History of Northian
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.
Pre-Galic and Proto-Northian
The term "Pre-Galic" is often interchangeable with "Proto-Northian" in literature. Typically, a Pre-Galic form is posited to account for alloforms of the same word whenever it is possible to do so within the history of Northian itself, by both phonological and morphological means. Such reconstructions are motivated by the divergent behaviour of various word-final sounds in interaction with following sounds, known as sandhi, e.g. the Pre-Galic form *zniδriyah is required to account for the sandhi form xanitriyaš-tə "and genitrices" and the independent form xanitriyā "genitrices". But it is cautioned that not every instance of sandhi can be considered proper to Northian, and a number of sandhi rules are probably present already within the parent language.
In the strictest sense, Pre-Galic is not considered by modern scholars a stage of language evolution local to any point in time; rather, it is the underlying phonological word-forms that, given fixed phonetic rules, generate the attested word-forms. Words like *zniδriyah may or may not have really been spoken by Northian speakers, and its evolution towards the attested forms xanitriyaš-tə and xanitriyā are not necessarily best explained by regular and sequential sound changes.
Some authors posit a separate Proto-Northian stage that represents the common ancestor of various Galic dialects. But as the existence of Galic dialects, as opposed to individual styles, is yet controversial, so too is the idea of Proto-Northian. Higher-order reconstruction work such as Proto-Nordic-Northian usually begin with Galic and does not take into account any dialectal variations between that and other Northian material, assuming that all subsequent forms of Northian developed out of a monolithic Galic language.
Other than the restoration of word final *-h, a word-final laryngeal following a vowel is usually accepted as well, as there are sporadic cases of a word-final laryngeal failing to create a long vowel in that position. Nr. 1004 has nom.sing humă | < *dumeh₂, where the laryngeal in auslaut has evidently been deleted before a natural pause and therefore could not lengthen the final -a. This fact alone could probably not justify writing this word as *humaH, but in intervocalic position the laryngeal must have survived into Galic times to cause the trisyllabic scansion of *wl̥kiH-es > urəkiHah "wolfess"; without the laryngeal in position, standard scansion would result in *urəz(i)yah, where the parenthetical /i/ is per Sievers's law and not a phonological syllable.
That being established, there is a question whether the laryngeal was an undistinguished *H or whether the (likely) three laryngeals of the parent language remained separate. Such is difficult to decide because most, if not all, of the words with an intervocalic laryngeal and attested in Galic turn out to be non-probative. The gen.sing ending -os cannot be coloured by the laryngeal, and the dat.sing ending regularly colours a preceding *e, so the effects of an *h₂ (which would be distinct from *h₁) cannot be verified. An inherited word with a vowel preceding the laryngeal would have been coloured in the parent language and thus cannot tell whether the laryngeal was distinct down to Pre-Galic times, and new coinages appear to waver. There is thus some reason to believe that intervocalic *h₂ was at least phonetically different from *h₁ for some speakers, but whether they were phonologically and grammatically different cannot be known for sure.
Galic
Early Galic
Early Galic is the language identified with the Period I Gales (G1), which are typically dated to 1750 – 1500 BCE, with perhaps some turns of phrase (the Old Material) predating and some emendations or recensions postdating that range. Being literature composed for solemnity and ritual, it is usually assumed that it is done in a highly polished register. Particularly, since Galic rituals usually involve two or more parties that may represent roving lineages or tribes, their shared literature is likely to have been composed in such a way to exclude innovations that would be alien to one or more parties, thus preferring the more archaic over the more innovative. This eye to conservatism is perhaps responsible for the highly uniform language that preserves very archaic morphology.
Aside from grammatical features, Early Galic has certain unique phonological features. While all Northian texts attest to the lengthening of final *-u to -ū, only Early Galic also has final *-i to -ī.
The Early Galic verb has formerly been analyzed according to the tripartite system of present, aorist, and perfect systems plus subsidiary stems, but though this analysis is not impossible, it has been found somewhat limiting and inadequately reflect the peculiar features of Galic and its diachronic developments. In works after the 1980s, it has been commoner to assume a more basic, bipartite system between the present-aorist and perfect. This re-analysis is founded on that present and aorist stems share a single set of endings, which stand in contrast to that of the perfect, and the only material difference between them is that only present verbs may take the primary marker *-i.
It should be cautioned that, owing to the limited size of the Galic corpus (only 413 verbs in 8,786 words), it is often difficult to conclude whether a given root is present or aorist. The root aorist is often left unaugmented in Galic, while the imperfect is usually augmented but can also appear unaugmented; unless the present be also attested, an unaugmented imperfect could not be distinguished by mere form from a root aorist. In this case, comparative information and semantics are relied upon to make the determination, but Galic verbs are at variance with the other Erani-Eracuran languages as to a given root's present/aorist affinity in a handful of instances.
The proportion of stem formations in Northian is such that the root aorist (including normal and xaŋzat types) accounts for three quarters of all aorist stems, while root present stems (normal and long) account for only about a third of present stems. Over 90% of all Galic verb stems are athematic, and in the aorist this percentage is 96%. There are no thematic stems built from the root that can be considered original in the present. This held to be a salient structural difference from Kankrit, where root stems preponderate in both categories and thematic root stems particularly in the present. Thus, while similar (and cognate) derivation strategies appear in both languages, their respective prominence differs greatly.
While in Northian some root presents with ablauting roots were certainly remade to suffixed presents ones with ablauting suffixes, reminiscing of the shift from amphikinetic to hysterokinetic nouns, the small corpus of root present stems (amongst which are very fundamental roots like Hes- "be") could also be understood to mean some root aorist were re-interpreted as root presents by the addition of hic et nunc -i.
When a verb was transmitted to the other aspect via a derivational marker, i.e. present to aorist or vice versa, only indicative and not modal forms are generated. Thus, there were sigmatic aorist indicatives but no sigmatic aorist subjunctives or optatives; when these were called for, they remain identical to the non-sigmatic present subjunctives and optatives. This behaviour is comparable to that of secondary derivations, which do not have their own modal forms and rely on those of the more primitive stem. Thus, a naw-present like gr̥nammi "I am making, I habitually or often make" was akin to a secondary form to the root aorist xarā "I make", just as the co-ordinating future xarištō "I hope to make" was.
This does not mean that a system of grammatical aspect did not exist, only that in ancient times roots were rarely used outside of their inherent aspects. On the contrary, there was a stringent rule that some roots could take the hic et nunc -i, while others not; if this was needed, the roots that could not were first marked by a derivational marker, as the -i was never to be added to such roots directly. This would appear to underlie the later tendency to re-characterize/derive both present and aorist roots with markers that were explicit as to their aspectual affinity (in most cases, the ablauting suffixes in the present and -s- in the aorist), even if the roots to which they were attached already implied this aspect.
Northian presents clues for the h₂e-conjugation theory. Directly preserved h₂e-conjugation endings are few and then only in the plural: 1 pl attested but once as -me, 2 pl as -s and -š(te), and 3 pl as -r more generally in the injunctive forms of xəŋzat-aorists and pluperfects.
Indirect clues in the form of mismatches with the mi-conjugation are more frequent. When the h₂e-aorists were renewed in the prehistory of Northian, forms of the mi-aorist became specialized indicative forms, while the old h₂e-aorist lingered as injunctives. The indicative mi-forms then slowly replaced the injunctives, leaving only the 3 pl in -r attested in situ. But the mi-forms created showed root e-grade in the 3 pl, which is only explainable via the e-grade in the 3 pl of the h₂e-aorist; otherwise, mi-conjugation root aorists had zero grade in the root there. Nevertheless, the h₂e-aorists may have played a role in the derivation of perfects stems with o-grade stems in the 1 & 2 pl, and the very archaic perfect imperative seems to reflect the imperative of the h₂e-conjugation.
It is a matter of active discourse if h₂e-stems had the same present/aorist distinction that m-stems had, and if so, if their primary derivatives also retained h₂e-endings in the other aspect. If they did, then it is probable that the markers used were distinct from those used for the m-stems, given the concentration of the i- and u-presents to xaŋzat-aorists, which are now recognized as descendants of the h₂e-aorists. However it should be noted that h₂e-aorists as a category that gave rise to xaŋzat-aorists are diagnosed only by their usage of the augment, and their contrast to h₂e-presents cannot be explicitly found since there are no obvious remnants of h₂e-presents in Northian.
Many verbs in Galic are defective in the middle or active. Indeed, philologist Kremann has observed that "non-defective verbs are exception" and called them "bivalent verbs". Such defects are particularly apparent when they are remedied in the late, prosaic Didaskalic Material. Verbs defective in the active are common in the other daughters, but those defective in the middle are only common in Northian. Such verbs are often cited as members of the h₂e-conjugation, primordially linked to the middle, but renewed with mi-endings in Northian. So, h₂e-turned-mi Northian verbs could not have an inherited middle voice, and where it is found it always shows signs of lateness, like the 3 pl ending -ā̆tro where -ro would be expected for an inherited form. But this explanation is only possible for some activa tantum, and others are ostensibly mi-verbs with firm word equations with mi-verbs Hittite.
There may be some h₂e-conjugation verbs that became media tantum, but this mechanism has received far less elaboration owing to its obscurity, though h₂e-conjugation endings are hardly extricable from the middle. Recently, Brent has argued that the entire perfect was formed on the basis of h₂e-conjugation verbs by analogy to reduplicated present verbs on the basis of mi-conjugation aorists. He further says all h₂e-verbs that have not become activae tanta automatically became perfects by assimilation within Northian, to explain the large number of perfect stems in Northian that have the o-grade root in the 1 pl and 2 pl. Brent points to the unreduplicated pluperfects mąšte and wizər[āt] as "formal identities with h₂e-aorists".
The earliest Gales could be dated to the same period as Early Hittite, but the Gales present only badly-mangled vestiges of a former h₂e-conjugation, while Hittite retains it as a fully productive category. This is held to indicate that Northian is much more closely related with other Erani-Eracuran languages.
Late Galic
Late Galic is defined as the language of the Period 2 and Period 3 Gales (G2 and G3), which occupy about 25% and 60% of the transmitted Gales. These are dated to around 1500 – 1300 BCE (G2) and 1300 – 1200 BCE (G3).
The language attested in this part of the Gales is very similar in structure to that in the earlier part of Period 1 Gales, but it has many grammatical and phonological differences. At one time it was thought Late Galic was simply a more evolved form of Early Galic, but this analysis has been demonstrated to be untenable in many ways. While some formations are certinaly to be sourced in Early Galic, there are also formations that do not appear in Early Galic or require some sort of internal reconstruction of Early Galic to obtain. Some of the more obvious examples are e.g. u-stem voc sing ending -aw as opposed to Early Galic -ō (though the sandhi behaviour in EG is as *-aw). Besides sandhi difference, the regular oxytone dative ending -ē in Late Galic can be seen as a monophthongization of Early Galic -ay, but where Early Galic has -ayay in the paroxytone i-stems, Late Galic has -ā < *-ay-i and not *-ayā < -ayay.
On the grammatical front, there has been significantly different treatment of verbs and nouns. Some active-only verbs in EG appear as middle-only verbs in LG, even if their meanings are exactly the same; the perfect imperative of EG is completely remade in LG, without leaving a single trace. Some EG root aorists reprise as sigmatic aorists in LG, though they retain their EG modal forms. The preposition *up appears as upū in EG but upā in LG; the former governs the ablative and genitive cases, and the latter only the ablative.
The grammatical consistency within the Late Galic corpus reminisces of that in the Early Galic corpus. Thus, grammatically, they present to two rather distinct foci. If the Early Galic foci is to be interpreted as a liturgical commonality within the tribes of the G1 period, some have inferred that the G2 and G3 periods represents a independent commonality that had important connections, geographic or temporal, to the G1 commonality. This is because the G2 hymns contain some themes, concepts, and items that are not found in the G1 hymns, and vice versa. For example, hymns to ϑaˀā (Earth) are almost exclusively found in G1, and the figure of Zyō (Heaven) is entirely absent from G3.
Perhaps the most compelling evidence to qualify Late Galic as something other than a more evolved form of Early Galic is their liturgical usage. In the transmitted Galic canon, the G1, G2, and G3 hymns are arranged in a very deliberate order that is anything but random. From G. 61 to 85 there is an unbroken block of 24 hymns that are all from the G1 period and dedicated to different gods. The block could not have coalesced by sheer chance or arisen by reference to the divine dedicatee. Astonishingly, G1 and G3 hymns do not border each other in the Canon, another fact that could not be due to chance. A particular variety of G3 hymn, dedicated to "all the holy gods" or ATHG, opens and closes the Galic chant ritual; this kind of hymn is evidently the latest type and rarely appear outside of the two ends of the canon. There is thus a very ancient awareness that G1, G2, and G3 hymns have distinct liturgical profiles, though what those were have been lost to time. In the modern Fentoi Wisto psalter, the Gales are chanted continuously, without ritual actions.
Epic
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 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 expected. For example, ā̊ "mouth" has gen hō etc.
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 3sg inj štō vs. Galic štawd "praises", where the diphthong /aw/ has monophthongized to /o/. This process must have commenced before Galic, since all diphthong have weakened allophones in front of stops compared to full forms in front of vowels; in the received text, which is phonetically oriented, this weakening of preconsonantal diphthongs is spelled differently from prevocalic and auslaut positions. The Galic vowel /ā̊/ becomes /ō/ in auslaut and /ā/ elsewhere.
The aorist system experienced a process of renewal earlier in the Epic period. The root aorist was the main aorist formation of the Galic language, but it had several varieties that owe above all to stem-final laryngeals. If a root ended in vowel and laryngeal, the root aorist appeared to have a long vowel before the ending in the singular active, which in 1p & 2p absorbed the personal ending, giving Epic -ō, -ō, -ās, e.g. štō "I/you stood", štās "he/she/it stood"; if it ended in a stop plus laryngeal, the laryngeal was vocalized and gave the endings -ā̆, -š, and -t ~ s in the singular. The dual and plural endings of both were identical, because in the former situation, as the root ablauted to the zero grade (2 pl act *-iϑé), the laryngeal always vocalized, just as in the latter (2 pl act *-iϑé).
If there was no root-final laryngeal, the endings -ā̆, -s, -t were encountered in the Gales. These endings were intensely prone to disfiguring the root as they were non-syllabic. Some roots adopted the short laryngeal endings, but more were transferred into the thematic aorist category, with endings -õm, -ā, -et; they are distinguished from genuine thematic aorists by their full-grade stem, the latter having a canonically zero-grade root.
The sigmatic aorist with -s-, which signified roots of present origin for aorist tense, also developed several varieties based on the shape of the root. All inherited sigmatic aorists have long-grade root in the singular active and full-grade elsewhere, and this contrast is preserved in the Epics. Once again, the laryngeal-final endings prevailed, and the 2p & 3p forms were whence introduced and used indiscriminately, thus áxānišṯ and áxānṯ, both "he/she/it sang". The 1p form in -zam or -ham (depending on the root shape) seems to have been left out of the replacement process and persisted through the Epic period.
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 three ablauting suffixes.
Post-epic changes
The language changed rapidly in all ways after the Epic period, with some perhaps having begun during the Epic period, to judge from Northian epigraphy that begin to appear more abundantly after the 2nd century BCE. Perhaps this suggests the Epic dialect was archaizing in some respects, which would be consonant with the stylistic premise of telling a story originating from long ago. Some authorities have sought to connect this change with the mass expulsion and consequent migration of Northians to the modern territories of DNS.
The largest post-Epic change was the replacement of athematic verbs with "activized" passive verbs in -y-. A verb like štuyōi "I am stood up (by)", which is the y-passive of the u-stem present štaHumi, came to acquire thematic active forms štuyō. štuyō then came to replace štaHumi in the active voice. Verbs like štuyō were also defective in the middle voice, since its middle forms were already used to form a specific passive. This defective middle then presaged the decline of the middle voice as a whole. The timeframe for this replacement is difficult to ascertain, as it is absent in the Epic corpus but common in Late Canon materials, even as they briefly overlapped in time; perhaps its use was dispreferred in poetry. It is likely to have existed in Epic times, since the commensurate co-ordination of an agent in the genitive case may underlie some Epic constructions.
The sigmatic aorist sign -s-, added to roots of present origin to produce an aorist stem, became a general preterite sign, built to most stems that ended in -y-, giving a composite past-tense stem in -ās-, to which was added the thematic secondary endings. The thematic aorist became productive during Epic times and left a considerable number of non-sigmatic past tenses. The imperfect to the present stem loses its vitality apparently quite early, and the sigmatic aorist itself disappeared whenever its marker was transferred to the present stem; however, both formations leave detectable relics in certain lexical items. The synthetic perfect, other than the item woyd- "know", remained in suspended animation for some time, only slowly losing ground to the periphrastic perfect based on the activized version of the perfect passive participle; the perfect active participle, on the other hand, disappeared almost completely after the Epics.
The subjunctive mood was already losing ground in Epic times and gradually became confined to frozen constructions. The future meaning of the subjunctive was transferred to a synthetic future tense already in the Epics. The other meanings of the subjunctive such as potentiality or uncertainty became indistinguishable from the optative, which became prominent in the Epics and very nearly supplanted it after then. The subjunctives of a few verbs survives in expressions indicating doubt or hesitation in Imperial and Medieval Northian. The third-person imperative is survived by the third person future imperative, which are often used in oaths and contracts; the future imperative for the second person disappears. The various uses of the injunctive were completely lost.