Basic clause structure

The most typical order for a Tsi sentence is SOV or SVO. The postverbal position is commonly used to introduce new information (i.e. indefinite constituents), particularly constituents with a significant amount of discourse relevance (specificity):

(1) Ya-dgo kto bà mte quuŋ-odʰę
round-mix start INAN INDF mud-PL
So he started mixing some mud

The immediate preverbal position is used for generics and old/background information. In these cases directional particles prefix to the object rather than appearing on the verb and the object must be unmarked and unmodified. (2) is a continuation of sentence (1):

(2) ya-quuŋ dgo-x cçę içç aŋŋo...
round-mud mix-REL ANIM NOM in...
And while he was mixing [that] mud….

Adverbial phrases can appear relatively freely before or after the verb:

(3) mdòt k!a-k!a tsą k!ǫǫ aŋŋo
wood chop ANIM quarter in
he is chopping wood in the quarter

(4) k!ǫǫ aŋŋo mdòt k!a-k!a tsą
quarter in wood chop ANIM
he is chopping wood in the quarter

Any nominal/adverbial constituent can be given contrastive emphasis by fronting, a change in intonation, and the insertion of the particle iç, which appears to be cognate to the (geminated) nominaliser içç. Presumably this is etymologically some kind of clefting construction:

(5) k!ǫǫ aŋŋo iç mdòt k!a-k!a tsą
quarter in EMPH wood chop ANIM
it's in the quarter that he's chopping wood!

Animacy

Tsi has a robust animacy distinction. Historically animacy existed alongside a more elaborate noun class system, which is retained in more archaising forms of the language by agreement of demonstratives, different plural suffixes and so on. In the Tsi of Tsat, however, this system has levelled out into a two-way distinction between animate and inanimate nouns. Animacy is distinguished morphologically in the following places:

Plural suffix: Inanimate -otʰ(a) vs animate -rą
Proximal: Inanimate tʰoobʰ/-tʰo vs animate roomʰ/ro
Distal: Inanimate tʰobʰų/-tʰų vs animate romʰų/rų
One/indefinite article: Inanimate mte~mde vs animate mre
Two: Inanimate bdelʰ vs animate brelʰ

The plural suffix -rą is not found in liturgical Tsi, where the two animate forms are class 6 -rǫ and class 5 -ąą. Possibly it is a dialectal form, a combined form, or a modified version of the former to avoid ambiguity with the demonstratives.

Note that whilst forms of the (historical) classifiers (o)q!ʰạ (used for large features of the landscape),(o)hạ (used for tools) and (o)hąą (used for animals) do appear frequently in spoken Tsi, they have been in a sense 'degrammaticalised' and are used as part of the honorific system rather than as classifiers or anything else. They can and do co-occur with true classifiers, which was impossible in older forms of Tsi.

Auxiliaries

More important than the limited system of agreement, however, is the complex series of auxiliaries that Tsi uses alongside main verbs to distinguish between verbs with animate and inanimate objects and subjects. The system is effectively split-ergative, with imperfective verbs accompanied by auxiliaries determined by the animacy of the subject and perfective verbs accompanied by auxiliaries determined by the animacy of the object.

Tsą/cçę (reduced versions of ‘to walk’) are used for verbs with animate subjects and imperfective meaning.

(6) ya-quuŋ dgo cçę
Round-mud mix ANIM
He is mixing mud

g|í (a reduced version of ‘to be’) is used for verbs with inanimate subjects and imperfective meaning:

(7) K!ek ša-ša g|í
Spring flow-ONOM be
The spring flows, is flowing

Bà(‘to take’) is used for verbs with inanimate objects and perfective meaning.

(8) Ya-dgo kto bà mte quuŋ-odʰę… .
Round-mix start INAN INDF mud-PL…
He began mixing mud

Generally, other verbs take no compulsory auxiliary. With some verbs, generally expressing violent action, dzo ‘give’ is used when the object is animate:

(9) Ka-ro-'n maŋ dzo ạnnạ
man=that=1sg hit give 1sg
The man hit me!

There are a number of other optional auxiliaries which can replace these forms, giving verbs additional nuances in meaning, but they are beyond the scope of this post to discuss in detail. Auxiliary verbs differ from normal serial verbs - although they are historically one and the same - in that they can appear governing a string of coordinated main verbs:

(10) mdòt k!a-k!a, ya-quuŋ dgo cçę
wood chop-ONOM round-mud mix ANIM
He is chopping wood and mixing mud

Location and direction in Tsi

Prepositions

Tsi has three 'true postpositions' whose meanings are quite broad and non-specific:

Hųų - directional ‘to, up to, into, towards’.

(1) Tsààd hųų - towards, into Tsat

Aŋŋo - locative ‘at, in, inside, by’

(2) Tsààd aŋŋo - in Tsat

Ro - affective (benefactive/malefactive) or genitive.

(3) Tsààd ro mà - the gate of Tsat

(4) Qʰotsààd ro - for the people of Tsat, the Tsatians

Locational verbs

It also possesses a set of 'locative verbs', which are similar in behaviour to the equivalent category in other regional languages. These have two forms, one historically derived from the old imperfective and one from the perfective. In liturgical/inscriptional Tsi, perfective forms (with -d) are used postpositionally:

(5) rlàŋŋạ ɢǃòn-tʰo lʰod to-srob là bà
squadron Qon=PROX be_around up-tent raise ANIM
the groups of soldiers camped around Qon

(6) bàq-bàq mad so-qǂʰo tšo
terror_bird be_before down-heel set
He braced himself (lit. 'put down his heel) in front of the terror bird

The -d forms can also be used as independent locative verbs with the meaning of 'to be' + a preposition in English:

(7) šušų yeb cçę ạnnạ ro qǂʰò
shushun be_in ANIM 1sg GEN house
Shushun is at my house

Introducing two new constituents these verbs typically appear initially, a rare example of verb-initial syntax:

(8) mad g|í Tsààd ro mà ŋ|oŋ|o àà!
be_before INAN Tsat GEN gate, clamour MOD
There's a great noise out in front of the gates of Tsat!

As has become the default construction in other Tsiic languages, Tsi can also essentially annex the forms without -d to a main verb in a serial construction. This is a valency-modifying operation, transforming the adverbial into an object of the verb. This often adds a different nuance to the sentence. Compare (9) with (5) for example - (9) serialises the verbal form for 'around'. Whilst the basic meaning is the same, (9) gives the implication of a deliberate surrounding, perhaps for a siege:

(9) rlàŋŋạ to-srob là lʰo bà ɢǃòn-tʰo
squadron  up-tent raise be_around ANIM Qon=PROX
the groups of soldiers camped around Qon

'Qon', as object, can now be incorporated into the verbal complex provided it meets the requirements (i.e. relative backgrounding, previous mentions etc). (10) shows prefixing of the directional affix to Qon, which is now unmodifiable and thus must lose its demonstrative:

(10) ɢǃòn to-srob là gi ŋìngga-ŋǃè! yè rlàŋŋạ to-ɢǃòn srob là lʰo bà
Qon up-tent raise said commander_of_80! so squadron up-Qon tent raise be_around ANIM
Set up camp around Qon! said the commander. So the soldiers put up their tents around Qon...

In normal spoken Tsi, the forms without -d are typically also used in 'postpositional' contexts. Likewise, the distinction between true postpositions and locative verbs is somewhat blurred, especially with ro, which is commonly productively serialised for both genitive and benefactive uses:

(11) qa-'ni ŋǃè ɢǃònù ro! (compare the more correct ạnni ro aqa-ŋǃè ɢǃònù)
apart=3sg head split DAT
split his head open! (=split his head up for him!)

Directional prefixes

There are several directional prefixes which attach, in the absence of any other elements, to the main verb. These should not be mistaken for adpositions or, strictly speaking, locatives of any kind. Inasmuch as they carry independent meaning, they express direction. The initial a- rarely appears in speech:

(a)ya-: around, backwards and forwards
(a)to-: upwards, forwards
(a)šo-: backwards, downwards
(a)qa-: apart, into parts, in all directions at once

However, these prefixes cannot be freely applied to any verb and with many verbs (šo-dòg 'spit', ya-dgo 'mix' etc) the combination of root and prefix is lexicalised. Some roots only appear with prefixes: aqa-qǃʰo 'scatter' for example.

Syntactically directional prefixes appear in the first position in the verbal complex. As such, they prefix to nouns incorporated in the verbal complex where they exist:

(12) ya-quuŋ dgo cçę
round-mud mix ANIM
He is mixing mud

The Tsi verbal complex can be quite elaborate. It consists of, minimally, a main verb (i.e. a single lexical item), which can be simple (a verbal element alone) or complex (including directional prefixes and lexicalised nominal elements). The verb itself can carry certain inflections, and the complex as a whole can also include auxiliaries and serial verbs. The order of elements is approximately:

Directional prefix - incorporated objects - negative - nominal element - main verb - serial verb - auxiliary

Main verb

A single main verb can appear on its own grammatically. Generally speaking, this is only acceptable if the verb is perfective and has an animate object or when the verb is a bare (and somewhat impolite) imperative:

(1) là
lift_up
(he) lifted (him) up/lift (him) up!

The main verb may be a 'compound' verb with a nominal element. In inscription Tsi these are predominantly formed with the largely semantically empty light verb χad.

(2a) ŋáá ọχad
teeth do
(he) bit (him)

In spoken Tsi this is typically reduced to a suffix (-χ or -ọχ after vowels), which is very productive in forming new verbal derivations:

(2b) ŋáá-χ
teeth-VB
(he) bit (him)

The main verb may also include a compulsory (lexicalised) directional prefix, which attaches to the first element in the verbal complex (with the exception of the negative):

(3a) ato-là
up-raise
erect (the tent)!

It can also include a causative prefix ampa- (often reduced to ma- or just m- in speech):

(4a) ato-ampa-là
up-CAUS-raise
make (them) put up (their tents)

In spoken Tsat Tsi these prefixes often trigger gemination of a non-geminated initial consonant:

(3b) to-llà
up-raise
erect (the tent)!

(4b) to-mma-là
up-CAUS-raise
make (them) put up (their tents)

Morphology of the verb

The imperfective-perfective and andative-venitive morphology which plays so prominent a part in the verbal morphology of related languages only now exists in Tsi in frozen lexical pairs. Other than the directional prefixes discussed elsewhere, there is only one major inflectional affix which attaches to verbs, -x/-ç (derived from the historical irrealis inflection). This is used with all subordinating conjunctions and with the classical negative:

(5) ŋǃè ọχ qǃʰạ!
head do.SUB NEG
do not give commands!

(6a) aya-dgo-x cçę-ç içç aŋŋo
around-mix-SUB ANIM-SUB NOM in
while he was mixing it...

Whilst in inscription Tsi it appears on all verbs, the tendency in spoken Tsi is towards two constructions. The first attaches the suffix to the final element of the verbal complex and leaves the other verbs unmarked:

(6b) ya-dgo cçę-ç (içç) aŋŋo
mix-SUB ANIM-SUB (NOM) in
while he was mixing it...

The other construction disallows serial and auxiliary verbs entirely:

(6c) ya-dgo-x (içç) aŋŋo
mix-SUB (NOM) in
while he was mixing it...

Both 4b and 4c show the common tendency in spoken Tsi to reanalyse -ç/x as a contraction of the nominaliser içç and drop the nominaliser (although etymologically they are totally separate).

In inscription Tsi some verbs have irregular subordinate forms, including the omnipresent ọχad whose form is ọχ. Other verbs containing a final dental derived from the perfect also regularly lose it in subordination. As is evident from the examples above, this affix is well into the process of reanalysis as a clitic and these irregular forms are generally being levelled out. Nonetheless, some speakers still use the irregular forms in constructions like the following:

(7) ma cçę-ç ro
sit.ANIM ANIM-SUB GEN
When he sat down

Other speakers however accept mad here.

Negative

In spoken Tsi, verbs are negated with pʰò, a reduced form of the word for 'never'. This is a relatively recent addition to the verbal complex, and is an exception to the general rule that the directional prefix is always the first element in the verbal complex:

(8) pʰò qa-k!a bà
NEG apart=chop take
He didn't chop it (the wood)

When nouns are incorporated into the verbal complex, however, this sometimes results in a shift of the directional prefix to before the negative:

(9a) qa-mdòt pʰò k!a bà
apart=wood NEG chop take
He didn't chop the wood

This coexists with the equivalent construction with an initial negative:

(9b) pʰò qa-mdòt k!a bà
NEG apart=wood chop take
He didn't chop the wood

Inscription Tsi uses the postposed qǃʰạ, as seen in (4).

Honorific

The three etymological classifiers mentioned previously, (o)q!ʰạ, (o)hạ and (o)hąą, are all used in modern Tsi as honorifics. Although in their etymological role as classifiers none of them referred primarily to humans, (o)q!ʰạ is used in speech to refer to significant social superiors and (o)hạ to refer to inferiors. (o)hąą, originally used for animals, is used in a very derogatory form of address. These honorifics are extremely versatile in their syntax, and one of the places they can appear is within the verbal complex:

(10) pʰò qa-mdòt k!a hąą bà jòò?
NEG apart=wood chop HNR take why?
Why didn't you chop the wood?! (from a master to a menial slave)

Although this use of classifiers was ungrammatical in the form of Tsi cultivated into Inscription Tsi, the semantic gap left by their absence - and the discomfort this inevitably caused Tsi scribes and bureaucrats - has meant that this usage has found its way into written language as well.

Serial verbs

The 'serial verb' slot is the freest and the vaguest. The name 'serial verb' is probably more of a diachronic one than a synchronically appropriate one: most serial verb constructions in Tsi as in related languages are grammaticalised to a greater or lesser degree. There are two major kinds of serial construction, locative (applicative) and adverbial.

We have already discussed the locative serial verbs elsewhere. These have an essentially applicative meaning, adding an additional locational object to verbs:

(11) rlàŋŋạ to-srob là lʰo bà ɢǃòn-tʰo
squadron up-tent raise be_around ANIM Qon=PROX
the groups of soldiers camped around Qon

Adverbial constructions typically add an additional nuance which can often be approximated by translating the serial as an adverb of manner. They rarely have entirely transparent meanings, however:

(12) ŋǃee ɢǂʰuɢǂʰu tsà Tsààd-tʰo ay!
go sweat ANIM Tsat-PROX MOD!
He's going (all that way) to Tsat!

Auxiliary verbs

I've already discussed the three main auxiliary verbs (tsa/cçę, g|í, bà) elsewhere, which are effectively aspect/animacy markers (imperfective/animate subject, imperfective/inanimate subject, perfective/inanimate object respectively). Auxiliary verbs differ from serial verbs in that the verbs that they govern (the 'main verb') can be coordinated with one another under one auxiliary. The majority of Tsi sentences will have one of these auxiliary verbs. In a main clause, only perfective verbs with animate objects can appear without an auxiliary, though as mentioned some 'affective' verbs typically take dzo:

(13) Ka-ro-'n maŋ dzo ạnnạ
man=that=1sg hit give 1sg
The man hit me!

There are a few auxiliaries which can replace these, adding additional nuance, as with k!ạ ('pat'), which is perfective, usually with inanimate objects, and expresses the ease or simplicity of doing something:

(14) g!u k!ạ mde bàq-bàq ay?!
string_up pat one terror_bird MOD!?
You caught a terror bird, just like that?

Causatives

Tsi has two causative prefixes, one 'distal' ((a)xo-) and one 'direct' ((a)mpa-). Unlike English Tsi does not distinguish between 'make' and 'let', at least not with its core causative morphology. Rather, the semantic distinction is to do with the extent of participation in the action by the causer. The 'direct' causative expresses a closer participation and is the default. Often the former can be translated with 'have X do something', although the most appropriate translation is very context-dependent. 15a might be said for example of an officer present at the scene, whilst 15b might perhaps be said of an order received from afar.

(15a) ato-srob ampa-là rlàŋŋạ hųų
up-tent CAUS-raise squadron to
he made the squadron put up their tents

(15b) ato-srob axo-là rlàŋŋạ hųų
up-tent DST-raise squadron to
he had the squadron put up their tents

Not only physical distance but also metaphorical distance plays a role, and the distal causative is often used to preserve the dignity of the ritually pure, sometimes to the extent of being added where a strictly causative reading is not necessarily accurate. (16) might be used even where the prince threw the objects out himself.

(16) ŋ|ʰu q!ʰạ axo-q!òp oq!ʰạ bà shạạq
prince HON DST-throw_out HON take stained_with_menstrual_blood
his majesty the prince threw out the bloodstained (clothes)
his majesty the prince had the clothes thrown out

The syntax of causatives is relatively straightforward. The object remains the same, but the original subject is demoted to an indirect object followed by the postposition hųų. This additional object can easily be dropped.

There are three main types of conditional in Tsi.

General

This kind of conditional - perhaps the most common - can express both open and counterfactual conditionals. In Inscription Tsi the conditional clause is subordinated with the conjunction ib:

(1a) ka ŋáá ọχ ib pʰá ŋ|o aya-qǂòp
man teeth do.SUB if, right_hand finger off-cut
if he bit the man, then (they will) cut off a finger from his right hand
if he bit the man, then cut off a finger from his right hand [i.e. imperative reading]
if he had bitten the man, then (they) would have cut off a finger from his right hand

(2a) jèjè roomʰ Tsààd ašo-lʰi ọχ ib tša ŋ|o aya-axo-qǂòp-qǂòp
Jeje this.ANIM Tsat back-forehead do.SUB if, right_foot finger off-DST-cut-DIST
if Jeje should turn back towards Tsat, I will have each of the toes of his right foot cut off
if Jeje had turned back towards Tsat, I would have had each of the toes of his right foot cut off

In spoken Tsi ib is typically dropped unless it is emphasised. The protasis is distinguished from a normal sentence by SOV syntax and by a distinct rising intonation and a pause:

(1b) ka ŋááχ-ax (ib), pʰá ro ŋ|o ya-qǂòp
man bite (if) right_hand GEN finger off-cut
if he bit/bites the man, cut off a finger from his right hand
if he had bitten the man, (I'd) have cut off a finger from his right hand

(2b) jèjè-rų Tsààd-tʰų šo-lʰiχ-hąą-x (ib), tša ro ŋ|o-rą ya-'n xo-qǂòp-qǂòp ro
Jeje-this.AN Tsat-DIST.INAN back-head-HON-SUB if, right_foot GEN finger-PL off-3sg DST-cut-DIST DAT
if Jeje should turn back towards Tsat, I will have each of the toes of his right foot cut off
if Jeje had turned back towards Tsat, I would have had each of the toes of his right foot cut off

Inscription Tsi also permits a construction where the protasis is also subordinated with ib. This rules out an imperative reading:

(3) ka ŋáá ọχ ib pʰá ŋ|o aya-qǂòp-ax ib
man teeth do.SUB if, right_hand finger off-cut-SUB if
if he bit the man, then they will cut off a finger from his right hand
if he had bitten the man, then (they) would have cut off a finger from his right hand

Counterfactual

Although the general conditional structure can be used for hypotheticals as well, spoken Tsi does possess a conditional structure with an unambiguously counterfactual reading - i.e. where the fulfilment of the condition is definitively impossible. This construction involves copying the main verb in the negative, complete with auxiliaries:

(4) ya-q|ʰą dòχ bà pʰò ya-ddòχ bà-x, srii mzạ-mzạ bèd nyè bà ŋǃoo?
away-skin steal INAN NEG away-steal INAN-SUB, strong barter-AUG sell bring INAN, right?
If he had stolen the (terror-bird) skin, he would have sold it for a lot of money, no?

Hypothetical-propositional

This form of conditional is used exclusively with an interrogative apodosis and is usually best translated as 'say I did X'. Confusingly for the learner, this is not a distinct structure in and of itself but rather is structurally identical to a perfective statement of fact:

(5) ya-q|ʰą bèd bà zu yè?
away-skin sell INAN then what?
Say I did sell it - what then?
If I sold it, what would happen then?

(6) qa-'t ŋǃè ɢǃònù ro zu yè?
apart=2sg head split DAT
And what about if I split your head open?

Nominalisations

Sentences can be nominalised straightforwardly using the all-purpose subordinator içç. In spoken Tsi this is typically (though not exclusively) followed by hą 'thing'. Nominalised clauses overwhelmingly have verb-final order. They also overwhelmingly - unless they are very light (i.e. with few internal constituents) - appear after the main verb, unless that too is subordinate:

(7)  mdò rlàŋŋạ ɢǃòn-tʰo to-srob là lʰo bà-x içç hą
see squadron Qon=PROX up-tent raise be_around ANIM-SUB NOM thing
I saw that the groups of soldiers had camped around Qon

Tsi does not have a 'small clause' structure equivalent to English 'I saw the groups of soldiers camp around Qon'. However, the nominalisation structure does behave like a small clause in one quite specific way - despite the fact that the clause as a whole (which has no animacy) or often the inanimate noun hą is formally the object of the main verb, for the purposes of animacy agreement the subject of the nominalised clause is treated as the object of the main verb, as illustrated by 7 as opposed to 8:

(8) qbò bà k!ek ša-ša g|í içç hą
watch INAN stream flow INAN NOM thing
I watched the stream flowing

Nominalised clauses can also be subordinated to postpositions and serial verbs etc, and this is the main strategy for producing adverbial clauses of various kinds:

(9) ya-quuŋ dgo cçę-ç içç aŋŋo...
round-mud mix ANIM-SUB NOM in...
And while he was mixing mud….

(10) Ka-ro-'t maŋ dzo-x içç dži
man=that=2sg hit give-SUB NOM come_from
because the man hit you

There is another subordinator for direct speech, bạạ. This one is a simple quotative and does not impose any changes on the quoted speech (obviously):

(11) g!a ka-ro-'n maŋ dzo ạnnạ bạạ
shout man=that=1sg hit give 1sg QUOT
He shouted: 'that man hit me!'

Although içç can be used for expressions of desire, hope etc, spoken Tsi also has another subordinator, fò (a reduced version of 'want'), often used for such structures:

(12a)  pfųų tsa mdòt k!a-k!a ççe-ç fò
want ANIM wood chop ANIM-SUB NOM
I want to chop some wood

This is synonymous with the (perhaps more formal) structure.

(12b)  pfųų tsa mdòt k!a-k!a ççe-ç içç
want ANIM wood chop ANIM-SUB NOM
I want to chop some wood

Tsi exists in a state of relatively differentiated diglossia, although the differences between the scriptual-bureaucratic inscription Tsi and spoken Tsi are not so gapingly distant as to be mutually incomprehensible. Perhaps it is better to think of inscription Tsi as an unusually differentiated, archaising register of Tsi. It is important to note that the 'ideal' of inscription Tsi (which reflects to a greater or lesser extent the cherry-picked spoken Tsi of centuries past) is typically diluted to a greater or lesser extent by influence from spoken Tsi.

Inscription Tsi makes use of some archaic or unusual vocabulary which is no longer in use in speech. It also makes use of a number of archaic constructions, some of which we have seen in the grammar presented so far. Most characteristically, it maintains the use of a differentiated classifier system (collapsed in spoken Tsi). It does not typically make use of the auxiliaries, and retains more of the morphological machinery used to distinguish perfectives and imperfectives (many perfectives have a final -d, which occurs much more rarely in normal spoken Tsi). In its spoken form it lacks many of the contractions found in normal speech and has some elongated versions of morphemes.

A form of spoken inscription Tsi is used in prayers and magical invocations, in formal addresses to government (which in a sense are the same thing), in legal proclamations and so on. It is in this use (the deliberate preservation of archaising and indirect language) that Inscription Tsi began, and it has a much longer tradition as a spoken register than in writing. However, since the Kangshi script was imported into the Tsi-speaking area, inscription Tsi has also been used in writing and of course on imperial inscriptions. These inscriptions are one of the most visible signs of imperial power throughout the archipelago - even if very few of its residents can read them.

Whilst very few people have any kind of active command of inscription Tsi and a whole small industry flourishes around those who do (magicians and folk healers, priests, scribes, official petitioners etc), there are many native speakers of Tsi - certainly more than any other language in the archipelago. The majority of these speakers are inhabitants of the largest of the southern islands of the archipelago, Haŋŋo-Dze. The coastline of these islands is dominated by the Tsi decapolis, which taken as a whole have historically dominated trade both within the archipelago and (more importantly) from the archipelago out to the rest of the world. The cities are all to some extent differentiated from one another dialectally, particularly the oldest cities. The inhabitants of the smaller rural settlements are famous for their distinctive and bizarre ways of speaking, which are a constant source of jokes for the urban classes. With the expansion of the empire, Tsi-speaking colonies have also sprung up in far-flung places, all of which are gradually developing their own ways of speaking - particularly with the adoption of Tsi by local elites.

The ~ideal Tsi vowel system~ vs Tsat

/i iː u uː/ <ị ịị ụ ụụ>
/ɪ ɪ̃ ɪː ɪ̃ː ʊ ʊ̃ ʊː ʊ̃ː/ <i į ii įį u ų uu ųų>
/e eː o oː/ <ẹ ẹẹ ọ ọọ>
/ə əː/ <ạ ạạ>
/ɛ ɛ̃ ɛː ɛ̃ː ɔ ɔ̃ ɔː ɔ̃ː/ <e ę ee ęę o ǫ oo ǫǫ>
/a aː ã ãː/ <a aa ą ąą>

Final long /aː/ merges with final long /ɛː/ and final long /ɪː/ with final long /i:/. On the other side of the vowel chart, final long /o: ɔː/ merge entirely to /u:/. Meanwhile, final nasal long vowels are shortened (but Tsi retains long final nasal vowels as a result of the later loss of final nasals; this is probably a chain shift):

ya-qquŋ [joq'ʊ̃ː]
k!ǫǫ [k!ǫ]
jòò /ɟù:/
šaa /ʃɛ:/

Short unstressed /a e ɛ/ merge to /ə/ and /o ɔ/ to /o/. Final unstressed short schwa is dropped (in the few derived words with this sound):

rlàŋŋạ ['rlàŋ(]
bek!ek [bək!ek]

The limited prefixial frontness-based vowel harmony observed in all western dialects of Tsi occurs in Tsat, too - the vowel /o/ shifts to /e/ before front vowels (including /a/) and vice versa. In Tsat however this is extended further because of the aforementioned shift, resulting in the alternation of /ə/ and /o/ in a series of prefixes (the harmony is also visible in some function words):

to-llà [təl'là]
aŋŋo [oŋ'ŋo]
ya-qquŋ [joq'ʊ̃ː]

Final nasals are lost, leaving behind nasalised long vowels. Final nasal long vowels are shortened in those words with no inherited final nasal consonant, but retained where they reflect the recent loss of a final nasal consonant:

ya-qquŋ [joq'ʊ̃ː]
k!ǫǫ [k!ǫ]

Consonants

[fʰ] mergest with inherited /w/: fháág [wáːk] 'older sister'

Final stops are devoiced: ktọp [ktop] 'pebble', ktọb [ktop] 'pure water'

Codaic /l lʰ/ are both labialised, producing diphthongs: ŋbą́lh [ŋbõũ] – dagger

Final nasals are lost, leaving behind long nasal vowels:

Final rhotics are debuccalised with lengthening: tbár [tbáːh]

Tsi has three 'honorifics' originally derived from classifiers, as mentioned previously. These honorifics are syntactically versatile: they can attach to nouns and pronouns, stand alone as anaphora, and appear within the verbal complex.

(o)q!ʰạ (ultra-elevated)
(o)hạ (respectful)
(o)hąą (derogatory/animal)

In addition, the absence of any classifier (which we'll call 'unmarked') expresses (rarely) equality or lower social standing.

The etymology of (o)q!ʰạ (originally a classifier for large geographical features and by extension their personified forms as deities) and (o)hąą (originally a classifier for animals) is fairly straightforward. (o)hạ, which derives from a classifier used originally for round objects and tools, is perhaps less transparent. Its usage is rooted in the Tsi equivalence of the concept of 'imperial slave' and 'citizen' (ngkòd) - that is, the elevated social class carrying the greatest amount of prestige are conceptualised simultaneously as property of the God-Emperor. The 'tool' classifier was extended metaphorically to refer to these individuals - a use attested in the oldest Tsi - before being (de?-)grammaticalised and extended by association to all social superiors.

Syntax of honorifics

With nouns and pronouns

The most simplistic use of honorifics is in conjunction with nouns. In this case they typically appear directly after the noun. The initial vowel o- of the ending is dropped after a vowel, and can optionally be dropped after a consonant as well (the complexity of the resulting consonant cluster may play a role in determining this):

srob-(o)hạ 'tent' (of someone respectable)
ka-hąą 'man' (derogatory)
tša-q!ʰạ 'left foot' (of the Emperor)

The plural ending typically follows the honorific. Although the border between independent particle, clitic and affix is particularly vague in Tsi, this is evidence for considering it in the latter group:

srob-(o)hạ-otʰ 'tents' (of someone respectable)
ka-hąą 'men' (derogatory)

Note that incorporated nouns cannot take honorifics.

Anaphoric usage

The anaphoric usage of honorifics is a fairly clear extension of the nominal usage, albeit without a noun attached. As with the 'unmarked' pronouns, the use of anaphoric classifiers is fairly limited to cases of disambiguation. The full form is always used:

(1) ohąą mdòt k!a-k!a-hąą tsą k!ǫǫ aŋŋo
HON wood chop ANIM quarter in
he (the menial servant) is chopping wood in the quarter

Where social superiors are being addressed, these forms are often used considerably more than is normal for pronouns in other contexts.

With verbs

The most complicated position that honorifics can appear is as part of a verbal complex. Here they follow the main verb directly. As with nouns, the o- is dropped after vowels and optionally after consonants:

(2) ktọb-ohạ-rų šo-k|ʰòò-hạ bà qǂeχ-ohạ ho?
water=HON=DEM down=pour-HON IN.PRF please-HON INT?
Should I pour sir some of this water [if I poured this water, would it please you?]?

(3) pʰò qa-mdòt k!a hąą bà jòò?
NEG apart=wood chop HNR take why?
Why didn't you chop the wood?! (from a master to a menial slave)

Usage

Tsi honorifics mark the social relationship between the speaker and the addressee (second person) and the referent (third person) of their conversation. Their usage varies to some extent dialectally, and misuse or underuse of honorifics is a stereotypical feature of rural speech. But for urban speakers in the central areas of the Tsi Empire at least they are fairly standardised.

Determining status

It is fairly unusual for two Tsi to occupy an identical position on the social ladder. Family, profession, gender (male > female) and age (older > younger) are all determinants of relative status. Within most social interactions, the difference is understood but not seriously marked: use of the unmarked (zero-honorific) forms and the respectful ((o)hạ) forms does not necessarily express a significant gulf, and in fact in most Tsi interactions is accompanied by use of familial forms of address expressing a kind of respectful affection.

Consider the following examples:

(4a) Dzo-hạ mdòt k!a-k!a-hạ tsą k!ǫǫ aŋŋo
Dzo=HON wood chop=HON ANIM quarter in
Dzo is chopping wood in the quarter

(4b) Dzo mdòt k!a-k!a tsą k!ǫǫ aŋŋo
Dzo wood chop ANIM quarter in
Dzo is chopping wood in the quarter

Both sentences express the same content, but reflect different relative positions between Dzo and the speaker. In a) Dzo has a higher social standing than the speaker, in b) a lower social standing. The social difference need not be that significant: in 4a the speaker could be his older sister and in 4b his older brother, for example.

4b could also be a long-term friend of the same gender and similar age - one of the few contexts in which it is possible to express near equality in status. In a relationship of this kind, both speakers using the unmarked form to address one another is normal. However, even in this kind of relationship, outside familiar situations a more conventional age differentiation or mutual use of the respectful form is likely to take the place of the unmarked form in the third person (i.e. when discussing the other party with someone else).

However, in interactions with or discussing the highest (imperial, deific) and lowest (menial slave, outlaw) levels of Tsi society, speakers have recourse to the ultra-elevated ((o)q!ʰạ) and derogatory ((o)hąą) forms. Consider (4c):

(4c) Dzo-hąą mdòt k!a-k!a-hąą tsą k!ǫǫ aŋŋo
Dzo=HON wood chop=HON ANIM quarter in
Dzo is chopping wood in the quarter

In this case, Dzo is assumed to occupy one of the lowest rungs of Tsi society. He is likely a slave or an outlaw - or if not, the speaker is deliberately adopting a highly insulting and derogatory form. Note that although this form derives from a classifier used for animals generally, its use with animals is variable (presumably since there is little need to express social difference between an animal and a human). It is almost universal with certain kinds of animals considered to be dirty or contemptible, however, particularly in contexts where the speaker is keen to emphasise the vast difference between the animal and an addressee or referent or themselves:

(5) k!oo-hąą ŋááχ-ohąą bà
dog=HON bite=HON IN.PRF
The dog bit it

The ultra-elevated form, meanwhile, is used when referring to the imperial person, the royal family, and high-ranking levels of the bureaucracy as well as deities (the exact rank after which (o)q!ʰạ has to be used is not entirely clear, and depends to some extent on the status of the speaker as well).

(6) ŋ|ʰu-q!ʰạ xo-q!òp-oq!ʰạ bà shạạq
prince=HON DST-throw_out=HON IN.PRF stained_with_menstrual_blood
his majesty the prince threw out the bloodstained clothes

Inanimate nouns associated with animate nouns often take honorifics through association:

(7) rlàŋŋạ to-llà-q!ʰạ bà srob-oq!ʰạ
squadron up-raise-HON INAN tent-HON
the group of soldiers raised [the Emperor's] tent

Verbal honorific

The underlying principle determining the use of verbal honorifics is essentially subjecthood. In an intransitive sentence, or a transitive sentence involving an inanimate object, the animate subject (or experiencer etc in some cases) determines the honorific to be used. For second and third person subjects, the honorific is straightforward: most commonly either respectful (for a higher-status person) or unmarked (for a lower-status person), with more unusual cases taking the highly elevated or derogatory form:

(8a) pʰò ktọb su-hạ bà
NEG water drink=HON IN.PRF
he (you) didn't drink the water (the person with higher status than me)

(8b) pʰò ktọb su bà
NEG water drink IN.PRF
he (you) didn't drink the water (the person with lower status than me)

(8c) pʰò ktọb su-q!ʰạ bà
NEG water drink=HON IN.PRF
he (you) didn't drink the water (person with extremely high status)

(8d) pʰò ktọb su-hąą bà
NEG water drink=HON IN.PRF
he (you) didn't drink the water (person with extremely low status)

For first-person subjects the default choice for people outside the lowest and highest status brackets is the unmarked form:

(9a) ktọb su bà
water drink IN.PRF
I drank the water

In interactions with social inferiors, however, those entitled to highly-elevated honorifics also use them to refer to themselves:

(9b) ktọb su-q!ʰạ bà
water drink=HON IN.PRF
I (e.g. the Emperor) drank the water

Likewise in interactions with social superiors (though not necessarily among themselves) those on the lowest levels of society are expected to use the derogatory forms with first person referents. A slave might speak to an innkeeper in this style. Even those who would otherwise normally be outside the scope of the derogatory form, however, can sometimes make use of it when addressing very elevated social superiors - the same innkeeper, when brought before the court as a supplicant, might use the same form:

(9c) ktọb su-hąą bà
water drink=HON IN.PRF
I (e.g. a chattel slave) drank the water

However, when more than one constituent of the verb has social status (i.e. is animate), the constituent with the highest social status takes priority, forcing the verb into that honorific form.

(10) bàq-bàq ŋááχ-ohạ
terror_bird bite-HON
The terror bird bit him (someone who is my social superior)

In some cases the superior may not even be a direct constituent syntactically. Consider sentence (2) again:

(11) ktọb-ohạ-rų šo-k|ʰòò-hạ bà qǂeχ-ohạ ho?
water=HON=DEM down=pour-HON IN.PRF please-HON INT?
Should I pour sir some of this water [if I poured this water, would it please you?]?

In this case the subject of šo-k|ʰòò 'pour' is clearly the speaker, but more importantly, the addressee is not an explicitly stated constituent or direct object. Nonetheless, as an implied beneficiary, the addressee still demands an honorific form.

Interrogative pronouns

The interrogative pronouns are as follows:

yèn - 'what'
son - 'who'
fon - 'where'
fondò - 'how'
jòò - 'why'

Their syntax is wh-in-situ:

(1) rlàŋŋạ to-srob là lʰo bà fon?
squadron up-tent raise be_around AN.IM where
Where did the soldiers set up camp?

(2) pʰò qa-mdòt k!a hąą bà jòò?
NEG apart=wood chop HNR take why?
Why didn't you chop the wood?!

(3a) dži-hạ fon?
come_from-HON where
Where did he come from?

Note that occasionally in sentences like (2) above, especially in cases of significant social differences, even interrogative forms can take honorifics:

(3b) dži-hạ fon-hạ?
come_from-HON where-HON
Where did he (that higher-ranking gentleman) come from?

There are also several atonic variants of yèn. One, -yè, attaches to nouns and acts as an interrogative determiner ('which' or 'what'):

(4) ka-hạ-yè dži-hạ Tsààd-tʰo?
man-HON-which come_from-HON Tsat-DST
Which men came from Tsat?

Another, -(y)in, is used in casual speech for interrogatives representing the direct object, and attaches to the final element of the verb phrase:

(5) mdò bà-yin haa?
see IN.PRF=what then?
What did you see then?

Another, ne-, is used with postpositions and locative verbs:

(6) rlàŋŋạ-tʰų kǂay-kǂay tsa ne-lʰo-d?
squadron-DST dance-IT AN.IM what=be_around-PRF
What is it that those soldiers are dancing around?

Closed questions

For unmarked closed questions the particle ho appears either at the end of the verbal complex or the end of the utterance:

(7) ktọb-ohạ-rų šo-k|ʰòò-hạ bà qǂeχ-ohạ ho?
water=HON=DEM down=pour=HON IN.PRF please=HON INT?
Should I pour sir some of this water [if I poured this water, would it please you?]?

To form negative questions the typical method is to replace the normal negative pʰò with dzè, combined with an interrogative (ho is optional but often appears in these constructions):

(8) dzè qa-mdòt k!a-hąą bà ho?
NEG.INT apart=wood chop=HNR IN.PRF ho?
Didn't you chop the wood?!

The particle omu can be used to form a rhetorical question where the implication is that both the speaker and their interlocutor know the answer, likewise with optional ho:

(9) omu qa-mdòt k!a-hąą bà ho?
INT.RHT apart=wood chop=HNR IN.PRF ho?
You didn't chop the wood?!

Personal pronouns

The Tsi personal pronouns are as follows. Each form has a tonic and an atonic form.

ạnnạ -n ndạ -dạ
ạtta -t tạta -ta
ạnni -ni ninạ -na

The atonic pronouns are fairly uncommon. One reason for this is that Tsi is typically fairly pro-drop, especially but by no means only for subjects. The other reason is that their use is somewhat limited by the same system of politeness and respect grammatically encoded with honorifics. In fact, as mentioned above, honorifics often appear in an anaphoric usage standing in for personal pronouns. The tonic personal pronouns are typically only used for people with whom the plain style is acceptable - typically the speaker and those lower on the social spectrum.

Third person tonic pronouns are also restricted to animates - inanimates are referred to with the inanimate demonstrative. (1a) can refer to animate or inanimate objects, whilst (1b) and (1c) are specified for the animacy of the object:

(1a) mdò bà-na ho haa?
see IN.PRF=3pl INT then?
Did you see them then?

(1b) mdò bà ninạ ho haa?
see IN.PRF 3pl INT then?
Did you see them then? (animate)

(1c) mdò bà tʰobʰų ho haa?
see IN.PRF DEM.IN INT then?
Did you see them/it then? (inanimate)

Vocatives

With most social relations locatives are typically (although not exclusively) preceded by ge (probably originally 'I say') in normal conversation and accompanied by honorifics and a respectful title:

(2) ge rlàŋŋạŋǃè-hạ mpe
VOC sergeant-HON father
Sergeant!

(3) yii ge q!ʰèèka zu?
yes clubman young_brother
Yes soldier?

With more distant social relationships - not necessarily ones which trigger remote honorifics - ge may be considered impolite.

Indefinites

Tsi marks indefinite, specific constituents optionally with a preposed mte~mde (inanimate) or mre~mạr (animate). These typically appear in the postverbal space. Consider the following examples:

(1a) nzu bà mte hạnzu
write IN.PRF INDF letter
he wrote (will write) a (specific) letter
he wrote (will write) some (specific) letters

(1b) nzu tsą mte hạnzu
write AN.IMP INDF letter
he was (is, will be) writing a (specific) letter
he was (is, will be) writing some (specific) letters

Although (1a) (as discussed above in the TAM section) is unambiguously specific because it is perfective, without mte (1b) could also be read as generic. mte marks the letter(s) as discourse-relevant ('specific'). Note that although this form is clearly etymologically identical with the postposed numeral mte, it is not restricted to singular constituents, as implied by the plural readings given above.

The plural suffixes -odʰę and rą also force a specific interpretation onto a newly introduced noun. Note that they too can freely co-occur with the indefinite. Compare the following sentences:

(2a) Ya-dgo tsa mte quuŋ-odʰę
round-mix AN.IMP INDF mud-PL
He mixes some mud
He's mixing some mud

(2b) Ya-dgo tsa quuŋ
round-mix AN.IMP mud
He mixes mud
He's mixing mud

(2c) Ya-quuŋ dgo tsa
round-mud mix AN.IMP
He mixes mud
He's mixing mud

The difference between (2a) and the other two sentences is that in (2a) the mud is specific and probably discourse-relevant. In the other two cases it may be background information or generic. We might propose a hierarchy of backgroundness/genericness from (2a) to (2c), but the exact difference between (2b) and (2c) in most cases is not clear. In any case, this distinction - if it does exist - is largely restricted to direct objects.


Definiteness and contrast in direct objects

The above post discussing indefinites and generics noted that indefinites appear in postverbal position. Definites, on the other hand, can appear in either the postverbal or the preverbal ('incorporated') position. There is little semantic distinction between the following two sentences, although 1b shows incorporation of the object into the verbal complex, while 1a has it in the postverbal position. Both can be read as either definite or generic depending on context, and both are very common, with incorporation perhaps slightly more frequent especially as discourse draws on.

(1a) Ya-dgo tsa quuŋ
round-mix AN.IMP mud
He mixes (the) mud

(1b) Ya-qquuŋ dgo tsa
round-mud mix AN.IMP
He mixes (the) mud

A third, albeit more colloquial variant also exists, with the shift of the directional prefix rightwards out of the verb complex to attach to a definite object in the postverbal position. This mainly occurs with the more common directional prefixes and is not acceptable with all prefix-verb combinations:

(1c) Dgo tsa ya-qquuŋ
mix AN.IMP round-mud
He mixes (the) mud

Although there is little semantic difference between these three variants as given above, 1b and 1c (which both display incorporation or genericisation of the noun) do not permit any modification of the noun. This is not the case for 1a - a definite object in postverbal position can be modified by adjectives, take plural and honorific marking, etc etc. Compare 2a and 2b - the latter has full incorporation, whereas the former

2a) kǂay-kǂay tsa lʰod srobhạ-tʰ
dance-IT IMP.AN around tent-HON-PL
They are dancing around the tents

2b) srob kǂay-kǂay tsa lʰod
tent dance-IT IMP.AN around
They are dancing around the tent(s)

Only a postposed definite object can be contrastive (in which case it appears with the demonstrative clitic):

2c) kǂay-kǂay tsa lʰod srobhạ-tʰ-rų
dance-IT IMP.AN around tent-HON-PL-DEM
They are dancing around the tents

3a) kahạ-yè džihạ Tsààd-tʰo?
man-HON-which come_from-HON Tsat-DST
Which men came from Tsat?

2c shows incorporation of a locative argument promoted to direct object by the addition of a locative verb. Note that proper nouns can be incorporated, as can substantivised adjectives:

3b) kahạ-yè Tsààd džihạ?
man-HON-which Tsat come_from-HON
Which men came from Tsat?

4) šo-kǂòd k|ʰòò bà
down=red pour IN.PRF
He poured out the red [water]

Note, however, that only one object can be incorporated into the verbal complex. Where a verb has more than one definite direct object, the choice of which to incorporate is typically conditioned by the relative establishedness of the different nouns in discourse. Both of the latter are possible:

5a) to-kǂòd k|ʰòò bà dù sạạq
up=red pour IN.PRF into jug
He poured the red [water] into the jug

5b) to-ssạạq k|ʰòò bà dù kǂòd
up=jug pour IN.PRF into red
He poured the red [water] into the jug

As well as the converb construction mentioned above, which expresses 'closely bound' purpose and implies the same subject, Tsi also has two more general constructions for expressing purpose. The more standard one is the fairly conventional use of içç, with or without the postposition ro.

(1) ka maŋ dzo ksù-x içç (ro)
man hit give shut_up-SUB NOM (for)
I hit the man so he'd shut up

The second construction uses the more colloquial optative subordinator fò:

(2) to-ddòòbh dži mdòt k!a-k!a ççe-ç fò
DIR-copse come wood chop ANIM-SUB NOM
She came to the copse to chop wood

(3) to-ddòòbh mpa-dži mdòt k!a-k!a ççe-ç fò
DIR-copse CAUS-come wood chop ANIM-SUB NOM
He sent her to the copse to chop wood

(4) ka maŋ dzo ksù-x fò
man hit give shut_up-SUB NOM
I hit the man so he'd shut up

Spoken Tsi has an impersonal perfective verb structure used to express non-volition, in which there is no grammatical subject and the 'logical subject' is expressed as an indirect object. The auxiliary used for this structure is ži, derived historically from dži 'come'.

(1) nạ'rro q!òyo ži
1sg=for get_hungry come
I got hungry/I'm hungry

As is common in other contexts, ro is often treated as a serial verb, allowing dropping of the pronoun:

(2) (ạnnạ) q!òyo ro ži
1sg get_hungry for come
I got hungry/I'm hungry

This structure is very common for non-volitive action in general - again only with perfectives, however:

(3) q!ʰạ-x ro ži
left_foot-do for come
(He) fell, tripped over

It is also used for weather structures:

(4) šo-Tsààd šèèšee ro ži
down=Tsat rain for come
It rained in Tsat

žop ŋǃè ru zò mte doo
whole world DAT language one stand
In the whole world, there was one language.

ninạ ya-mbạ dži ɢǂʰuɢǂʰu-x aŋŋo / dži šạnar q!ab ye-ç dʰędʰę-tʰa hoy / to-srob là này
3pl away-sun come sweat-SUB in / come shinar land be_in-SUB flatland-PL MOD / up-tent raise MOD
And it came about that in their wandering from the east, they came to a stretch of flat country in the land of Shinar, and there they put their tents up.

And they said one to another, Come, let us make bricks, burning them well. And they had bricks for stone, putting them together with sticky earth.
And they said, Come, let us make a town, and a tower whose top will go up as high as heaven; and let us make a great name for ourselves, so that we may not be wanderers over the face of the earth.
And the Lord came down to see the town and the tower which the children of men were building.
And the Lord said, See, they are all one people and have all one language; and this is only the start of what they may do: and now it will not be possible to keep them from any purpose of theirs.
Come, let us go down and take away the sense of their language, so that they will not be able to make themselves clear to one another.
So the Lord God sent them away into every part of the earth: and they gave up building their town.
So it was named Babel, because there the Lord took away the sense of all languages and from there the Lord sent them away over all the face of the earth.

With many verbs reciprocals can be simply expressed with reduplication in context, but often co-occur with the reciprocal pronoun ŋǃèŋǃè (literally 'head-head'). Like other verbs formed with reduplication, all are imperfective.

(1) Kahạ-ro ŋǃèŋǃè maŋ-maŋ-orą tsą
man-HON=that REC hit-REC-HON AN.IM
Him and that man are hitting one another

Note that they force the subject into the plural when there is only the one subject:

(2) Kahạ-rą-ro ŋǃèŋǃè maŋ-maŋ-orą tsą
man-HON-PL=that REC hit-REC-HON AN.IM
Those men are hitting one another!

Reflexives are formed similarly with a single iŋǃè, etymologically a possessed form of 'head':

(3) Kahạ-ro iŋǃè maŋ-orą tsą
man-HON-that REF hit-HON AN.IN
That man hit himself