had ablaut.

3 is a regular PX i-stem noun and is only declined in the plural. Nom. ϑráiiāḫ shows regular development of *e > a bordering yod. As with other animate PX nouns, the accusative plural has a zero-grade suffix followed by a zero-grade ending: *tri-ns > ϑrī́s. The sequence *-ins developed irregularly, usually appearing as -ī in Northian; it is also a notorious false friend to Nordic þrīz, which was not the accusative but the nominative = Northian ϑráiiāḫ. The feminine forms employ the feminizing infix -sr-, which is always found in the zero grade, and take regular athematic endings. There is also a particular form for three women or goddesses, as in δaēuuiyāḫ ϑraḗšrāḫ "three goddesses".

4 behaves like most athematic nouns and also employs the feminizing infix -sr- for its feminine forms. Note however that the ablauting element was the second syllable of the stem -tuuor-, which in zero grade appears would be -tuur-. Which of the two resonants vocalize depends on the phonetic environment. Where the suffix stands alone the *-w- is vocalized, as in neuter nominative kotur < *kʷetw̥r, but where an obstruent follows the suffix it is the *-r- that becomes syllabic, as in kóśwṛmuš < *kʷetwr̥mus. There was also a singular form košuuō < *kʷetwōr = Venetian quattuor.

The feminine forms for "four" have the particularly long stem of hošwṛzṛ-, which is for *kʷétwr̥-sr̥- where the ending begins with a consonant. The masculine stem for "four" frequently supplants the feminine owing to the sheer length of the etymological stem, which is metrically unusable. Note that the accent is on the suffix syllable in the strong forms owing to the effects of the eponymous kʷetwóres rule, which shifts the accent from a preceding *e to the following *o if followed by only one other syllable.

5 and higher

5 fəṇka is from *pénkʷe.

6 xšuuāxš from *kswéks, a match with Xevdenite xšuuah.

7 haftam from *septm̥.

8 oxθō, the proto-form of this word is disputed. Northian oxθō can be traced back to both *(H)oktow and *(H)oktoH, with or without an initial laryngeal. Morphologically, it is the dual of óxθō "fingers", in ei-stem.

9 nauuam

10 dekam

Pronouns

First person

	sing		du		pl
	stressed	enclitic	stressed	enclitic	stressed	enclitic
nom	áxa, áɣā̊		wōḥ		wāi
acc	mḗ	mi	āŋhō	nō	ənśmé	nāḫ
gen	méni	mai	nō		ənśr-(I/II)
dat	méjiia		nanā́		ā̊(s)

nom sg The term for "I", usually áxa, comes from Erani-Eracuran *éǵ-h₂, with regular devoicing of a stop before *h₂. The long form áɣā̊ must have *éǵ-ōm, without laryngeal, but cognate extensions to the pronoun with this suffix all have the laryngeal. This would suggest that an unattested Northian form of *ák < *éǵ may have existed independently for the suffix to be added.

Second person

	sing		du		pl
	stressed	enclitic	stressed	enclitic	stressed	enclitic
nom	tū́		yṓḥ		yā̊́ḫ
acc	suuó	ti	ūmé	wō	ušmé	wā̊
gen	ϑáiia	toi	yuϑr-(I/II)		yušr-(I/II)
dat	ϑə̄mβiiō		wanā́		ušmái(iai)

Demonstrative

	sing			du			pl
	m	n	f	m	n	f	m	n	f
nom	ha	ϑaṯ	hā	ϑō	ϑoyī	ϑāyī	ϑoi	ϑa	ϑai
acc	ϑõm		ϑā̊				ϑā̊		ϑáō
loc	ϑoi		ϑahiiai	ϑṓhō			ϑohū		ϑāhū
gen	ϑōiio		ϑahiiā̊	ϑṓhōš			ϑoiiõm		ϑāõm
dat	ϑōžmōi		ϑahiiayi	ϑṓzma			ϑoiiomuš		ϑāmuš
ins	ϑō		ϑahiiā				ϑoiiomβīš		ϑāβīš

Interrogative

	sing		du		pl
	m f	n	m f	n	m f	n
nom	kiš	kiṯ	kə̄		ϑoi	ϑa
acc	kim				ϑā̊
loc	kāsmi		kiyō		ϑohū
gen	kāiio		kiyōš		ϑoiiõm
dat	kāsmai		kimō		ϑoiiomuš
abl	kāṯ				ϑoiiõm
ins	koi		ϑoiiomβīš

Indefinite article

The Northian indefinite article, which introduces an indefininte noun phrase, is derived from the PEE root *oiwos, meaning "one". Note that the endings are those of demonstratives.

	aēuuōḫ, "a, an"
	M & F	N
nom	aēuuōḫ	aēuuó
voc
acc	aēuuõm
loc	aēuuōi
dat
abl	aēuuōṯ
gen	aēuuōiš
ins	aēuuō

Verbs

See also: Northian verbs

Endings

Athematic

The athematic verb endings, like their noun counterparts, are directly attached to the verbal stem without an intervening theme vowel. The primary endings are used for the athematic present, and the secondary endings for the athematic present injunctive, the imperfect, and (with the suffix) the optative, as well as the aorist indicative, injunctive, and optative.

There are two sets of (phonologically conditioned) parallel endings called the long and short endings. The long endings (noted below in grey) arise from a laryngeal between the stem and ending, whose effects are determined by neighbouring sounds. Where it precedes a guttural consonant, it becomes a, a labialized element, u, and elsewhere, i. If it preceded e, the e is coloured according to the laryngeal's identity.

The two sets of endings further vary depending on the position of the accent, since unaccented *e regularly becomes /i/ and unaccented *o > /a/ in most positions. These changes are reflected quite regularly in Galic, but in the Epic language most of the phonetic variants had disappeared.

Primary active, mobile [static]				Primary middle endings, mobile [static]
	sing	du	pl		sing		du		pl
					trans	intrans	trans	intrans	trans	intrans
1p	-aē-mi	-ū-wəiñi	-a-maŋhi [mahi]	1p	-ai		-ū-uuōδa		-a-mōiδa
2p	-i-šti	-i-tāḫ	-i-te [ti]	2p	-i-tai		-ātiϑai		-i-dūvó
3p	-i-ti		-éṇti [āti]	3p	-i-tói	-ó	-ātāyi	-ā	-éṇtro [ātro]	-ṓ [r]

Secondary active endings, mobile [static]				Secondary middle endings, mobile [static]
	sing	du	pl		sing		du		pl
					trans	intrans	trans	intrans	trans	intrans
1p	-ā̆m	-ū-vṓḫ	-a-mṓḫ	1p	-a		-ū-voδa		-a-meδa
2p	-i-s	-i-tõm	-i-té	2p	-i-ta		-i-tāδi		-i-duvó
3p	-i-t	-i-tā̊	-enṯ [ā̆ṯ]	3p	-i-to	-o	-ātā	-ā	-arə / ār

1 sing The primary and secondary active endings differ with the hic et nunc particle *-i in the proro-language, for the singular active. The element m is accepted in mainstream reconstructions of Proto-Erani-Eracuran to signify the first person. As m is a resonant, the ending -i in the primary conjugation can trigger mutation in the preceding syllable in the usual manner, particularly in a syllable generated by an interconsonantal laryngeal. In the secondary conjugation, final -m can vocalize to -ā̆m if following a stop; the final nasal is insensitive to length and can be written long should the Galic meter demand it. But if the verb stem ended in a full- or long-grade vowel plus resonant, the final -m triggers Stang's law resulting in a lengthened vowel that subsequently loses the final -m. In late texts, this -m is usually restored following the long vowel.

In the middle voice, the ending evolves from *-h₂ey > -ai. This ending is agnostic as to any preceding laryngeal. The secondary middle ending loses the hic et nunc particle, as with the rest of the singular middle.

2 sing In the primary conjugation, the signifying element of the active second singular *s can become [h] or [š] depending on the phonetic context; if the latter, epenthetic [t] is introduced to separate it from the following -i. In secondary sequence it usually triggers compensatory lengthening in resonant stems. In the case of *-H or semivowel stems, it usually becomes identical to the 1 sg form, but in contrast thereto, final -s is never restored.

The middle ending here is *-th₂ey > -tai. If there is a preceding laryngeal, it appears as -itai.

3 sing This -ti ending is usually retained in the primary conjugation. If the stem ended in a dental, the ending was liable to mutate in several ways. In the secondary, -t can displace preceding stops or be dropped in some contexts.

The middle ending of the third singular depends on the meaning of the word and the stem used, which is peculiar. In root verbs and many stem-classes, a middle verb with intransitive menaing will take the ending -o, and those with transitive meaning, -toi. In other cases, such as the -naō ~ nu- stems, the ending -toi is always used, regardless of meaning. If a laryngeal preceded the ending, it appears as -itoi. Where the ending is not accented, it appears as trans. -itai or intrans. -a.

1 du The primary active ending is from *-wen-i and appears as accented -uuóni and unaccented -uuiñi. If a laryngean preceded the ending, it became *-u-weni, whereupon nasalization induced -u-mβóni.

In the middle, the form -vozδa < *-wesdʰh₂ is found. The long ending was regular -u-vózδa; nasalization did not intervene here as the vowel was originally a back vowel *o.

2 du Here the active ending -tāḫ is for *-th₂es. An epenthentic -s- is sometimes found if the stem ended in a dental to avoid a sequence of two dentals together, and the resulting combination is sometimes resolved to prehistoric *-ss-. But this was not a universal phenomenon, and sometimes the geminate dental either drops or evne surfaces. Such examples are often interpreted by analogical restoration. Secondary -tõm is found in the middle for *-tom.

The middle ending is -ātiϑayi, which is structurally complex and the subject of much debate. First, the final -i must have been added only after the final laryngeal vocalized; otherwise, the monosyllabic ending *-ϑi would be expected for *-dʰH-i; indeed, it is often omitted in Galic. The element -ϑa- is often considered identical to that found in the 1 pl mid ending -mōi-δa, with the initial dental devoiced following a laryngeal reflected as -i-. That this element should be deemed a particle is clarified by the development of 1 pl mid -mōi-δa < *-mes-dʰH, which is only regular word-finally, and also that it is shared with the 1 du mid ending. The distinct part of the ending is thus -āti-, which has the zero-grade variant -(i)ti that appears after roots with persistent accent. The element -ti- < *-tH- has been identified as a zero-grade variant of the Kankrit 2 pl act ending -tha < *-tHe.

Curiously, Northian presents both parallel and contradictory information to Kankrit comparanda, which has 2 du mid primary athematic -āthai̯ and thematic -a-i̯thai̯. Kankrit has distinct secondary -ātham, while Northian attests no distinct secondary form. If the particle -ϑa- were to be omitted in Northian, the resulting sequence *-ātiyi would be very similar in structure to the Kankrit, especially if a full-grade vowel can be posited in the second syllable and superficially deleted in unaccented position. Disputes cloud the identification of the first part of the ending, which behaves differently in both languages. In Kankrit, the variant appearing after the thematic vowel cannot be identified as a laryngeal, but that is nearly required in Northian.

3 du -tāḫ reflects *-tes. As with all endings which begin with /t/, it is liable to an epenthetic -s- following another dental. There is thus a superficial identity between the 2 du and 3 du primary endings; this identity was often extended to the secondary where it is not a regular outcome in later materials, usually at the expense of the 2 du ending, which was apparently less frequently used.

The secondary ending is -tā̊ < *tā́m < *teh₂m. The -m ending is confirmed by the co-ordinating imperative ending, which shows the particle -u attached, producing *-tā́mū.

In the middle, this personal form is also sensitive to the transitivity of the verb stem. Where the 3 sg & pl forms require transitive endings, this form will canonically take the ending -ātā, and the intransitive ending is -ā, with the particle -i added in the primary sequence. But in the received text, -ātā is often seen in place of expected -ā; considering they have a differing number of syllables, this could hardly be a metrical alteration.

1 pl In the primary active one finds -máŋhi < *-mensi. This is usually explained as a concactenation of the 1 pl. suffix *-men plus the (redundant) plural marker *-s, with the hic et nunc particle *-i. For verbs with recessive accent, a different form -mahi is used; this would be from *-mesi. The secondary form is always -mo.

The subjunctive does not take the normal primary ending of -máŋhi but rather the ending -omōhi, which is best explained as the thematic ending -omō plus the segment -hi extracted from the athematic.

In the primary middle, the ending -mōiδa is encountered, usually thought to be for *-mesdʰh₂. The expected phonetic outcome is *-mezδa > *-mēδa, but it seems the *z was elided in such a way that it caused the preceding vowel to lengthen, which then resolved as though it were at the end of a word *-ē > -ōi. Alternatively, the ending could have been -meh₁dʰh₂, which would produce the same result. In either event, it indicates the *-dʰh₂ could have been considered an independent particle, thus triggering the word-final phonetic change for the long vowel. While *-mes is preferred in the interest of comparison to archaic Syaran -μέσθα, *-meh₁ would compare very well with the 1 pl perf ending -mōi < *-meH.

In the secondary, the ending was -méδa is used.

2 pl The allomorphs are -te or -ti. -ti is from unaccented *-te and is regularly seen on verbs of recessive accent. However, a deviant form -ta also exists for some verbs, and its origin is unexplained.

The middle endings primary -dūvó and secondary -duvó has caused some controversy amongst academics as its provenance is open to many interpretations. No other Erani-Eracuran language attests a difference between primary and secondary endings in this position, and much Galic material also does not distinguish between them. But in the Early Galic, -dūvó is clearly preferred as the primary ending, being attested ten times over the two times of -duvó. In secondary sequence, -dūvó never appears at all. Some prefer to see the length difference as militated by that found in the 1 pl, where the elision of *-z created a long vowel in the primary but not the secondary. But the quantitative difference did not disappear in that form, while the putatively connected contrast disappeared rapidly.

The general shape of these two endings also require some comment. The u-vocalism itself could have two origins. First, as in Kankrit, it could be attributed to a form of Sievers's law that created a syllabic *u before non-syllabic *w following a heavy syllable, but this variety of Sievers's law did not operate generally in Northian. Second, the pre-form *-dʰh₂wé would regularly vocalize as *-δiwó > -δuwó, since /i/ before /u/ is always assimilated to it. Because *w always follows two consonants and thus a heavy syllable, the Sievers's form *-dʰh₂uwé is generated, which has been argued as the source of primary -dūvó by way of metathesis to *-dʰuh₂wé, though this hypothesis does not explain secondary -duvó.

3 pl In the active, the ending -ənti is used, which is written as -anti if following h-. The vowel at the beginning is susceptible to colouring by a stem-final laryngeal. In verbs with persistent accent, this ending takes the zero-grade form of *-n̥t > -āt.

For the middle voice, there are several endings that share (what is usually interpreted as) a morpheme *-r. Most present, and all secondary, verbs show -ntro, though a few merely -ro. -r reprises as the standard ending for the i- and u-stem present verbs (regardless of meaning) in secondary sequence, as -ir and -ur, and it further appears as the standard ending for root aorists in both the indicative and optative. The ending is furthermore found in the same place in the perfect. It is thus unclear in which direction the borrowing occurred.

Thematic

The primary and secondary thematic endings include a theme vowel between the stem and the ending-proper, varying between *e ~ o. The thematic endings formally differ in the active singular and third plural from the athematic ones but are transparently the same, with the addition of the theme vowel, in others. It is still a matter of active debate what the contrast between athematic and thematic endings was in the proto-language. The primary and secondary thematic endings are used in present and aorist stems in the same manner as the athematic ones.

Thematic active endings				Thematic middle endings
	sing	du	pl		sing	du	pl
1p	-ōḥ	-ouuōḫ [auuōḫ]	-omō [amō]	1p	-oai [aā̊i]	-ouuazδa [auuazδa]	-omazδa [amazδa]
2p	-ai	-etāḫ [itāḫ]	-eti [iti]	2p	-etai [itai]	-etõm [itõm]	-eθō
3p	-ati [iti]	-etiš [itiš]	-oṇti	3p	-etai [itai]	-etā̊ [itā̊]	-ō

1 sg The first singular active ending is -ō. The middle ending is -aai for *o-h₂e-i—the ending is disyllabic in Northian.

2 sg The ending for the second active singular is -ai. The middle ending is the same as the athematic one, with the theme vowel /e/ inserted.

3 sg In the third singular one finds the ending -eyi; note that this ending is disyllabic, unlike that of the second singular; ditto for the middle.

1 - 3 du and 1, 2 pl For all these items the thematic forms are the same as the athematic ones, with thematic /e ~ o/ added.

1 pl Ending -omō does not show -s, in contrast to the 1 du.

3 pl The endings here are active -o and middle -ō.

Thematic secondary endings, active or middle, are all the same as athematic ones, with thematic vowel inserted in like manner as the primary.

Imperative

The imperative in Northian does not have opposition between primary and secondary. It is observed that the imperative usually implies immediacy, while the stem has aspectual value regarding the action required. The first person imperative is always defective: a speaker expressing a requirement for oneself would use the future tense. For all dual forms, the imperative is the same as the indicative, there being no sign that these ever had distinct imperative endings in Northian.

Imperative active endings				Imperative middle endings
	sing	du	pl		sing		du		pl
					trans	intrans	trans	intrans	trans	intrans
2p	-δi / -ϑi / -Ø	-i-tāḫ	-i-ti	2p	-i-zuua		-ātaθa		-i-duua
3p	-i-tū	-i-tāmū	-eṇtū [ātū]	3p	-i-to	-o	-tā	-ātā	-eṇtrō [āθō]	-ārə / -ir

2 sg either endingless or *-dʰí, which normally yielded -δí. A preceding laryngeal devoices the voiced stop and disappears, giving -ϑí. All nasal-suffix verbs (but not the nasal infix) have the endingless form.

2 du & pl endings mimic the indicative endings; suffixed verbs drop the suffix.

3 sg, du, & pl appear to be the corresponding secondary ending plus the particle *-u, which is used in all 3p forms. 3 pl ending is ablauting, ātū < *-n̥t-u when accent is in the stem. Unlike the 2p forms which have initial accent, 3p forms have accents that correspond to indicative forms.

The imperative forms for thematic verbs are as follows:

Imperative active endings				Imperative middle endings
	sing	du	pl		sing	du	pl
2p	-Ø	-etāḫ	-eti	2p	-ezuua	-ātaθa	-ezδuuo
3p	-etū [itū]	-etiš [itiš]	-oṇtū [aṇtū]	3p	-eta [ita]	-ā̊tā	-onδrō [anθō]

Future imperative

The future imperative probably only had active forms in the proto-language, but a middle voice was created on the pattern of the active in Northian. Both forms were created by appending the particle *-tod to the corresponding present imperative stem with no alteration. The particulate character of this particle is still quite clear in Northian, as it is often attached to the full present imperative word rather than only its stem.

The only point worthy of note is probably the 3 pl. form, where the final particle exceptionally counts as a full syllable and draws the accent away from the normally full-grade ending. Thus, where it occurs behind -entū, the ending is -atu-tót, as in G.Nr. 113 apuyātutót "let them go away", as opposed to apuyántū. This form is to be regarded as inherited, as the zero-grade variant of the 3 pl. imperative has been superseded by the full-grade variant even in Late Galic.

Future imperative active endings				Future imperative middle endings
	sing	du	pl		sing		du		pl
					trans	intrans	trans	intrans	trans	intrans
2p	-i-tót, -δi-tót	-i-taš-tót	-i-ti-tót	2p	-i-zuua		-ātaθa		-i-duua
3p	-i-tu-tót	-i-tāmu-tót	-ātu-tót	3p	-i-to	-o	-tā	-ātā	-eṇδrō [āθō]	-i-rō

Perfect

The perfect was an athematic formation, irrespective of the thematicity of the present or aorist stems.

Perfect active endings				Perfect middle endings
	sing	du	pl		sing	du	pl
1p	-a	-wō	-mōi	1p	If the perfect active is transitive, the middle is formed from the perfect subjunctive
2p	-ta	-ō	-ōi	2p
3p	-e	-iϑō	-r	3p

3 sg the ending -i is the unaccented outcome of *-e, which is coloured by any preceding laryngeal and therefore may surface as -a if the stem ended in *-h₂ or *-h₃.

2 du This form is at the centre of philological debate as it yields up so very little information at its surface. In contradiction to the usual distribution of ablaut grades in the perfect, in both its appearances in the Gales (both Early Galic) it does so with the strong stem rather than weak, i.e. momṓnō "you two recall" and woyδō "you two know". A final -ō could represent so many pre-forms (e.g. -ow, -ōw, -ōy, -eh₃, -oH, -h₃eH, etc.) that it is totally uninformative, and given this form's rarity, it unfortunately lacks sandhi clues. Sandy recommendeds that the particle *-h₃ is involved, shared with the pronouns, but not necessarily a borrowing from pronouns; his argument partly proceeds from the 3 du ending -iϑō, which he reconstructs as *-h₃-teh₂, with the last particle identical to that from the present.

On the other hand, Patrickson argues this aberrant ending must be a new creation, formed by removing the -w- of the 1 du ending that was (via the present system) associated with the first person. He observes the same pattern in the plural, where the 2 pl ending lacks an identifying element that precedes the ending in the way the 1 pl ending does. Regarding this line of thinking, Mortimer amongst others point to Kankrit endings 1 pl -mā < *-meH and 2 pl -a < *-e and concludes Northian has probably extended the prehistoric laryngeal in the 1 pl to the 2 pl. Thus, the general direction of analogy at least does not oppose Patrickson's argument.

2 pl -e is the accented outcome of *-e, which is also liable to be coloured by laryngeals and may appear as -a or -o with stems ending in *-h₂ or *-h₃, respectively.

Moods

Subjunctive

The subjunctive mood had a variety of functions in Galic. In direct discourse, the subjunctive most often expresses "neutral potentiality" without indicating the speaker's personal wish, standing in contrast to the optative that does so. In subordinate clauses, the subjunctive often expresses futurity rather than mere potentiality.

It is agreed the sine qua non of the subjunctive, across the Erani-Eracuran languages, is the thematic vowel, which in most daughter languages was added directly to the full-grade stem. In Early Galic, there is usually a contrast between the indicative stem and the modal stems which, synchronically, appear to be unpredictable. In the past, it was often thought the modal stem was an innovation tending towards abbreviation from the indicative, but more recent scholarship has preferred to view the indicative stem as more innovative. Differences in modal stems were levelled out in the transition from the Galic language to the Epic language by 650 BCE, always in favour of the indicative, and so a distinct subjunctive stem is also called the "Galic subjunctive".

In the Tennite languages, primary and secondary endings are applied to the subjunctive stem without discrimination or an obvious difference in meaning, while the other daughters exclusively apply the primary endings. This anomaly of the Tennite languages reminisces of the subjunctive endings used in Galic Northian, which are primary only by the addition of the hic et nunc particle *-i to the secondary thematic endings. "Genuine" primary endings are associated with the athematic indicative, partiuclarly 1 du ending -woiñi and 1 pl -máŋhi, contra subjunctive -ouuōḫ and -omōḫ.

All Erani-Eracuran languages other than Northian bind the thematic vowel to the full-grade stem, if a suffix appears, but in Galic it is bound often to the zero grade (with a few instances of the full grade). The zero-grade is otherwise used with the optative in the present, though the full-grade is found in the aorist. Thus Northian xšāuuasi < *ksn̥w-e-ti is to be compared with kṣṇávati < *ksnew-e-ti. To explain this deviation, one view is the zero-grade subjunctive was a back-formation from the optative, and another that the "zero-grading" of unaccented syllables was an active rule in Northian for longer than in other daughters.

The Northian subjunctive has informed some scholarly opinions that characterizes the subjunctive, in the parent language, as a derivational strategy rather than a morphological part of a verb.

Opatative

The optative is the other principal modality apart from the subjunctive. In terms of functionality, it expresses the wish of the speaker: if in the first person, the speaker wishes themself do something, and in the third, the speaker wishes the named or implied person do so. It is usually translated into Shalumite as "I wish..." or "would that..."

The optative is signified by the suffix -ī-, which ablauts to -yā- under the accent, and to which is added secondary endings. The accent of the optative is as follows: if the root took mobile accent, the modal suffix takes the accent from the root in the singular active, while the ending takes it in all other forms; if it took persistent accent, the accent remains persistent. The correspondence between the accent and the full grade form is totally predictable. Thus, for verbs with persistent accent, the suffix is always -ī-. While this morpheme looks somewhat like the feminizing suffix, they contain a different prehistoric laryngeal and are, as far as linguists are aware, not related.

Within the present system, the optative has the same stem as the present indicative. In the aorist system, it is formed from the root exclusively in Galic, though later texts may have the optative suffix added to the sigmatic stem. The perfect optative, like other modal forms of the perfect, is rare in Early Galic but becomes reasonably common in Late Galic and continues to be productive into Epic times. From whichever stem the optative is made, the secondary endings are always used, even in the perfect.

The behaviour of the optative in the present system thus differs from that of the subjunctive but is like it in the aorist. The significance of this divergence is still debated by researchers. At any rate, derivative verbs (that is, the desiderative, future, passive, future perfect, perfect passive, and causative) did not form corresponding optatives until the very end of the Epic period.

Imperative

The imperative mood encodes the speaker's demands. The difference from the optative, which encompasses the speaker's mere wish (which the speaker may or may not intend to be fulfilled), is encapsulated in the following timeless quote by Himinastainas:

It is lawful to say xāiiā̊ but not xāδí.

(It is lawful to say "I wish you would kill..." but not to say "kill!")

The Northian imperative is a fairly straightforward continuation of the parent language's largely-agreed imperative structure, where there are second and third person forms in the singular, dual, and plural numbers.

Conjugational paradigm

Unlike nouns, verbs may form more than one stem and be still considered the same lexical item, whereas nouns are typically restricted to one stem including any allomorphic variants (such as occasioned by ablaut), save suppletive items. It is thus necessary to discuss the relationship between the various stem-formations as they are attached to the root. The relationship between verb-stems and endings they receive are as follows in the Galic language:

From each root, which is agnostic as to part of speech in the proto-language, can arise multiple stems classified as present, aorist, or perfect, differentiated by their affixes. To these stems are attached endings to constitute the finite verb, which conveys tense/aspect, mood, voice, person, and number.

Tenses attested in Early Galic are in bold.

Present stem	Aorist stem	Perfect stem	Root
Present indicative
Present injunctive	Aorist injunctive	Perfect injunctive	Prohibitive
Imperfect	Aorist	Pluperfect
		Perfect
Optative	Aorist optative	Perfect optative	Second optative
Present subjunctive	Aorist subjunctive	Perfect subjunctive	First subjunctive
Imperative	Aorist imperative	Perfect imperative	Root imperative
Future imperative		Future perfect imperative	Root future imperative
Derivatives
Passive I	Passive II	Future perfect	Future
Desiderative		Perfect passive	Intensive
			Causative
Non-finite forms
Present active participle	Aorist active participle
		Perfect active participle
Present middle participle	Aorist middle participle	Perfect middle participle
Present active infinitive	Aorist active infinitive
		Perfect active infinitive
Present middle infinitive	Aorist middle infinitive	Perfect middle infinitive

Stem shapes

The shapes of the stem in the Erani-Eracuran parent language decides the resulting forms in Northian:

• VC—these take the short (regular) endings.
• VH—these take the short (regular) endings in the strong forms and long (augmented) endings in the weak forms, with any laryngeal-colouring effects visible on both stem and endings which interface with the stem-final laryngeal.
• VCH—these take the long (augmented) endings, with any laryngeal-colouring effects visible on endings which interface with the stem-final laryngeal.

Aside from the shape of the stem, there are two accentual patterns found in present stems:

• Mobile accent—this pattern is found within all stem-classes and is the most common by far; the accent is on the stem in the singular active and on the ending in all other forms.
• Persistent accent—some root stems take this accent pattern, and the accent is on the stem in all cases; the singular indicative and imperfect active usually has a lengthened vowel, in contrast to the short vowel of all other forms. The injunctive can sometimes have short vowel throughout, but this is completely unpredictable. The 3 pl form, which is sensitive to ablaut, takes the zero grade form.

Present stems

The present stems, with imperfective or durative aspect, appears to have largely continued the system of reconstructed Proto-Erani-Eracuran. With regard to the history of Northian, the imperfective stem was the most productive, buttressed by many derivational strategies that specified this aspect. Each strategy designates either athematic or thematic endings be used, which are synchronically arbitrary.

After the influential grammarian Praetorianius (fl. 1200s), the present stems in Northian are classified by the size of their roots and endings; the most extended stems appear first, and the most extended endings last. Thus, the fully-reduplicated intensives are first on the list, then the partially-reduplicated present stems, and then the root stems. Praetorianius regarded any derivational suffix as part of the ending (he called them "augmented endings") and listed the shortest suffix first and then longer ones. Where ablaut changes the length of these elements, the longest form determines the position on the Praetorianian list. Lastly, the athematic classes come before thematic, as athematic -mi is one letter longer than thematic -ō.

At the heart of Praetorianius's understanding of the verbal system is his analysis of all finite verb into a combination of a xaput "head" (giving meaning) and a ϑénū "body" (giving function). On the one hand, this gave him the insight that some roots, superficially different, are really varieties of each other, differing only by a vowel; this insight peers into the same phenomenon as that termed the guṇa and vṛddhi grades of Kankrit grammarians. On the other hand, Praetorianius did not segment his "bodies" into suffixes and simpler personal endings, resulting in many co-equal sets of "bodies" that not only blended the suffix and ending but sometimes even part of the etymological root itself; that is, to Praetorianius, the 1 sg endings -āmi and -mi were equivalents, as much as -mi and -ō were.

Phonetic opaqueness and the loss of semantic difference in suffixes are responsible for his inability to analyze his "bodies" further. As an example of this mis-segmentation, Praetorianius analyzed the word nauuāsi "renews" as nauu-āsi but also ϑištāsi "stands up" as ϑi-št-āsi. In the former case, the suffix (now known to be descended from the factitive suffix *-eh₂-) along with the ending is understood to constitute the "body"; yet in the latter case, the ā is actually part of the etymological root. The pattern of singular -āsi and plural -ité is strengthened by other "bodies" that have similar shapes, such as the -nāsi and -nité endings, which he also did not further segment.

Many athematic stems display ablaut, and in Northian it is customary to give the first person singular active to show the form of the strong stem and the second person plural, the weak stem. Where a stem does not display ablaut, only the first person singular active is given. These forms, that is 1 sg and 2 pl, are chosen with the view they might exemplify the more frequent superficial forms in the received texts. Aside from them, the 3 pl is regularly irregular owing to the vowel-initial shape of the ending, and the 1 du triggers some (at least) orthographic variations that are, for the most part, well-described by simple rules.

#	Class	Pre-form	Characteristics	Examples
I	Intensives	CoC-CéC-mi	These denote repetitive actions. Historically, these are certainly derived stems, but Praetorianius understood them as separate lexical items. Intensive stems do not always display ablaut, and if they do, it tends to be confined to the syllable immediately before the endings.
II	Reduplicated	Ci-CéC-mi / Ci-CØC-té Cé-CeC-mi / Cé-CØC-ti	This stem type is associated mostly with root aorists. They always show root ablaut, varying between *e ~ Ø. The accent is on the root if it is in full-grade, and on the ending if the root is in zero-grade. A sub-type of this class, not associated with root aorists and not separately classified by Praetorianus, had an accent that persisted on the reduplication vowel, which is /é/ rather than /i/.
III		Cí-CØC-ōḥ	Like the athematic verbs in class III, these are imperfective verbs formed from eventive roots, but with thematic endings. The root syllable is in zero-grade. The accent is consistently on the reduplication syllable.	pí-p-i- "to drink"
IV	Nasal infix	CR̥-né-C-mi / CR̥-nØ-C-té	Associated with root aorists, the affix *ne ~ n is inserted between the vowel and final consonant of the root. The root itself is always in the zero-grade, while the infix undergoes ablaut, which likewise has a full-grade vowel in the singular active and zero-grade elsewhere. Roots which take this infix obligatorily have a resonant before the infix, and the resonant vocalizes but never the infix itself in zero grade. Praetorianius classified roots ending in laryngeals with this formation with the -nā́mi class, as to him the long vowel generated was part of the ending and not the root.
V	Root	CḗC-mi / CéC-ti	These roots show a lengthened root vowel in the singular active and the full grade in all other forms. The accent persists over the root syllable in all forms, at least in the indicative.	ēs- "to sit, stay"
VI		CéC-mi / CØC-té	These roots, the commonest, show a full-grade vowel in the root in the singular active and a zero-grade root in the other forms; the accent is over the root where it is in full grade, and over the ending when the root is in zero grade.	es- "to be"
VII		CØC-ṓḥ	This marginally-attested class binds the thematic endings to the zero-grade root, which have aorist origin. Praetorianius called them "like-subjunctives" after subjunctive forms of other verbs, which could also take the zero-grade root.
VIII		CéC-ōḥ	This even more marginally-attested class, of which only six members are known, has thematic endings to the full-grade root. Its rarity stands in contrast with its frequency in other Erani-Eracuran languages, and none of the six members have clear cognates at the stem level in the language family.
IX	-imi / -umi	CéC-i-mi / CØC-i-té CéC-u-mi / CØC-u-té	A number of present stems have a non-ablauting -i- or -u- added between root and ending, of no discernable function. In the 3 pl it causes the ending to become -ónti. All verbs of this class that have active forms have mobile accent, but some deponents in this class have persistent accent on the root instead.	tāi- "distribute"
X	-aⁱmi	CéCH-mi / CØCH-té Cé-CH-mi	These verbs have a stem-final laryngeal following a consonant. This laryngeal vocalizes regularly and is mutated by any following vowel /i/ separated by a resonant. There are also reduplicated verbs behaving this way. The special case is for reduplicated roots ending in laryngeal that take pesistent accent: as the root never takes the full grade, the zero-grade root behaves exactly like full-grade roots ending in a stop plus laryngeal. They may also be regarded as a sub-type of the stems that have an ablauting vowel in front of the laryngeal—their zero-grade forms are identical.	ϑé-δi- "give"
XI	-ā́mi	CéH-mi / CØH-té CVC-éH-mi / CVC-ØH-té	This class is a combination of multiple types of verbs that ended in an ablauting vowel and laryngeal. They descend from suffixed factitives in *-h₂-, statives in *-eh₁-, and any non-suffixed roots that merely happened to end in the same sequence of sounds. Praetorianius wistfully notes the same vowel of the singular active "haunts" the 3 pl and subjunctive. *-h₂- attached to e-, u-, and i-stems created stems in -ā́mi, -ūmi, and -īmi. *-h₂- also caused a following -t to become aspirated (thus 2 du act -āϑāḫ), but -h₁ did not.
XII	-ā́mi	Ci-CéH-mi / Ci-CØH-té	The same, but with reduplication.	ϑi-št-ā́- "stand up"
XIII	-nā́mi	CØ-néH-mi / CØ-nØH-té CØC-néH-mi / CØC-nØ-té	These behave in exactly the same way as V-VI, except with an extra -n-.	zrβ-nā́- "seize"
XIV	-(n)ammi	Céw-mi / CØw-té CØC-néw-mi / CØC-nØw-té	This class is also underlyingly the same as V-VI, but with an extant *u taking the place of the prehistoric laryngeal. It also contains both roots that naturally end in -w and suffixed stems ending in the same letter, which apophonically behave identically. Preconsonantally, there are several phonetic variants, being 1 sg act -nammi, 2 sg -naušti, 3 sg -nəōti. In weak forms, the suffix is -nu- preconsonantally and can become -nw- or -auu- prevocalically, e.g. xrauuəṇti < *krn̥wenti "they make". This class of verbs typically (but not universally) drop the suffix to form the non-indicative voices, thus xarōṇti "they may make" xariiāṯ "would that they make", etc.	xarə-naw- "to make"
8a	y-	CéC-y-ōḥ	This very prolific suffix creates present stems from aorist roots. They characteristically had a full-grade and consistently-accented root. Under Sievers's law, this suffix has the allomorph -iy- after heavy syllables, that is to say those containing a long vowel or closed by two or more consonants.	špah-y-e- "observe, spectate"
8b		CØC-y-ṓḥ	Similar to the above, but this type had a zero-grade root and consistently accented endings. Together with this fact is observed that the root aorists from which these stems derive most often lack an active voice, even though their meanings may be active; the zero-grade root characterizing root aorists middle voice is held to explain the presence of the zero-grade root in the derived present stem, even if it acquires an active voice there.
9	šk-	CØC-šk-ṓḥ	Creates imperfective stems from perfective roots.
10	s-	CéC-s-ōḥ	Only a few examples are known from this class.
12	áii-	CóC-ay-ōḥ	Stems of this class have root syllables in the o-grade.
15	aii-	CVC-ay-ṓḥ	Creates verbs from o-stem (2nd declension) nouns.
16	āii-	CVC-āy-ṓḥ	Creates verbs from ā-stem (1st declension) nouns.

Perfective stems

The perfective stems, also aorist stems in some literature, are a relic class in Epic Northian, but they are more plentiful in the Galic language. When used in the indicative, there are four formations as follows. But when used outside of the indicative, and particularly in the injunctive and optative moods, aorists in s- are often "conjugated from the root", that is behave like root aorists. Roots which have this behaviour are called "split aorists" and are associated with a contrasting ablaut pattern, with only the third plural exhibiting a weak grade.

#	Class	1 sg. / 2 pl.	Function	Examples
1a	root	CéC-(m) / CØC-té	Contains roots that are inherently perfective. Most perfective verbs in the parent language appear to have been of this type, and Galic Northian attests hundreds. The root syllable typically shows ablaut between *e ~ Ø. Many suffixed imperfective stems are derived from a root perfective stem.	dō(i)- "give" xr(a)w- "hear"
1b		CḗC-(m) / CéC-té	The same, but with ablaut variants in long and short grades. This type was comparatively rare, about as rare as the long/short present stems to the normal type. In Northian this type was often in free variation with the s-aorist, which only exists in the indicative; that is, the modal forms of class 4 are in 1b.
2		CØC-õ / CØC-té	Contains roots that are inherently perfective. This type appears to have been rare.	luθ- "listen"
3	reduplicated	Cé-CØC-õ / Cé-CØC-ti	This pattern is known from only one root, though it is widely-attested in other Erani-Eracuran languages.	və̄t- "say"
4	s-	CḗC-s-(m) / CéC-s-ti	The s- suffix creates perfective stems from roots which were inherently imperfective; notably, the ablauting root syllable contrasts the long vowel in the strong forms and the short in the weak.	daik-s- "point"

Stative stems

The features of the perfect stem in the proto-language were reduplication and the o-vocalism in the perfect indicative. This is true of Northian as well. In the Epic language, the o-vocalism has spread into the subjunctive, but in Galic the subjunctive has the inherited e-vocalism.

#	Class	1 sg. / 2 pl.	Function	Examples
1	root	CóC-a / CØC-é	This type is known from only one verb, namely u̯oi̯d- "to know". The weak stem is u̯id-. Accent is on the root syllable for strong forms, and endings for weak forms.	u̯oi̯d- / u̯id- "to know"
2	reduplicated	Ce-CóC-a / Ce-CØC-é	Creates stative stems from perfective and imperfective roots. An extra syllable is prefixed, consisting of the consonant immediately preceding the next vowel and the vowel *e. Due to phonetic change, this vowel can appear as o following a labialized consonant or become more obscure if a lost consonant such as a laryngeal is reduplicated, cf. ā̊ha "I am desiccated" < PEE *h₂e-h₂oh₂s-h₂e. The strong forms have characteristic accented o-grade in the root syllable and zero-grade there in the weak forms with accented endings.	ve-voi̯d- / ve-vid- "to believe" ko-kor- / ko-kr- "to be working" ā̊s- / ā̊s- "to be desiccated"
3	long-vowel	Ce-CṓC-a / Ce-CóC-i	This type may have developed in Northian on the pattern of the sigmatic aorist as it has no like cognates in other branches of the language. It contrasts a long vowel in the strong forms and a short vowel in the weak, and the vowel is often, but not always, the o that is characteristic of the perfect.

Historical development

Periodization

The Galic corpus, divided by genre, is usually analyzed to obtain three language periods, named Early Galic, Late Galic, and Didaskalic Galic. Early Galic is the language of the Period I Gales and any scattered Old Material in them, Didaskalic Galic was the language of the Didaskalic Material, and Late Galic refers to everything else.

Compensatory lengthening

In the prehistory of Northian, which includes the Erani-Eracuran proto-language, there was sweeping compensatory lengthening of a preceding vowel following the dropping of final resonants. These long syllables then generally lost their final consonants in pre-Galic, creating preponderating quantities of open final syllables for which Galic is known. Named Szemerenyi's law, this dropping of final resonants is reflected in one way or another in all Erani-Eracuran languages.

On the one hand, owing to its early attestation, Northian reflects the law's operation with greater regularity and fewer analogical restorations than most other languages; on the other hand, Northian presents idiosyncratic restorations that post-date the split-up of the proto-language and the generalization of some outcomes of Szemerenyi's law as regular forms, in the process probably deleting an *-s restored in the parent language. Thus, the absence of final *-s is not conclusive evidence that Szemerenyi's law operated in that phonetic environment in the parent language or that *-s was not then restored. A second wave of dropping final resonants in Pre-Galic has rendered many forms open to different interpretations and possible analogy.

For example, the genitive huiiō from huiiuš is usually held to be evidence that the genitive *-s was deleted in the proterokinetic u-stems *-ow-s > *-ōw > Northian -ō. While outwardly this form is similar to the hysterokinetic ending *-os > Northian -ō, the possibility that this ending was transposed to the u-stems can be excluded because huiiō never shows final *-s in sandhi, while the -ō ending from *-os does. Thus, the operation of the law is consonant to the deletion of final -i and perhaps part of the same process. Yet all the other daughter languages show final -s or a reflex thereof, so Northian is held to be archaic.

But in the case of zmerū from zmerun "marrow", the final long syllable would reflect *-ūn < *-un-s, which is unexpected in the proterokinetic paradigm (expected is *-wen-s > *wən-h). *-un-s must therefore have arisen from an otherwise-unattested acrostatic form, if it existed in the proto-language. Linguists are therefore ambivalent to ascribe zmerū as an archaicism and instead are wont to think of it as a non-ablauting innovation. The observation that most un-stem terms are Northian derivations buttresses that assessment, even though the root of this word is a sure inheritance from the parent language.

In principle, the following final consonants are dropped at the Proto-Northian level:

1. *-s
   • always after *w- and *y-
     • gen huiiōʷ < *huyōw "child's"; gen mātō < *mātōi or *matēi "mind's"
   • after *un- and *in-, but not after *en- or *on-
     • gen hmerūš < *smerun-s "marrow's"; gen parī < *parin-s "doorway's"
     • gen huwənh < *sh₂wen-s "sun's"
   • not dropped after *m-
     • gen tə̄mps < *dem-s "house's"
2. *-i
   • after *w- and *y-
     • loc huiiō < *huyōw "in a child"; loc mātō < *matēi "in mind"
3. *-h₂
   • after another laryngeal -h₂ is dropped without a trace
     • nom du fóṇtōi ahō tuš-tə "two paths, [one] of good and [one] of evil"
   • after *n-, *r-, and *s-

Vocalization of resonants

Like most Erani-Eracuran languages, Northian vocalized the resonants that took the place of vowels in their absence. But in contrast to most other languages, the vocalized resoants (at least as they appear in manuscripts) apparently had no phonemic length and could be metrically long or short as required. Older texts usually state the vocalized resonant was long in initial and final position, as well as internally except around laryngeals or nasals, but too many exceptions peppered even such a specific rule. The matter was only formally addressed in the 1950s, since the <a> resulting from *n̥ was clearly different from the <a> resulting from, for example, *h₂e; a more recent view provides for residual consonantal quality when the Gales were composed, such that <a> was /ā ~ an ~ ąn ~ ən/, which quality disappeared centuries later when committed to writing.

Inter-galic changes

The language of the Gales, across its three generally-recognized periods, is internally consistent, whose morphologies and vocabularies have consistent meanings. Nevertheless, scholars have argued there have been diachronic differences amongst the three periods, seen most clearly in the creation of suffixed subjunctive formations, first attested in Period III (c. 1200 BCE) material and used in conjunction with root formations of the subjunctive. Most descriptions of Galic language considers it to be fully contemporary to the Galic poets, since few, if any, difference in register are detectable. Little study has been done for the False Gales and the Sacerdotal Verses, though they are thought to be alike to the Gales.

Between the Galic poetry and prose, exemplified by the Didaskalic Material, a noticeable change is the disappearance of the sigmatic present stems, e.g. fā-s- "protect" and wak-š- "say", resulting in the confinement of that sign to the aorist aspect. The only s-stem that still took primary endings in the DM was rax-š- "protect".

Galic to Epic changes

It has been well noted that the change from the Galic to Epic language was a gradual but non-linear process that mostly occurred between 1200 and 600 BCE. This is based on the assumption that at least the Epic language was contemporary and not an artificial dialect. There is considerable difference in opinions whether the language of the Gales was spoken natively by its poets. On the one hand, there seem to be mild changes in grammar that suggests the language was not dogmatically taught; on the other hand, analogical replacements are nearly unknown, and exceedingly opaque forms surface when their replacement should have been only too certain. For example, ā̊ "mouth" has gen hōḫ, loc hā̊ etc.; forms are so perfectly regular as to be astounding.

In terms of morphology, the Epic language outright lost very little of the richness of the Galic language and, additionally, continued to reflect faithfully the grammatical peculiarities occasioned by ablaut. A few immediately-noticeable changes can be interpreted solely phonetically, e.g. Epic 3 sg inj štōd "praises" vs. Galic štəōd "id", where the diphthong /aw/ has monophthongized to /ō/. This process must have commenced before the earliest Galic, since all diphthong have weakened allophones in front of stops compared to full forms in front of vowels. Compare Galic 2 sg štō < *staws, with regular dropping of final *-s.

The main difference noted by linguists was in the frequency and productivity of various formations. Like other Erani-Eracuran languages, most athematic stems and their derivational strategies became vestigial and unproductive; the exception was the nəō ~ nu suffix, which was attached to a variety of both native and borrowed terms.

Syntax

Copula

In Early Galic, nominal sentences were the normal construction to express the sense of equivalence or identity as found in the word "is", e.g. GNr 112 maxrō tu-at "but thou art tall". The finite verb ešti is only rarely used for this function. When it does appear, it often connotes contrast with a previous statement.

See also

• Northian language
  • Northian nominals
  • Northian verbs