A sketch of Ājat he-Heloun

Ājat he-Heloun is a language of Akana, in the Edastean family, and an immediate descendant of Mavakhalan, spoken in Khalan and its environs. This sketch describes Ājat he-Heloun (henceforth AhH) as it existed circa 1700 YP.

Contents

Phonology

Segment inventory

The consonants of AhH are arrayed in the following table, together with their romanizations. (The headings in the tables are not attempting to be phonetically precise.)

The existence of ṗ ṗh as separate phonemes /pf pfʰ/ is a feature confined to nonstandard dialects (especially those spoken nearest where Ayāsth is), and is generally stigmatised. In the standard they fall together with p ph /p pʰ/.

On the other hand, the rendering of d as [d̪] is strongly acrolectal, and I use the letter d as much for its convenience as a cover symbol as anything else. In common speech d either falls together with one or more of l r ř j or zero, or is distinguished as [ɮ], according to dialect. The most progressive dialects go one step further and eliminate voicing contrasts altogether by merging z ž into s š; in these d may be [ɬ].

The monophthongal vowels are:

There are also a number of falling diphthongs, yi ei əi ai ou au. The nuclei of these diphthongs have the same qualities that they do as independent vowels, and the offglides i u are [ɪ̯ ʊ̯].

[[Include frequency charts.]]

Phonotactics

Syllables are minimally V and maximally CCVGC, where G is the offglide of a diphthong. Note that when v j occur in a coda they're to be distinguished from offglides of diphthongs, and in fact they can cooccur with an offglide. (Sometimes their realisations are somewhat fricatival, to magnify this distinction.)

Valid complex onsets are: anything but v j followed by v j; a nonresonant (including a nasal) followed by a resonant; two nasals; or a voiceless stop (or affricate) followed by anything other than a voiced or aspirated obstruent. Stop + fricative sequences are distinct from affricates. Complex onsets, overall, are less frequent than in Mavakhalan.

Aspirated stops do not occur in coda position, and in fact they're rather rare in roots altogether.

Segmental allophony and active phonological processes

(Note that the romanised orthography recognises no subphonemic variation.)

h /h/ has a number of realisations, according to the following vowel, or the preceding one when in a coda. In a syllable with a low vowel it's [χ]; with a front vowel it's [ç]; with a rounded vowel it's [ʍ]; elsewhere it's simply [h]. The postalveolars č čh š ž are [tɕ tɕʰ ɕ ʑ] before front vowels.

Voiced consonants finding themselves before voiceless consonants are devoiced. The same happens following voiceless consonants other than unaspirated stops (and affricates); this exception provides for a greater contrast between such pairs as kr [kɹ] and khr [kʰɹ̥]. z ž devoice finally.

ř in a coda is [ɾ], and represents the neutralization of all of d r ř in this position.

Nasals assimilate in place to following obstruents. Sibilants assimilate in place to following sibilants. Clusters containing one of s š z ž d r ř followed by one of r ř assimilate to řř. The cluster lr becomes ll.

h cannot stand after an obstruent. The two collapse into an aspirated obstruent, fricatives turning to affricates and another h to čh.

Vowels in hiatus are disallowed, certainly within a word, but also in fluent speech at word boundaries. Situations where two vowels come together are resolved as follows:

• if the first vowel is a diphthong and the second is high, the offglide of the diphthong drops, and the sequence is then subject to the adjustments in the next bullet. Otherwise, diphthongs behave exactly as if there were a consonant separating their two parts.
• if the second vowel is high, the two coalesce to a diphthong or i u < ii uu. If the second vowel is y it's replaced by i; then if the result isn't a legal diphthong or ii uu the first vowel is fronted or backed just enough to turn it into one.
• if the first vowel is high but the second isn't, j (after i) or v (after y u) is inserted.
• if both vowels are non-high, the first one drops.

No word ends in e. In borrowings final e tends to become i.

Register

AhH distinguishes four registers, i.e. tone-phonation complexes. The domain of association of register is the word: each word bears exactly one register. The tonal component of the register is realised with any contour centered on the root-initial syllable (when it's final) or its right edge (otherwise), except that in the first two tones the slight contour tends to just vanish when the accent is final. The phonational component is realised throughout the word. The registers are:

High

Tone 43; normal phonation or slightly stiff. No diacritic (a). This is the unmarked and commonest register.

Mid

Tone 32; normal phonation; stressed syllable may be lengthened. Denoted by a macron (ā).

Low-rising

Tone 13; heavily breathy. Denoted by an acute (á).

High-falling

Tone 51; somewhat breathy. Denoted by an grave (à).

The diacritic marking the register is placed on the stressed vowel.

If one takes the location of the tonal contour to be an exponent of stress, then stress always falls on the root-initial syllable (this is the sense in which this description uses the word "stress"). There was formerly a robust system of secondary stress, but it's no longer well motivated to posit one.

As for timing AhH falls somewhere between syllable- and mora-timed, where coda consonants are moraic but offglides of diphthongs are not.

At the phrasal level, declaratives with neutral intonation are characterised by a certain amount of downdrift, and an exaggerated fall on the main stress of the final prosodic unit (or, if this stress falls on a low-rising word, a flatlined low tone there). Questions are characterised be the absence of any drift, commands by a relatively low pitch through most of the utterance with a rise before the final prosodic unit. Topics are characterised by their pitch beginning higher, foci by beginning lower. [[What else? Information structure factors? Clause combinations?]]

Morphology ... plus a smidgen of usage

It remains sensible to analyse morphophonological alternations in AhH in terms of capital and caudal inflectional classes, determining the form respectively of prefixes and suffixes. Mavakhalan's capital classes are rather well preserved and have been augmented by the action of one vocalic mutation, although significant analogy has applied to the actual allomorphs they select. On the other hand, though the Mavakhalan caudal classes are still somewhat visible they have come to be overshadowed by new distinctions arising from syncope sensitive to earlier secondary stress patterns. Mavakhalan's class of subsyllabic stems with reduplicating augment has been lost without synchronic trace.

The capital classes, together with their abbreviations, are

• Cs2, voiceless cluster: stems beginning with a consonant cluster whose first member is voiceless, and stems beginning with an aspirated consonant, clustered or otherwise.
• Cs, voiceless: stems beginning with a single unaspirated voiceless consonant.
• Cv2, voiced cluster: stems beginning with a consonant cluster whose first member is voiced and not a nasal.
• Cv, voiced: stems beginning with a single voiced consonant other than a nasal.
• N2, nasal cluster: stems beginning with a nasal-initial cluster.
• N, nasal: stems beginning with a single nasal.
• j, yod vocalic: most stems beginning with j.
• PN, prothetic vowel–nasal: stems beginning with a vowel which is historically prothetic, followed by a nasal.
• P, prothetic vowel: stems beginning with a prothetic vowel followed by anything else.
• V, vocalic: stems beginning with a vowel, and certain stems beginning with v, which latter group represents the Mavakhalan waw vocalic class.
• i, yod consonantal: stems of various forms, continuing that part of the Mavakhalan yod class with no explicit initial j.
• *a, a-mutating, where * is one of the preceding classes: stems whose first vowel in an early stage of AhH was low.

Mavakhalan's waw consonantal class was very poorly represented and does not survive. Its members have been redistributed according to their form.

Prefixes containing a vowel must themselves be regarded as being either a-mutating or not. We mark a-mutating prefixes with ↓. All other prefixes which either contain a vowel or are an allomorph of such mandatorily select the non-a-mutated variants of preceding vowels. All prefixes which contain no vowel (in any allomorph) are transparent to a-mutatation.

Prefixes may select a new register for the prefixed word; this is specified by listing the diacritic for the register in the table, along with any segmental material. Additionally, some prefixes occurring before voiceless consonants are realised as aspiration, a consonantal mutation. Aspiration is tabulated as ʰ. Its effect is identical to infixing an h after the onset; thus it changes stops (and affricates) to their aspirated equivalents (leaving already aspirated ones alone), fricatives other than h to aspirated affricates, and h to čh.

For the purposes of presenting prefixal allomorphy the initial vowel in a P or PN stem, and the initial j in a j stem, will be presented as part of the prefix, since material in the prefix tends to replace them. For all other purposes, such as determining stress, they're still part of the stem.

There are three caudal superclasses, the strong, weak, and mixed groups. Within each superclass a number of classes can be distinguished based on stem-final vowels. In all the classes which must be distinguished are

• the i strong (iS), y strong (yS), and consonantal strong (S) classes;
• the i weak (iW), y weak (yW), schwa weak (sW), and consonantal weak (W) classes; and
• the (consonantal) mixed (M) class.

Suffixes may induce a-mutation in the stems they attach to; this is the same process as that caused by a-mutating stems. We tabulate this also as ↓. Only certain stems are susceptible to a-mutation. In those that are the observed alternations are as follows, with those with colored background most frequent.

Suffixes may also induce a new register in the suffixed word. In instances where a word is assigned a register by multiple affixes, the general rule is that the register coming from the furthest left morpheme prevails, but there is much uncertainty among speakers regarding these forms. Certain innovative varieties simply ignore the register component of most suffixes.

Nouns

The nominal word consists of several components, as follows:

[case] – number, definiteness – stem – [possessor]

Number and definiteness

The material in the nominal paradigm coming immediately before the stem expresses definiteness and number, and distinguishes oblique and nominative stems in the definite. In the indefinite the two stems can be taken to be identical. All non-nominative cases are formed from the oblique stem; the nominative case is zero-derived from the nominative stem.

The bracketed (ou) and (ei) appear when necessary to prevent illegal onset clusters starting with š; they drop out in the presence of a case prefix which allows the š to serve as a coda. The (y) only drops before single resonants, case prefix or no.

The inflectional collective of Mavakhalan has been lost, though a derivational one persists. The singulative remains. The singulative is productively formed from the bare stem of the noun (which agrees with its indefinite singular form) by changing the register to mid; a certain number of exceptional nouns suffix i or y to the stem as well.

Case

Case markers occur as prefixes to the head of a noun phrase. The one most variable in form is the accusative:

Its form is determined by the shape of any number and definiteness prefix present, not necessarily by the capital class of the root. (The parenthesised vowels are dropped where permissible.) Observe that its most common allomorph is zero, which is selected in particular by any definite oblique prefix. The nominative is zero everywhere, but in the definite it takes a different stem to every other case, maintaining the distinction from the accusative.

The only morphophonemic variation displayed by the remaining case markers is that certain among their number are subject to a-mutation (caused, recall, by *a stems to which no vowel has been prefixed or any stem with an a-mutating prefix). These case markers are as follows.

In romanization, case markers aside from the nominative and accusative are followed by a hyphen.

Nonaccusative case markers never displace a prothetic vowel, and so are never a-mutated by words with one. Compare honda 'a loaf of bread (acc)', with mutation, and zy-inda 'of a loaf of bread', without, from the PNa-capital stem inda.

Possessor marking

There is no distinct 1st plural category of possessor marking. After its phonetic form fell together with the 1st singular, the 3rd plural took over its role; the 3rd plural is still used for 1st plural referents. Where this would lead to an uncomfortable ambiguity a genitive personal pronoun can be supplied, h-əizei or zyv-əizei as appropriate. Such a strategy is also often necessary in the 1st singular with stems in i.

Adjectives

Adjectives in AhH are a subclass of nouns, distinguished from general nouns by their capability to modify nouns attributively. Used independently adjectives have semantics like 'Xness': some typical ones are dyzi 'whiteness', nunzei 'speed', hīju 'cruelty', ṗhy 'obesity'. These nouns only denote the state of possessing some property, and are never used to refer to a position along a continuum (unlike in English where e.g. "height" is ambiguous between 'the state of being high' and 'the extent to which (a thing) is high'). For the latter meaning AhH juxtaposes two adjectival nouns specifying both ends of the scale, positive first: e.g. hu ṗei '(degree of) height', lit. 'high(ness)-low(ness)'.

When adjectives are used attributively they receive no marking of their own for case or number or definiteness or possessor: [[example]].

There is no morphology confined to adjectives. The intensive jəm- characteristically attaches to adjectives, as in jəndyzi 'extreme whiteness', but even so combines with other nouns as well: kory 'friend', jəňkory 'best friend'. The same can be said of modifiers like jal 'very': these prefer adjectives, but even so one finds constructions like mumij jal 'utter fool', lit. 'very fool'.

Pronouns and determiners

Personal pronouns distinguish nominative and oblique stems, from which case forms are produced in the same fashion as nouns. The plural personal pronouns have singulative forms: in the third person these are ā ōu 'one of them' as if formally from the singular, and in the first and second persons they are ə̄izi ə̄izei 'one of us', lōh lōhu 'one of you' from the plural. The personal pronouns also have distinct distributive forms.

The ? forms in the table are the interrogative pronoun 'who'.

Demonstratives

AhH has a five-way system of deictic contrasts, manifested in a series of pronouns and one of determiners. The pronouns show no distinction between nominative and oblique stems but receive other case markings normally. The following table again includes interrogatives.

Quantifiers

Numbers

We tabulate here only those numbers that are determiners, and derivations from them. For usage see later.

Ordinals are regularly formed by prefixing əl- (of capital class P). In multipart numbers the first word gets the prefix. 2 4 6 and their multiples by powers of ten, and numbers starting with these, instead take just l-, deleting any initial v and in the cases of 2 · 10ⁿ changing the register to low-rising. 'First' is suppletive.

There are also adverbs in the sense 'in groups of n', existing only when n is one of 1 2 3 4 5 10: keskəi jezja zozzy vòuzvu dozdy rařřo 'one by one, two by two, ..., ten by ten'.

The nullar quantifier ouk can be glossed 'zero' but doesn't really form a part of the number system, having for instance no ordinal.

Verbs

There are two different word classes in AhH which have reasonable claims to the label "verb". One of them is an open class with striking morphological similarities (but not identity) to the noun, whose members in fact historically derive from deverbal nouns. The other is a closed class continuing the verbal class of Mavakhalan, but whose roster has much diminished and now includes only auxiliary verbs, light verbs, and a handful of other frequent verbs, mainly unaccusatives or those which commonly occurred first in serial chains, which are represented in both classes.

In this sketch we apply the name p-verb to the former class and the name s-verb to the latter; these are supposed to suggest "periphrastic" and "synthetic" respectively. The lexicon abbreviates these classes respectively as v and s, using v/s for stems that take forms of either class.

P-verbs

The p-verbal word, mirroring the noun, consists of the following components:

[adverbialiser] – phase, polarity, number – [compounded s-verbs] – stem – [progressive aspect] – person, number

Inherited morphologically simple p-verbs are only found in the consonantal classes and the schwa weak class, though through derivation and borrowing p-verbs have come to exist in all classes.

The compounded s-verbs in this structure are inner auxiliaries and the first components of separable p-verbs, but both may be regarded as in fact compounded to the stem (from which perspective they really shouldn't appear in this skeleton at all). On the other hand the position of the stress, on the stem and not on any compounded s-verbs, speaks to the analysis here.

Phase, polarity, and number

The category of phase (terminology borrowed from this description of Tepa) has two values, called indefinite and definite after the values of the nominal definiteness distinction they arise from. Phase doesn't map well to a single standard verbal category, and its usage is rather subtle. It overlaps somewhat with aspect: definite phase correlates with perfectives and perfects, and is categorical on experiential perfects. It overlaps with mode: definite p-verbs are mainly realis. It even overlaps with evidentiality: definite phase is most common with directly witnessed happenings. Miratives always take indefinite phase. Perhaps the most reliable guide is still definiteness: most (but not all) definite p-verbs refer to events or states which are identifiable to the listener and speaker, and most (but not all) indefinite p-verbs to unidentifiable events or states.

Only positive p-verbs display phase marking; negative p-verbs have an invariable negative marker in the relevant slot. The definite phase marker reflects the number of the most patientive participant (S in intransitives, O in transitives). The usage of the pprn forms will be elucidated later.

As before, (parenthesised) material appears only where phonotactically necessary. The {m} drops out only when the prefix stands before a stem of capital class V in v.

These markings are entirely isomorphic to the possessive markings on nouns; we reprise the earlier table. Recall that the third person plural subsumes the first person plural.

The "unmarked" category here is only used in special circumstances, such as in imperatives and some clauses with outer auxiliary verbs. The main p-verb of a canonical clause must have a person marker.

Progressive aspect

A number of progressive forms of the p-verb are built on what we might call the progressive stem, which is formed from the unmarked stem by suffixing -in, absorbing any stem-final vowels. The resulting progressive stem falls into a consonantal caudal class, but the superclass (strong, weak, mixed) is preserved from the unsuffixed form. So ezzə 'knead', sW caudal, yields ezzin, W caudal, and ləizi 'utter', iS caudal, forms ləizin, S caudal.

A very few verbs would have inherited an exceptional progressive stem, with syncope: one such is fazys 'undress', progressive stem fassin. These verbs have regularised by generalising the syncope to occur with all suffixes.

Progressives inflecting to agree with non-third persons apply to the progressive stem the suffixes introduced in the last section. Thus including the progressive -in we get

In the third person, the progressive suffix reacts with the person suffixes and we get these fused endings instead, which replace the -in but still absorb any final vowel.

In particular the progressive does make a distinction between first and third plural: the exponents of 1pl agreement, -inak/-inko, include the n from the possessive stem formant where the exponents of 3pl agreement, -ijak/-iko, do not.

Adverbialisers

Various constructions require p-verbs with an adverbialiser. The use for which they're named is that a verb phrase with adverbialiser can occur alone, making up an adverbial clause. In this use they always take progressive forms, and therefore make no TAM distinctions.

The adverbialisers are formally identical to a subset of the oblique case markers on nouns; two of the nominal case markers (et-, raf-) are not found as adverbialisers. We present here the functions of the adverbialisers when they appear heading adverbial clauses. Il- has no such function and is only ever licensed by outer auxiliary verbs.

Of these at- and al- have the broadest usage. In many functions they're not much more than polarity markers (thus the glosses pos and neg).

[[Constructions where a gerund would have been called for in Mavakhalan survive too, and in taking no s-verb have perhaps the flavour of small clauses. P-verbs may still have gerundive sorts of uses too, but I'm tempted to say that most constructions require a true noun there (at could be one), along the lines of "the fact that X".]]

S-verbs

The s-verbal word consists of the following components:

mood – stem – tense, number

Since the s-verbs are few and their functions diverse, we'll often want to refer to individual ones. As their names we take the forms that they would have had unaffixed; this nearly always coincides with their habitual singular form (with no mood prefix).

Mood

For an Edastean language of its age and milieu, the mood-marking system of AhH is rather degraded. Only the indicative, optative, and conditional survive; the last has taken on functions of the obligative, and has certain uses where it's simply a general irrealis. [[Note this thing that protases might be rendered by some proposition.]] Furthermore, in casual style use of the optative is flagging. Sentences with optative force can be and increasingly are cast in the conditional with the word kótəi or its negative uňkōtəi; these are in origin optative forms of ot 'come'.

The indicative affirmative on V-capital words, listed as r*, may change the initial vowel of its base. We'll give the modified vowel of the indicative affirmative of every V-capital s-verb for which this happens. (Many of the columns in the above table must be regarded as purely theoretical.) [[Once the list is complete, remove these.]]

In informal styles, an initial unstressed syllable of form rV(G)- contained (fully) in any of these prefixes is deleted (where phonotactically possible). Thus the indicative affirmative of nun 'go', formally based on ronun (stress on the u), in everyday speech is simply based on nun, as is the optative affirmative, formally rounun. Indeed, this collapse of the affirmative optative and indicative is frequent, but it's no big deal, given that the optative doesn't get much informal use anyway. As a nonexample the rV(G)- syllables can't be trimmed from the indicative affirmative forms based on ryt of ot 'come'.

On the other side of the coin, by analogical extension, every s-verb can take an rV(G)- syllable as a mark of (hyper)formality. In the unetymological cases not exhibited with such a prefix in the table above, the prefix takes the form ro-. (No forms with two rV(G)- syllables arising from prefixes ever occur.)

Tense and number

The s-verb inflects in fusional fashion for three tenses, present, past, and habitual, and two numbers of the subject, singular and plural. The caudal stem classes recognised elsewhere in AhH aren't very useful in presenting these suffixes. Accordingly, since the s-verbs are so few, we won't exhibit the relevant suffixes here but will simply list all six forms of each s-verb when we enumerate them shortly below.

The complete roster of s-verbs

Among s-verbs, the copula is unique in reflecting the animacy of its subject. Its inanimate forms are based on the stem at, while its animate forms are based on the two stems ap and má. In some animate forms both the latter two stems survive, the choice between them being determined by the construction. For this reason and a few others of syntactic nature, it might be argued that AhH has two copular verbs, most of whose forms happen to coincide. We won't take it quite that far.

All remaining s-verbs have only one stem with its six suffixal forms (and only five distinct: the present singular and habitual plural always coincide). The glosses we give here are only representative.

[[Finish the list.]]

A certain amount of the derivational morphology that has been innovated since Mavakhalan times is suffixal, contravening the general head-first pattern of compounds. (Adāta had a few instances of this as well, and it's just possible that the same underlying motivating factor has survived.)

Action nominalizations are obtained by zero-derivation. Recall also that adjectives are fundamentally nouns with abstract sense, so that a derivational operation for quality nominalization is unnecessary.

In the list below, the prefixes bearing the annotation {acc} attach to the accusative indefinite singular of their stem; other prefixes attach directly to the stem. When a suffix is presented in two forms with a slash, which one occurs is determined by the caudal superclass of the base: the former form occurs on strong stems, the latter on weak and mixed. The annotation {sw} means :S stems ending in non-high vowels should be treated weak, while mixed and most consonantal weak stems (but not those with an .|n-extension) should be treated strong, and that any .|n-extension is deleted before a consonant. {f↓} means that base-final vowels should be a-mutated by all allomorphs. Other allomorphy is explained in the entry. The notations in the second column specify the new capital (for prefixes) or caudal (for suffixes) class, and have the same meaning they do in the lexicon.

The derivations in the next group create separable p-verbs. Each is cited in two forms; the first is a prefix to a p-verb, the second is a standalone s-verb. Which of the two is selected — whether one finds a prefixed p-verb or an unprefixed one in a clause with s-verb separate — is contingent on the particular construction the separable p-verb appears in. In case the prefixes cause modifications to the root, these modifications don't occur in the separated form: sau-čhēt 'receive hospitably', from the base het, occurs as het with an s-verb (based on) sāu when separated.

The semantics of verbs using especially the last three of these prefixes tend to be very idiomatic. The first two have a number of predictable uses; in particular most adjectives and most stative intransitives form an inceptive and a factitive, and most transitives in tez- with typical case selection can form a sort of passive by replacing this with t(a)-.

Compounding

Compounding remains productive, as attested by the significant number of compounds the AhH lexicon has gathered since the Mavakhalan stage. However, speakers never create new compounds nonce (except perhaps in poetry, for their striking effect); compound-formation should be thought of as stripping the internal morphology from a fixed collocation, which generally already has high frequency when the compound is formed.

Compounds are still head-first. The development of p-verbs out of Mavakhalan nominals has resulted in practice in a substantial loosening of the restrictions on compound types. [[I should go through pulling examples from the lexicon, conversely making examples of the other types. I will strike the dvandva and the two noun+deverbal and the verb+verb.]]

Syntax, and more usage

Case and adpositions

Case usage

AhH has accusative alignment: normally, subjects of intransitive or transitive clauses take the nominative, objects of transitive clauses take the accusative. Several verbs select cases for an arguments which the gloss suggests is an object other than the accusative, but such verbs are syntactically intransitive. [[Has every verb a subject? Is there real quirky case?]]

[[Discuss topicalisation with ni, and other methods of topicalisation.]] [[predicative == appositive, also for some depictives]] [[ ́ju- for focussing in the construction developing from a cleft]]

Positional nouns

In place of a refined series of oblique cases, or a dedicated class of adpositions, AhH employs positional nouns – a slight misnomer, since they don't all describe position. A couple examples are mot 'middle, inside', tessot 'exclusion, exception'. Positional nouns, and positional senses of nouns, are picked out with (pos.) in the lexicon.

The usual contruction involving the positional noun has it appear in an oblique case followed by its complement in the accusative. This complement can equally be a noun or a p-verbal subordinate clause [[link?]], as semantics permit. For positional nouns which do describe position, the case can be freely chosen from the local cases, and the sense is compositional: so there's u-mót 'in the middle of', et-mot '(in)to the middle of', ha-mot 'from the middle of', raf-mot 'through the middle of'. Other positional nouns will have a fixed case (or several such, if there are several different uses): 'except for' is nə-tessot.

Things are conceived of as moving forward through time. Thus the positional noun rolkə appearing in 'in front of' is also used to mean 'after', vah in 'in back of' to mean 'before', and mot to mean 'at the same time as'. (Conveniently, mot is homophonous with the noun 'time'.)

The noun phrase

The composition of the noun phrase has little changed since Mavakhalan, aside from the conversion of old cliticising prepositions to case. Noun phrases are essentially head-first. (< denotes 'precedes' here.)

quantifiers < demonstratives < head noun < adjectives < nouns in oblique cases < relative clauses

Here's one with everything:

Nouns with any oblique case can modify nouns, not just the genitive as above: azza u-šidú 'a city in the mountains'.

Since adjectives are nouns, they are equally followed by their modifiers: pīmou jal 'very brave' lit. 'brave very'.

[[Relative clauses below.]]

Genitives no longer demand 3rd person possessive marking on their head. It's still permissible to provide this possessive marking, though it's rather emphatic and usually reserved for cases where the possessum is topical. So 'the father of the children' is usually məis zy-šidou with məis 'father' in its bare form, but can be mesko zy-šidou with a 3pl possessive -ko.

Number and definiteness

The indefinite singular is used in place of the plural when a determiner is present supplying the plural sense. This does not happen in the definite. So afik 'cats', zy ik 'three cats', vs. khik 'the cats' (nom), zy khik 'the three cats'.

Proper names only occur in the zero-marked form that we're here calling the indefinite singular; of course, they have definite sense. Possessed nouns, however, do take definite marking as appropriate.

It is generally more idiomatic to use singulars with collective sense, where these exist, than plurals; in particular reverting from a singulative to its base is frequenter than repluralising it: cíncik 'the stars (nom)' is likelier than chīncik, though cf. ō chīncik 'each of the stars'.

Non-count nouns also generically take formally singular inflection.

Quantifiers

A noun phrase may contain up to one quantifier specifying quantity, be it a numeral or one with relative sense (oř 'all', hjeř 'many', ...), as well as collective əis or distributive ō: thus ō zy šeiməij 'each of the three cows'. It may not contain two of the same sort: *oř zy šeiməij 'all three cows'.

Any quantifier may occur as the head of a noun phrase. It then takes nominal case-marking.

This provides the workaround for the restrictions on quantifier stacking: oř [zy ha-šijakrou] 'all three enemies', more closely 'all (of them) from among the three enemies'. Note that without devices like these, the distinction between 'the three enemies' and 'three of the enemies' is not made.

The single-word numbers less than 1000 are determiners, and so precede the noun, which occurs in the expected case but is singular: vù ydy 'four eyes' (nom). Numbers inexpressible in a single word are expressed additively, the components running from greatest to least, with nothing in between: jāju dyř ə̀itu mev '256 people'. By contrast, the number taiheč Cs:S '1000' is a noun which modifies its head adjectivally, and thus appears following its noun, which may or may not be plural. With counts that are an exact multiple of 1000, the determiners modify taiheč and that's the end of the story: (af)ty ro taiheč '10000 lakes'.

Formal style demands that large nonmultiples of 1000 be split, thousands appearing afterwards with an u 'and', the remainder as a determiner, along the model of jař vù mev u taiheč '1024 people', lit. '24 people and a thousand'. It's a common informal trend to put the thousands out front in this case (without u) to achieve contiguity and monotonicity of the number words: taiheč jař vù mev id. But note that when such numbers are used with nominal force, the contiguous monotonic pattern of taiheč jař vù-kə '1024' is entirely normal, while jař vù-kə u taiheč sounds horribly affected.

These methods go so far as 10⁶-1. Beyond that there are no standard words. There does exist a sequence of names for higher numbers (syntactically like taiheč), namely motnja N:S 'zillion', motnja ljà lit. 'second zillion', motnj əlzy lit. 'third zillion', etc. The catch is that different traditions assign them different values: for some the nth zillion is 10⁶ⁿ, for others 10⁴ⁿ. If one wants an unambiguous 'million' one's best recourse is aftaiheč taiheč 'thousand thousands'.

Numbers derive nominalisations with the sense 'the number N' with -kə. [[example, perhaps the old 'he threw a ten on the dice']].

Of the two words for 10, ro tends to be used in counting or enumerating things in sequence, with nominalising -kə, and with a few nouns (such as units of measurement or time, or any higher numbers which it modifies multiplicatively). Kəiř is used with most nouns.

Fundamental verbal constructions

The majority of verbal constructions in AhH are so fundamentally periphrastic, requiring both an s-verb and a p-verb, that we will have to introduce the constructions on which they're based before we can treat the construction of canonical clauses. In the simplest cases the necessary s-verb is the copula, so we treat the uses of the copula now.

The copula serves as an existential verb, including for locational senses:

With a complement in the predicative case, this serves for predicate nominals, both equational and instance-of-class.

Predication of possession uses one of two constructions, depending on whether the possessor is expressed as a pronoun. If it is, AhH uses a possessed noun and the copula in the sense of 'exist', and the possessor may be omitted, manifesting only as the marking on the possessed noun (this is uncommon in the first person plural):

Otherwise, the formula for 'X has Y' is 'X is with Y', Y appearing in the comitative. 'X does not have Y' has a corresponding formula with Y in the caritive; the copula remains positive.

The same pair of constructions is used for adjectival predicates, which makes sense, bearing in mind the semantics of adjectives as nouns.

With animate subjects, the equational construction uses the stem ap of the copula (type 0 in the table), and the possessive construction uses the stem má (type 1).

Canonical clauses

The unmarked word order of AhH is sV S pV O (obliques, adverbs). sV and pV denote the s- and p-verbs; one of each usually occurs. This is the expected development of the Mavakhalan order once the former pattern of topic fronting had lost its predominance (perhaps under Ayāsth influence); current AhH prefers a newer method of topic-marking. [[Link that.]] Adverbial elements, however, can still occur fairly freely in clause-initial position, and S will occasionally.

Canonical clauses with p-verbs are formed on either the existential or the possessive construction. An example of the existential is

Observe that the p-verb here, apsijei rendered 'is making', takes no adverbialiser (it historically took one in very early AhH, but this has been lost without trace).

The possessive construction continues to display its variation in formation according to whether the subject is pronominal. In the non-pronominal case the construction looks like

The p-verb (here at-valei 'is resting') obligatorily bears an adverbialiser at- (in the positive) or al- (in the negative) cognate to the case marker used in true possessive clauses. When the subject is pronominal, or is omitted entirely, the construction instead looks like

I've recast this example in definite phase in order to exhibit a pprn phase marker (on válei), this and the absence of the adverbialiser (and of any explicit subject) being the only points where the two variants differ. The pprn phase markers appear on s-verbs of definite phase in the possessive construction whose subjects are pronouns (indeed, they're cognate to the nominative definite forms in the nominal definiteness paradigm, and accordingly occur exactly where true possessive clauses have a nominative).

In both constructions, the number agreement on the s-verb is determined by the subject of the p-verb, not by the p-verb itself (in the few cases where syntax makes that seem plausible). So mó valiko 'they are resting' takes the plural copula mó and not the singular mái as the parallel with the nominal possessive construction would suggest.

Of course, clauses built around one of the few s-verbs that retain lexical uses (most of which have double lives as p-verbs) will contain no p-verb themselves. Such clauses continue the nonperiphrastic pattern of Mavakhalan; the choice between existential and possessive constructions doesn't arise. We will call these nonperiphrastic clauses. Vel 'rest' happens to be one of these s-verbs, so we can also have

The only clauses which need not contain an explicit subject are those in the possessive construction. In the existential construction and in nonperiphrastic clauses, an explicit subject is necessary, and agreement on the p-verb does not suffice, whether third person or non-third. Thus *rōp apsijei řəisi for 'he is making a chair' and *vēlai u-řə́isi for 'he is resting in a chair' are ungrammatical, and should be rōp a apsijei řəisi, respectively vēlai a u-řə́isi.

Clausal tense and aspect

We have seen that that the p-verb has a two-way progressive vs. unmarked distinction and that the s-verb has a three-way distinction of tense. Taken together with the choice between the existential construction and the possessive construction, this makes 2 · 3 · 2 = 12 ways of marking any given canonical p-verbal clause. AhH makes good use of these, associating each with one of its twelve basic tense-aspect categories:

The general scheme is that the existential construction tends to be dynamic and the possessive construction static. In the existential construction the absence of the progressive gives punctual senses; in the possessive the absence of the progressive tends to displace the event back in time.

Clauses with no p-verb simply collapse many of these distinctions; only the tense distinction on the s-verb is available.

Non-simple clauses

The constructions we present in this section have no particular common thread, aside from being more complicated than the canonical clauses discussed earlier.

Separable p-verbs

In the existential construction, separable p-verbs appear separated, the s-verb part replacing the copula. So a separable verb such as tez-aľ 'cook' might appear as

In the possessive construction, the copula is not replaced and the separable p-verb appears compounded. Using sau-phjá 'prefer' we have

Auxiliary verbs

All auxiliaries are s-verbs. There are two broadly different syntactic patterns they follow, inner and outer, treated in the next two sections.

Inner auxiliaries behave syntactically just like the first components of separable p-verbs. Thus, in the existential construction the auxiliary stands in place of the copula. In the possessive construction the auxiliary appears compounded with the p-verb. Inner auxiliaries cannot occur in nonperiphrastic clauses. Using causative tes as exemplar we have:

in the habitual dynamic, taking the existential construction, and

in the aorist, taking the possessive.

These auxiliaries can stack, and can occur with a separable p-verb; in either case the clause will have a string of s-verbs. When there's such a string, in the existential construction the first s-verb takes the place of the copula. All other s-verbs are compounded to the p-verb. So in

there are three s-verbs, causative tes serving as the clause's main s-verb, and at 'stop' and sau the separable prefix of 'blame' both compounded onto the p-verb. This said, doubly-compounded p-verbs are awkward. Triply- and more-compounded p-verbs are avoided and are best reckoned as outside the verges of grammaticality.

Known inner auxiliary verbs, listed in their compound and independent lexical forms, are [[more?]]

There are a few constructions that can be regarded as intermediate between clauses with inner auxiliaries and those with separable p-verbs. [[example: productive auxes for nouns; also plain old equivocation between n and v]]

[[What stands in for the passive? This is a question of information structure.]]

Outer auxiliary verbs are s-verbs that always take clause-initial place in canonical clauses. They may specify an adverbialiser which must appear on the following verb, which must be a p-verb. If the clause taking this auxiliary is in the existential construction and would contain the copula as s-verb (i.e. has neither any other auxiliary or a separable p-verb), the outer auxiliary replaces the copula and the main p-verb gets the adverbialiser. Otherwise, the outer auxiliary converts the following s-verb to a p-verb and gives this its adverbialiser. Any later p-verbs lose their person marking. Taking as example the volitional future nōu we get

Since the copula has no p-verbal form, it requires special treatment in the possessive construction (where it's not deleted) after an outer auxiliary. In lieu of its p-verbal form one instead gets the p-verb run (which I guess can be considered suppletion), with the remainder of the clause reformed as if following the existential construction (without changing whether it's marked for progressive).

Nonperiphrastic clauses may take an outer auxiliary verb, unless their s-verb is the existential copula.

The outer auxiliary verbs include the tense auxiliaries, as well as many with modal semantics. Known examples, with the adverbialiser they take, are [[more?]]

[[what tense-aspect categories are selected inside tense outer auxes?]]

There are a small number of verbs which participate in an object-introducing construction much like the outer auxiliary construction, but with an extra object inserted before the p-verb, making the clause structure Aux (sV) S O₁ pV (O₂) (...). These constructions are mostly of pragmatic utility: among other things they indicate a greater degree of focus on the first object O₁.

Known verbs of this sort are

[[Serialisation did not develop into auxiliaries. It's now an improductive strategy, and surviving instances of it want to work as sequences of two ordinary verbs, maybe with some of the shared marking reduced. Instances: 'take to', 'bring'. The only trick might be that one of them survives as an s-verb, and so falls in with that pattern...]]

More constructions

Reflexives and reciprocals are expressed by means of the noun vařkə '(him/her/it)self', which needs the usual definiteness marking but isn't marked as possessed. None of the other constituents vary in form between reflexive and non-reflexive clauses. In the plural vařkə is indifferently 'themselves' and 'each other'. If the distinction is necessary it's made with the collective and distributive quantifiers.

Negation

There are at least three different exponents of clausal negation in AhH. Only one is ever used at once. Clauses without a p-verb, and only these, use the s-verbal negative system.

P-verbal clauses built on the existential construction negate only the p-verb.

And p-verbal clauses built on the possessive construction simply use the adverbialiser al- in place of at-.

In line with this, AhH exhibits no negative concord. Clauses containing a non-verbal element, such as a quantifier, with negative force take no further negation. (When presented with them, speakers tend to find clauses containing multiple negative elements confusing.)

Questions

[[Do this.]]

Imperatives

Imperative clauses contain no s-verb, and so minimally consist of just a p-verb alone. The imperative verb generally entirely lacks person agreement, though for emphasis second person markers may appear. A progressive imperative, from the bare progressive stem, has the expected sense 'do X continuously over a period of time'.

Cohortatives are formally imperatives with first singular marking. (This was historically the first plural; the construction is a rare example of the third plural not taking over.)

Negative imperatives use the negative marking on the p-verb.

[[Does serialization survive elsewhere? Maybe there are still traces of it in coordinationy constructions, or in pivots, or something.]]

Adverbial constructions

Adverbial phrases within a clause are ordered place manner time.

Manner is generically expressed with a prepositional phrase containing an adjective, used adverbially. [[Example: 'with e.g. haste' and 'with e.g. a knife' are entirely analogous. Also 'in my haste'.]]

Many underived sentential adverbs are frequently sentence-initial. [[For adv/n words expressing time, the unmarked options are initially as an adverb or in M adjunct position with the circumstantial. When they occur initially with the circumstantial they're fronted as topics.]] [[For that matter, general topicalization with the circumstantial? Above.]]

P-verbs may take (oblique stems of) quantifiers, including numbers, to express number of repetitions of an event. As in their other uses these quantifiers precede the p-verb.

[[example of an adverbialised clause headed by a p-verb with adverbialiser. link morphology to here]]

Clause combinations

[[is there really all that much to say here?]]

[[Relatives. (Do honest participles exist?) p-verbs with at- or al- work when the subject is relativised on, and get the pprn markers. Maybe the object behaves the same way, with just a preceding subject and gapping on the object. Otherwise we could get jōu, but I wonder if the pattern has been reformed.]]

[[Subordination.]]

[[If-then constructions. There is a lexical 'then' ňei.]]

[[The idiom for comparisons: 'first / before Std in Xness', for appropriate 'before'.]]

Conjunctions

To render 'and', ȳi is used between clauses or verb phrases, u elsewhere. In a few fixed collocations with dvandva-like semantics, u can appear as un before a vowel. [[example?]]

Many conjunctions can be used with only a single argument. [[How does this interact with the case of more than two conjuncts?]]