Sint Verbal Morphology

[chris_notts]

This is the reorganised overview of verbal categories in my new project, provisionally called 'Sint'. I've converted it from XeTeX to plaintext... there are some specific areas I'd love feedback on that I'll add to a following post. The actual ordering of the morphemes, which isn't given in the excerpt below, is:

[NEGATIVE-AGR1-PREVERB-AGR2]-[INCORP]-[ROOT-VOICE-ORIENTATION/ASSOCIATED.MOTION-MOOD-AGR3-PLURAL]

(This was also posted to the CBB yesterday, but the CBB seems to have temporarily died...)

Summary of Verbal Inflectional Categories

1. Agreement and Pragmatic Status

In a Sint verb there are four slots which related to verbal agreement, namely AGR1, AGR2, AGR3 and PLURAL, but the verb only agrees with at most two arguments. Which slots are occupied depends on the person and pragmatic status of the core arguments.

Focus

The most important rule is that contrastively focused arguments do not control agreement. Focus is defined in more detail elsewhere, but in simple terms a focused noun phrase occurs pre-verbally in the same clause. Post-verbal noun phrases are not contrastively focused and do control verb agreement.

Inherent and Discourse Topicality

In the case were a verb agrees with both the actor and patient, the agreement pattern depends on which argument is highest ranked,
with speech act participants ranking higher than third persons, and more discourse topical third persons ranking higher than less topical third persons. The highest ranked argument is called the \emph{pivot}, as described in more detail in section \ref{sec-pivot-definition}.

The rest of this section will use the following terminology:

S = non-focal argument of intransitive
(S) = focal argument of intransitive
A = non-focal actor
P = non-focal patient
(A) = focal or non-referential actor (`somebody')
(P) = focal patient

Taking into account verbal transitivity and the constraint that focal arguments always rank low, there are seven possible cases of actor and patient focus and pivot selection, which are shown in table \ref{tbl-agreement-patterns}. The table is divided into three sections to reflect similar verbal inflectional patterns.

Direct
S
A > P
A > (P)

Inverse
A < P
(A) < P

Neutral
(S)
(A) = (P)

The following sections will describe the three different agreement patterns.

1.1 Direct Agreement

In direct agreement forms, the single intransitive argument or transitive actor control the AGR2 and PLURAL agreement slots. If there is a non-focal patient, the patient controls the AGR3 slot. If the patient is focal (pre-verbal), then the verb shows no agreement with it.

The following demonstrate the S, A > P, and A > (P) cases:

niterk
n-i-terk
1EXC-P-sing
`I sang'

nimoolah
n-i-moola-h
1EXC-P-pour-3N
`I poured it'

la seenk navàrk
la seenk na-∅-vàrk
D.F bowl DOWN-3F-put
`She put THE BOWL down'

1.2 Inverse Agreement

In inverse agreement, the transitive actor controls AGR1 agreement. Agreement with the patient is split based on person: a speech act participant controls the AGR2 and PLURAL slots, whereas a third person patient controls the AGR3 slot.

If the actor is focused or non-specific (`somebody'), then special forms are used in AGR1 which neutralise person, gender and number distinctions. The following examples show inverse forms with focal, non-focal, and impersonal actors:

weronterkım
wer-o-n-terk-m
3PL-P-1EXC-sing-ANTI
`they sang to me'

weraterkmıt
wer-a-terk-m-t
3PL-D-sing-ANTI-3F
`they sang to her'

čo kens jaterkmıt
čo kens e-a-terk-m-t
D.PL people FOC-D-sing-ANTI-3F
`THE PEOPLE sang to her'

inaterkmıt
in-a-terk-m-t
IMPER-D-sing-ANTI-3F
`someone sang to her'

The inverse agreement pattern with a focal agent marker is also used in the imperative mood. Compare the following with the example above:

jaterkmıt
e-a-terk-m-t
FOC-D-sing-ANTI-3F
`Sing to her!'

1.3 Neutralised Agreement

Agreement is completely neutralised when all arguments are either focal or impersonal (`someone'). In this case, e- or in(e)- occur in AGR1, and no other agreement markers occur in any other slot. In intransitive clauses this is quite common:

na kiins jonaa
na kiins j-o-naa
D.M man FOC-P-go
`THE MAN came'

inonaa
in-o-naa
IMPER-P-go
`Someone came'

Transitive clauses with two focal arguments don't occur. However, clauses with impersonal in(e)- and a focal argument are possible:

ki čamka inenitsvek
ki čamka ine-nits-vek
I.PL rabbit IMPER-BEHIND-follow
`Someone's hunting RABBITS'

1.4 Interaction Between AGR2 and PLURAL Slots

In verbs where the AGR2 slot is occupied, the interaction between AGR2 and PLURAL is complicated. The basic function of PLURAL is to mark a plural controller, but there are the following complications:

1. The first person inclusive is inherently plural, so in this case PLURAL marks the difference between a minimal inclusive (2 people) and a more numerous inclusive (3+ people).
2. Sint has a single plural for masculine and feminine genders, and no plural at all for neuter nouns. Verbal agreement reflects this, so in the case third person plural agreement, AGR2 cannot contain an overt gender marker.

Taking into account these constraints, all possible combinations of AGR2 and PLURAL are given in the following table. In the table, the word initial allomorph of the AGR2 morphemes is listed; for the other allomorphs see the section on AGR2 agreement.

AGR2, PLURAL, Number, Meaning
n-, -, singular, I
n-, -n, plural, We (excluding you)
ječ-, -, dual, You and I
ječ-, -n, plural, You, I and others
w-, -, singular, You
w-, -n, plural, You all
k-, -, singular, He (masculine)
-, -, singular, She (feminine)
s-, -, any, It, they (neuter)
-, -n, plural, They (non-neuter)

1.5 Summary of Agreement Patterns

In order to select an agreement pattern, the speaker must determine:

1. Which argument is the pivot, if any, according to the criteria in chapter 2
2. Which argument, if any, is contrastively focused
3. Whether any arguments are speech act participants (SAPs)

Based on this information, the controllers of the different agreement slots can be summarised as follows:

AGR1
This slot is controlled by both non-pivot 3rd person A agreement, by any A/S when it is contrastively focused, and is also used
to mark impersonal A/S.

AGR2 and PLURAL
These slots are controlled by a third person A/S pivot, and by a SAP pivot in any role.

AGR3
This slot is controlled by a 3rd person P, and by a SAP non-pivot P.

2. Mood

All verbs are marked for one of four moods: indicative, optative, potential and imperative. The fundamental split is between actions which are presented as realised, marked by the indicative, from actions which are not. Unrealised actions are then subdivided into the potential mood, which deals with possibility, and the optative and imperatives moods which deal with desirability and obligation.

2.1 Indicative

The indicative is used for specific, realised actions in the past and present. It does not cover future events, habitual events, or events which the speaker is unsure of or does not assert.

2.2 Potential

The potential is used with events about which the speaker has less confidence, and which are in some sense predictions. This includes not just past and present uncertainty, but also all future references and habitual aspect. The link between the habitual and uncertainty is that habitual behaviour is fundamental inductive: the speaker identifies or infers an ongoing pattern from limited observations.

2.3 Optative and Imperative

The optative is used for events which the speaker thinks should happen or desires. It is used in deontic expressions, polite orders, hortatives, and subordinate clauses of purpose. The imperative is used to give direct orders, but otherwise has a much more limited distribution than the optative.

2.4 Examples of Mood Distinctions

The following examples demonstrate the contrast between the four moods:

katerk
k-a-terk
3M-D-sing
`He sang'

katerkai
k-a-terk-ai
3M-D-sing-POT
`He (might, will, usually, ...) sing(s)'

katerkıl
k-a-terk-l
3M-D-sing-OPT
`Let him sing!'

jaterk
j-a-terk
FOC/IMP-D-sing
`Sing!'

3. Tense and Aspect

Tense and aspect are much less grammaticalised than mood and are not marked by dedicated morphemes within the verbal word. Tense is marked indirectly by mood in some cases, since the indicative is only used for past and present actions, but this is limited.

The indirect marking of aspect is more complex, and involves four different groups of morphemes. The first is the mood suffixes, since habitual aspect is one meaning of the potential mood. The alternation between the potential and indicative can distinguish between a bounded number of events and an inductive pattern of events.

The second source of aspect marking is the posture and associated motion suffixes. These suffixes may sometimes mark a distinction between perfective and progressive/stative aspect, but their use to express aspect is complex and depends on the verb's aktionsart and semantics. The following table summarises the possible aspectual uses of these suffixes.

Stance = stative, perfect, progressive
Ongoing motion = progressive, continuative, prospective
Preceding motion = inchoative

Because aspect is not the primary meaning of these suffixes, the aspectual meaning is not always available or the morpheme may be ambiguous between an aspectual and motion interpretation.

The last two sources of aspect are different in nature to the first two, because they are more derivational and less inflectional. These affect the lexical aspect of the verb, the distinction between atelic, unbounded activitiesa and telic, bounded activities. The first of these is the preverb system. Many of the preverbs, which encode direction and path of motion, may mark telicity / boundedness when added to non-motion verbs. This is a change in lexical aspect, and as such the preverb(s) that a given verb root selects may be unpredictable and adding a preverb may cause other changes in meaning.

An example which combines the first three sources is:

nakwakaneh ki kars
na-k-wak-kan-e-h ki kars
DOWN-3M-burn-GO.ALONG-POT-3N I.N house
`he would keep on burning down houses'

To break this down:

1. wak- is an atelic verb meaning burn, which does not require that the patient is completely burnt
2. na-wak- `burn down' is a telic compound verb in which the patient is completely burnt
3. na-wak-kan- includes the motion suffix -kan which either means `do continously while moving' or `keep on doing'
4. na-wak-kan-e includes the potential suffix -e, which adds the habitual meaning

The final derivational aspect marking system is patient noun incorporation. If the patient of a transitive verb is incorporated then a new verb is derived that is non-specific as to the number of patient entities affected. That is, a telic transitive verb describing a bounded action can be converted into an unbounded activity. For example:

nitsınčamkavek
nits-n-čamka-vek
BEHIND-1EXC-rabbit-follow
`I rabbit hunted'

4. Negation

Negation is marked by verbal prefixes. There are two prefixes, one v(i)- which is used in the indicative and potential moods, and another is-, which is used with optative and imperative moods.

5. Path, Motion, Orientation and Movement

Notions of space are strongly grammaticalised within the Sint verb, and are more prominent than either tense or aspect. There are two
slots in the verb which relate to space and motion, but with differences in function and semantics. As discussed in the previous sections, these morphemes also have other grammatical functions, but their core semantics are spatial.

5.1 Stance and Associated Motion

The first, and simplest, are the stance and associated motion suffixes. These are inflectional in nature and mark that either that motion preceded, coincided with, or followed the action, or describe the stance taken when performing the action. They show nominative alignment: they always relate to the motion or stance of the A/S argument. A morpheme of this type is sufficient to unambiguously add or removed a motion component to/from an action. For example:

niterkten
n-i-terk-te-n
1EXC-P-sing-GO.AND-PL
`we went and sang'

The verb terk- `sing' alone is ambiguous between a stationary singing action and one concomitant with motion. However, when the associated motion suffix -te is added, the singing is unambigously related to motion of the A/S.

The purpose of these suffixes is not to be the primary means of describing motion and position, in the same way that grammatical tense is not the primary means of providing detailed temporal descriptions in other languages. The fact that a language has a past tense does not obviate the need for very common expressions such as `now', `yesterday', `last week' etc. In the same way, motion in Sint is normally established using motion verbs in separate clauses, not suffixes, and other actions are then anchored to the trajectory using the associated motion suffixes. The suffixes serve to integrate motion and non-motion verbs into a unified spatial flow, in the same way that tense suffixes integrate events into a unified temporal flow.

5.2 Preverbs

The second set of spatial morphemes are the preverbs. These are different to the associated motion suffixes in the following ways:

1. Preverbs do not encode motion. They are ambiguous between describing static spatial relationships and motion, and this ambiguity can only be resolved by the verb root with which they collocate.
2. Preverbs describe the spatial configuration or path of the absolutive (S/P) argument.
3. With non-motion and non-locative verbs, preverbs tend to lexicalise into compound verbs with differences in lexical aspect, transitivity, or other semantic changes compared to the unmarked verb root.

To illustrate the first two these points, consider the following two examples:

menvàrkıt o na kotla
me-n-vàrk-t o na kotla
ON.P-1EXC-put-3F OBL D.M table
`I put it on the table'

menki na kotla
me-n-ki na kotla
ON.P-1EXC-stand D.M table
`I stood on the table'

In both cases, the preverb me- describes a spatial configuration in which the figure is supported on the horizontal surface
of the ground. But one example describes a motion event which puts the figure into that configuration, and the other describes a
static spatial arrangement. Similarly, in the transitive example it is the patient (P) which enters into the configuration, not the actor (A).

The most common use of the preverbs with non-motion verbs is an extension of their core semantics. Preverbs vary in whether they imply
a definitive change of state or not, and to what extend they imply the presence of a specific ground. Those describing the arrival or crossing of a boundary such as -ma- `on' or -as- `out' are extended to indicate non-spatial changes of state, and therefore
to derive telic versions of atelic verbs. And those that imply a specific, bounded ground such as -ma- `on' may also have
a transitivising function. In addition, there are large numbers of semantically more unpredictable preverb-verb pairs to form semantically
distinct compound verbs.

5.3 Interaction Between Preverbs and Associated Motion

While it might be expected that preverbs and associated motion suffixes interact semantically to describe motion, in fact their co-occurence is limited and their semantic combination even more so. Preverbs almost never elaborate on the path of associated motion, and associated motion morphemes do not license the spatial meanings of preverbs.

There are several reasons for this. The first is that in transitive clauses associated motion and preverbs relate to different arguments of the verb. Consider the following example:

šekvàrkıčkit
še-k-vàrk-ı-čki-t
UP-3M-put-DO.PASSING-3F
`he put it up in passing'

In this example, the preverb specifies the path of the patient, whereas the associated motion suffix relates to the actor.
In order for the preverb to modify the path of the associated motion suffix, the verb would have to be intransitive, but then collocational
restrictions come into play. Associated motion suffixes are typically not used on verbs which already encode motion of the S/A, and
conversely on non-S/A-motion verbs the preverbs often encode aktionsart or perform non-spatial functions.

Finally, static locational expressions can take both a preverb with spatial semantics and an associated motion suffix, but here the
preverb expresses a static locational meaning related to the verb root, instead of modifying the associated motion:

šenaite
še-n-nai-te
UP-1EXC-perch-GO.AND
`I went and perched up (there)'

There is one exception to this general separation of functions. The deictic notions of `here' is encoded either by a deictic vowel or the preverb ti-, and this may have scope over either the verb stem, the associated motion suffix, or both. For example:

šikvàrktet
ši-k-vàrk-te-t
UP.P-3M-put-GO.AND-3F
`He went and put it up hither'
`He came and put it up hither'

6. Incorporation

Incorporation of nouns and adjectives is very productive in Sint. A single stem without inflections can be incorporated before the verb root to perform the following functions:

1. Possessor raising (limited to body parts and inalienably possessed objects)
2. Classification of the underlying patient of transitive verbs
3. Valency reduction
4. Expression of manner, event type or circumstance

The incorporated noun is just that, a noun or adjectival stem. It cannot be a syntactic phrase, e.g. a noun phrase, and simple, generic nouns are favoured. In cases where a morphologically complex stem is incorporated, this is normally a well-established nominal compound in its own right.

While a lot of noun roots can be incorporated, not all are used equally frequently, and some nouns aren't incorporated at all or rarely so. Body parts are very common in possessor raising constructions, and for classificatory and valence reduction purposes basic vocabulary with a generic sense are preferred. The most diverse array of nouns occur in manner, depictive and resultative incorporation, where the incorporated stem can range from a concrete instrument (knife-cut), characteristic location (church-marry, mountain-climb), to an adjective (quiet-go).

Incorporated nouns do not refer in the same way that noun phrases do: they characterise the event in some way rather than identifying a referrent as participating in it. Noun stem may show a shift in meaning when incorporated, and some also show irregular phonological reduction such shortening and loss of unstressed syllables. As an example of both phenomena, čaao `water' shows shortening to čao- when incorporated, and it may also be used in the sense of `liquid':

arınčaoseečat
ar-n-čaao-seečat
IN.DENSE-1EXC-water-jump
`I jumped in (to any liquid)'

Often, there are alternative ways of expressing the same event with and without incorporation. Incorporation is favoured:

1. When it allows the promotion of a more animate referent to direct clausal argument (possessor raising)
2. To reduce the prominence of backgrounded referents (classification, valency reduction)
3. When it removes a non-specific / non-referential argument from the clause (valency reduction)
4. As a way of packing complex actions into a single clause or deriving new event types (classification, manner, depictive, resultative)

The first three of these all have in common reduced prominence of backgrounded material. The choice between an incorporated and non-incorporated version of the same clause is often driven by pragmatic factors such as the specificity and referentiality of nouns on the one hand, and whether they are in focus or backgrounded on the other hand. The final use of incorporation relates to the lack of other ways to form complex clauses in Sint compared to a language like English.

6.1 Possessor Raising

In possessor raising incorporation, a possessed noun in absolutive function is incorporated into the verb, and the possessor becomes an argument of the verb. This is most common with body parts. Compare the following with and without incorporation:

osaìlkan tles jerım
∅-o-saìl-kan tles jer-m
3F-P-hurt-GO.ALONG P.F hand-1SG
`my hand keeps on hurting'

nijersaìlkan
n-i-jer-saìl-kan
1EXC-P-hand-hurt-GO.ALONG
`my hand keeps on hurting (lit: I keep on hand-hurting)'

The difference between these is whether the exact location of the pain is in focus, or its effect on the person feeling the pain. The first example would be an appropriate answer to `What hurts?', with the location in focus, whereas the second example would be a better answer to `What's wrong?'.

Compare also the following example:

kamàščelıč čo čamka
k-a-màts(a)-čel-č čo čamka
3M-D-flesh-eat-3PL D.PL rabbit
`He ate the rabbits (their flesh)'

6.2 Valency Reduction and Absolutive Classification

These two functions are similar apart from the transitivity of the resulting verb. In classificatory incorporation, a generic noun derives a more specific transitive verb which nevertheless retains patient agreement and its ability to take an overt patient argument. The verb may also be made morphologically intransitive, in which case the single remaining argument is the actor. Compare:

kametsktaailıh ke šìvoi
k-a-metsk-taail-h ke šìvoi
3M-D-fish-buy-3N D.N trout
`he (fish-)bought trout'

kametsktaail
k-a-metsk-taail
3M-D-fish-buy
`he fish-bought'

6.3 Circumstantial, Manner and Resultative Incorporation

A variety of stems describing the circumstances of the action can be incorporated. Commonly these stems can express the instrument
or general location of an action. In the following example, a well established instrument-verb compound tarnaa `walk', literally `foot-go', recursively incorporates the location tsem(hır) `mountain':

waktsemtarnaa
wa-k-tsem(hır)-tar(s)-naa
AROUND-3M-mountain-foot-go
`He mountain-walked around'

Adjectival or nominal roots may be incorporated to express manner of action:

tlašaìhnaa
tla--šaìh-naa
ALONG-3F-quiet-go
`she snuck along'

7. Transitivity, Grammatical Voice and Valence

In Sint, clausal transitivity is overtly marked either by agreement morphemes or, in the case of focal patients, by the presence of a
fronted focal noun-phrase. This overt marking of transitivity allows, for some verbs, zero derivation between transitive and intransitive
uses. Verbs which allow this are almost always S = P ambitransitives such as wak `burn'.

In addition to zero-marked transitivity alternations, there are five other other ways to change the argument structure of a verb. There are three suffixes, the anti-passive, causative, and reflexive, which are dedicated voice morphemes, and incorporation and preverbs also have valence adjustment functions.

7.1 Antipassive - Dative

The suffix -m- is used to either convert a transitive root into an controlled intransitive root by eliminating the patient (antipassive), or to promote a recipient, beneficiary or other oblique to control the patient agreement.

Most ambitransitive verbs in Sint are S=P ambitransitives, and -m- in this case is the only way to produce an intransitive whose only argument is the actor. Compare the following three examples:

awak
Ø-a-wak
3F-D-burn
`it burnt'

awakıt
Ø-a-wak-t
3F-D-burn-3F
`she burnt it'

awakım
Ø-a-wak-m
3F-D-burn-ANTI
`she burnt (things)'

When -m- co-occurs with patient agreement or a patientive noun phrase, it's interpreted as the recipient, beneficiary, or some other interested party:

tonwakım
t-o-n-wak-m
3F-P-1EXC-burn-ANTI
`she burnt (things) for me'

Intransitive motion verbs, including detransitised motion verbs, can take a destination or source locative with core case marking. The role of a promoted locative is determined by the preverb used with the motion verb. -m- can be used to promote locatives on transitive motion verbs:

nitlevlım ke ikaar
n-i-tlevıl-m ke ikaar
1EXC-P-carry-ANTI D.N camp
`I carried (it) to camp'

Semantically, -m- combines the functions +agent and -patient. If the verb is transitive anyway, then by implicature the patient-like NP must be occupying another semantic role. The exact role of this NP is underspecified, apart from that by Gricean implicature the role is not one which could be covered by the other argument promoting voices.

Because -m- is +agent, it only combines with verbs describing volitional actions. The following example sounds odd because it implies that being asleep is a volitional activity:

tontohom
t-o-n-toho-m
3F-P-1EXC-sleep-ANTI
`?she slept for me'

The only interpretation that makes sense in this case is inchoative, that the actor went to sleep for someone else.

7.2 Causative - Instrumental - Comitative

The suffix -(a)s-}functions as both a causative and applicative. In its causative meaning it can be applied to any verb stem to indicate causation. With verbs describing volition actions, whether transitive or intransitive, -(a)s- can also promote an oblique comitative or instrument to control patient agreement on the verb. However, this reading isn't available with non-volitional (unaccusative) intransitives.

The following examples show the possible meanings with intransitive unergative and unaccusative verbs and with a transitive verb:

nitohosıt
n-i-toho-s-t
1EXC-P-sleep-CAUS-3F
`I slept her (e.g. of a baby)'
`*I slept with her'

nikaasasıt
n-i-kaas-as-t
1EXC-P-run-CAUS-3F
`I made her run'
`I ran with her'

ničoksasıt
n-i-čoks-as-t
1EXC-P-hit-CAUS-3F
`I made her hit (something)'
`I hit (something) with her'

Note that the first example cannot have the comitative reading because the verb root is unaccusative, i.e. it doesn't describe a volitional action. In some cases, a similar meaning can be obtained using the preverb -kal- `together':

kalıntohon
kal-n-toho-n
TOGETHER-1EXC-sleep-PL
`We slept together'

The use of the preverb kal- together with -(a)s to strengthen the comitative interpretation and rule out the causative is also common:

kalınčoksasıt
kal-n-čoks-as-t
TOGETHER-1EXC-hit-CAUS-3F
`I hit (something) with her'

A unified account of the semantics of -(a)s- is as follows. -(a)s- adds a new argument which has partial or complete control over the event. If the event already had an unaffected, actor-like argument, the resulting predicate is ambiguous between the case where the original actor is robbed of control (the causative ~ instrumental meaning), and the case where it retains some control (the comitative meaning). On the other hand, if the event only had a patient-like argument, adding a new controlling argument doesn't affect it, so only the causative meaning is possible.

Another important point to note that the suffix -(a)s- can't create formally trivalent verbs. When applied to a transitive verb, it demotes the original patient to an oblique in favour of the argument it adds. For the same reason, it can't be applied recursively.

7.3 Reflexive and Reciprocal

The marker -ša- marks reflexive or reciprocal action. It attaches to transitive verbs and forms intransitives, as in the following example:

niwakša
n-i-wak-ša
1EXC-P-burn-REFL
`I burnt myself'
`I got burned'

As the gloss above suggests, \emph{-ša-} also has an anti-causative function.

7.4 Preverbs

Preverbs may be used to make intransitive verbs transitive. Which preverb(s), if any, an intransitive verb selects can be unpredictable, and there may be multiple preverb-verb combinations with different semantics. For example, terk `sing' is an intransitive verb, but as-terk `sing out' is a transitive verb whose patient is the song or words which are sung. Typically, preverbs used to change valence also convert atelic verbs into telic verbs.

7.5 Incorporation

One of the functions of incorporation is valency reduction. By incorporating the patient noun into the verb, a formally intransitive can be created which describes the activity of performing the action on an unspecified number of patients.

[chris_notts]

There are a few areas feedback would be good:

1. Complexity of the agreement system

The agreement system is fairly complicated, because of a combination of covert hierarchical agreement (there is no inverse marking, but choice of 2 out of the 3 agreement slots determines relative topicality etc.) and anti-agreement effects with both focal As and Ps. I'm not aware of any language that definitely combines some form hierarchical agreement and anti-agreement effects, but I think both features exist in Salish (IIRC some Salish languages have obligatory passive use with SAP patients) so maybe there's a Salish language which has both?

It's also verb hard to find good examples of languages which show fronted/focused P anti-agreement, although gach gave me some ideas where to look in a different thread.

2. Interactions between fixed order (templatic) derivational markers

Both the preverbs and incorporated nouns can serve derivational functions, but these occur in a fixed order which raises certain issues. For example, the verb "hunt" is nits-vek, literally "behind follow". If I wanted to incorporate a patient noun into this, e.g. čamka "rabbit", which literally falls between the preverb and the verb root, then is the order of derivation:

nits-vek + čamka = nits-čamka-vek
OR nits + čamka-vek = nits-čamka-vek

In other words, does the irregular meaning of preverb-verb unit survive having a noun inserted into the middle of it, or is čamka-vek a new root which could maybe have a different meaning when combined with the preverb nits-?

Of course, many of the Germanic languages have separable verb + preposition/particle combinations with unpredictable meanings, but these languages don't have productive noun incorporation so they don't really answer the question of how natlangs do it. Neither, as far as I'm aware, do Laz or other Caucasian languages with similar spatial preverbs. I'm not aware of any language which combines (i) spatial preverbs which also serve derivational functions, (ii) noun incorporation, and (iii) ordered in such a way that the incorporated noun is placed between the root and the preverb.

[chris_notts]

3. Use of the deictic vowel

I don't think I covered it in the overview, but the pre-stem (pre-verb + agreement markers) includes an often obligatory hither vs thither distinction marked by vowel alternation (vowel raising for motion towards the deictic centre). The alternation:

1. Is present if there is no preverb (but see restrictions below)
2. Is shown with some preverbs but not with others
3. Can be optionally reinforced / marked for non-alternating preverbs by ti- "hither"
4. Must take the proximate form if there's a first person argument

"Describing Morphosyntax" suggests that Otomi has a similar inflectional alternation between actions "here" vs "there", but I haven't really had chance to look it up yet. What worried me, beside the general concept, is whether marking it via vowel alternation is not prominent enough, and also whether this would be stable. I can see two obvious pathways for evolution:

1. Since the "hither" form is obligatory with 1st person core arguments when the vowel alternation is available, it could evolve into secondary 1st person agreement marking
2. Since it suggests an on-stage / off-stage distinction (especially if the deictic centre doesn't have to be the speech location), it could be a secondary marker of 3rd person argument status, e.g. a covert proximate/obviative, animate/inanimate, or high/low empathy marker of some kind.

[cedh]

First of all: This is awesome!
(And, I believe, very realistic.)

2. Interactions between fixed order (templatic) derivational markers

Both the preverbs and incorporated nouns can serve derivational functions, but these occur in a fixed order which raises certain issues. For example, the verb "hunt" is nits-vek, literally "behind follow". If I wanted to incorporate a patient noun into this, e.g. čamka "rabbit", which literally falls between the preverb and the verb root, then is the order of derivation:

nits-vek + čamka = nits-čamka-vek
OR nits + čamka-vek = nits-čamka-vek

In other words, does the irregular meaning of preverb-verb unit survive having a noun inserted into the middle of it, or is čamka-vek a new root which could maybe have a different meaning when combined with the preverb nits-?

Of course, many of the Germanic languages have separable verb + preposition/particle combinations with unpredictable meanings, but these languages don't have productive noun incorporation so they don't really answer the question of how natlangs do it. Neither, as far as I'm aware, do Laz or other Caucasian languages with similar spatial preverbs. I'm not aware of any language which combines (i) spatial preverbs which also serve derivational functions, (ii) noun incorporation, and (iii) ordered in such a way that the incorporated noun is placed between the root and the preverb.

Here's an example from Winnebago (Siouan):

Kook-ra	ha-nan-zhin-je-enan.
box-DEF	SUPRAESSIVE-by_foot-stand-AUX-DECL

'It is standing on the box.'

I'm not sure whether -nan- 'by foot' represents an instance of productive noun incorporation here, but even if it is not, it seems likely that it has originated from an incorporated body part noun. If that is true, the order of morphemes is clearly PREVERB - INCORP - ROOT. Semantically, the incorporated noun seems to be part of a complex verb stem here, with the preverb added later in the derivation.

(Source: Craig, Colette & Hale, Kenneth, Relational preverbs in some languages of the Americas, Language 64/2 (1988), p. 312-344; here: p. 317)

3. Use of the deictic vowel

I don't think I covered it in the overview, but the pre-stem (pre-verb + agreement markers) includes an often obligatory hither vs thither distinction marked by vowel alternation (vowel raising for motion towards the deictic centre). The alternation:

1. Is present if there is no preverb (but see restrictions below)
2. Is shown with some preverbs but not with others
3. Can be optionally reinforced / marked for non-alternating preverbs by ti- "hither"
4. Must take the proximate form if there's a first person argument

"Describing Morphosyntax" suggests that Otomi has a similar inflectional alternation between actions "here" vs "there", but I haven't really had chance to look it up yet. What worried me, beside the general concept, is whether marking it via vowel alternation is not prominent enough, and also whether this would be stable. I can see two obvious pathways for evolution:

1. Since the "hither" form is obligatory with 1st person core arguments when the vowel alternation is available, it could evolve into secondary 1st person agreement marking
2. Since it suggests an on-stage / off-stage distinction (especially if the deictic centre doesn't have to be the speech location), it could be a secondary marker of 3rd person argument status, e.g. a covert proximate/obviative, animate/inanimate, or high/low empathy marker of some kind.

It seems that your language distinguishes only five vowel qualities, /i e a o ɨ/. In this configuration, I think marking an important distinction by vowel alternation should be prominent enough. (I've been having similar worries with regard to my own conlang Buruya Nzaysa, but there the situation is much less distinctive phonetically, often relying on a /e o/ vs. /ɛ ɔ/ difference in word-final unstressed open syllables.)

As for whether the "here" vs. "there" distinction would be stable semantically, I think you don't have to worry. The two possible pathways you listed are good options if you want to create a descendant language at some point, but neither of them is inevitable. "Here" vs. "there" is salient enough in many discourse situations that it could easily remain stable for a long time, especially if your language does not use separate "here" vs. "there" adverbs very often. That said, I think it's not unlikely that the reinforcement with ti- would eventually become obligatory given enough time, leaving the vowel alternation free to be reanalysed as marking something else. (Also, past or future actions involving a 1st person argument could easily start allowing distal marking too, for example to indicate that the speaker did or experienced something while being at a different location.)

Here's some (quite basic) information about Otomi:
(Source: Palancar, Enrique, The conjugation classes of Tilapa Otomi, Linguistics 50/4 (2012), p. 783-832; here: p. 792-793)

In addition, in Table 2 there are some tenses with local semantics that provide motion information about the subject: the ambulative (an alternative term for “perlocative”) portrays the subject as moving about and the andative depicts the subject as moving away from the speech act situation in order to perform the action.

In other tenses, local values may be realized in the inflected form by means of additional morphemes associated with a specific tense, as shown in Table 3. For example, an andative value can be encoded in the ambulative realis by means of the element -r. Similarly, a cislocative value that portrays the subject moving towards the speech act situation can be encoded in the ambulative realis by means of a labial affix, which surfaces as an infix, and which is at times complemented by the formative tí. In other tenses, the same labial affix is used to express a translocative value. This value portrays the subject as either doing the action in a different place than the speech act situation or as moving away to such a place (i.e., it may convey andative semantics). This is shown in Table 3, where the labial affix appears as an infix.

Table 2. The grammatical tenses of T-Oto

Code: Select all

Realis
Present     continuous    grá hpeni       ‘you’re washing it now’
            habitual      grú hpeni       ‘you commonly wash it’
Ambulative                gá hpeni        ‘you wash it away (here and there)’
Imperfect   continuous    grá má hpeni    ‘you were washing it’
            habitual      grú mú hpeni    ‘you used to wash it’
            ambulative    gá má tí hpeni  ‘you were washing it away/long ago’
Past                      gú hpeni        ‘you washed it’
Perfect                   xkú hpeni       ‘you’ve already washed it’
Pluperfect                xkí hpeni       ‘you’d already washed it’

Code: Select all

Irrealis
Present                   gi hpeni        ‘you’ll wash it’
Immediative               xta gi hpeni    ‘you’re about to wash it’
Ambulative                gi tí hpeni     ‘you’ll wash it away (here and there)’
Andative                  gri hpeni       ‘you’ll go wash it’
Past                      gi gi hpeni     ‘you’d wash it’
Perfect                   xki gi hpeni    ‘you’d have washed it’

Table 3. Local values

Code: Select all

Realis
Ambulative  andative      gá-r p.eni      ‘you wash it away (here and there)’
            cisloc.       gwá tí hpeni    ‘you’re washing it as you come’
Past        transloc.     gwú hpeni       ‘you washed it somewhere else’
Perfect     transloc.     xkwú hpeni      ‘you’ve already washed it somewhere else’
Pluperfect  transloc.     xkwú hpeni      ‘you’d already washed it somewhere else’

Code: Select all

Irr
Present     transloc.     gwu hpeni       ‘you’ll (go and) wash it somewhere else’
Past        transloc.     gwu gwu hpeni   ‘you’d (go and) wash it somewhere else’

Also, verbs in the Nilotic language Datooga have an opposition between "centrifugal" and "centripetal" orientation, which in some situations appears to correspond to a distinction between "here" vs. "there".

[chris_notts]

First of all: This is awesome!
(And, I believe, very realistic.)

Thanks!

Here's an example from Winnebago (Siouan):

Kook-ra	ha-nan-zhin-je-enan.
box-DEF	SUPRAESSIVE-by_foot-stand-AUX-DECL

'It is standing on the box.'

I'm not sure whether -nan- 'by foot' represents an instance of productive noun incorporation here, but even if it is not, it seems likely that it has originated from an incorporated body part noun. If that is true, the order of morphemes is clearly PREVERB - INCORP - ROOT. Semantically, the incorporated noun seems to be part of a complex verb stem here, with the preverb added later in the derivation.

This is a good example of a similar system, but I think my worry is about less transparent combinations. The example above is more or less compositional, i.e. on-foot-stand "stand on by foot". Do you know of Winnebago has any semantically non-transparent preverb-verb combinations? And if it does, what happens when you incorporate nouns into them?

(Source: Craig, Colette & Hale, Kenneth, Relational preverbs in some languages of the Americas, Language 64/2 (1988), p. 312-344; here: p. 317)

I wonder if I can get my hands on that... the most documented spatial preverb systems I've seen are for IE and Caucasian languages, and for languages influenced by them like Hungarian. They do seem to be an areal feature of Europe.

It seems that your language distinguishes only five vowel qualities, /i e a o ɨ/. In this configuration, I think marking an important distinction by vowel alternation should be prominent enough. (I've been having similar worries with regard to my own conlang Buruya Nzaysa, but there the situation is much less distinctive phonetically, often relying on a /e o/ vs. /ɛ ɔ/ difference in word-final unstressed open syllables.)

In unstressed syllables, yes, plus a few diphthongs. In stressed syllables length and creakiness are also distinguished, but the pre-stem is never stressed.

An /e o/ vs. /ɛ ɔ/ alternation is very suggestive of a +/-ATR distinction. Does Buruya Nzaysa show a similar contrast elsewhere, or am I barking up the wrong tree?

As for whether the "here" vs. "there" distinction would be stable semantically, I think you don't have to worry. The two possible pathways you listed are good options if you want to create a descendant language at some point, but neither of them is inevitable. "Here" vs. "there" is salient enough in many discourse situations that it could easily remain stable for a long time, especially if your language does not use separate "here" vs. "there" adverbs very often. That said, I think it's not unlikely that the reinforcement with ti- would eventually become obligatory given enough time, leaving the vowel alternation free to be reanalysed as marking something else. (Also, past or future actions involving a 1st person argument could easily start allowing distal marking too, for example to indicate that the speaker did or experienced something while being at a different location.)

Yes, you're right, it could also become associated with tense or mood. It could even be a redundant marker of direct vs inverse agreement in verbs where a deictic vowel is present.

Here's some (quite basic) information about Otomi:

Very interesting! It at least reassures me a bit that widespread marking of stationary deixis (not just direction of motion) in the verbal system is attested. It would be great to see these forms in use, e.g. in a transcription. There is a book, "The Function of Verb Prefixes in Southwestern Otomi", available on Amazon for £8, which appears to focus on deixis in the Otomi verbal system.

Also, verbs in the Nilotic language Datooga have an opposition between "centrifugal" and "centripetal" orientation, which in some situations appears to correspond to a distinction between "here" vs. "there".

Again, that document is really interesting because it describes the semantic extensions of these suffixes. And one question it raises is what should be the default. If I understand correctly, the most overtly marked case is the "elsewhere" case, which makes a lot of sense since the deictic centre is in some sense the spatial centre of interest. In the same way, in languages with proximate/obviative marking, it's always the obviative that's overtly marked. Perhaps instead of overtly marking action "here", I should be overtly marking action elsewhere in the case where the vowel alternation is not available.

[Frislander]

I've gotta be honest I haven't even tried to read your whole post because it is a gigantic wall of text, you perhaps could have done with splitting it u across several posts.

Here's an example from Winnebago (Siouan):

Kook-ra	ha-nan-zhin-je-enan.
box-DEF	SUPRAESSIVE-by_foot-stand-AUX-DECL

'It is standing on the box.'

I'm not sure whether -nan- 'by foot' represents an instance of productive noun incorporation here, but even if it is not, it seems likely that it has originated from an incorporated body part noun. If that is true, the order of morphemes is clearly PREVERB - INCORP - ROOT. Semantically, the incorporated noun seems to be part of a complex verb stem here, with the preverb added later in the derivation.

This is a good example of a similar system, but I think my worry is about less transparent combinations. The example above is more or less compositional, i.e. on-foot-stand "stand on by foot". Do you know of Winnebago has any semantically non-transparent preverb-verb combinations? And if it does, what happens when you incorporate nouns into them?

I wouldn't worry about it honestly, it seems natural to me. I'm pretty sure this happens in Iroquoian to some extent, where some of the pre-pronominal prefixes are mildly lexicalised and dependent on the verb and they precede both the pronominal prefixes and the incorporated nouns, but the best examples are probably to be found in Athabaskan (only the northern languages though, Apachean having lost NI), where the preverbs are almost entirely lexicalised and at the same time separated from the root by loads of morphology including person and TAM markers, and likely incorporated nouns as well.

[mèþru]

I've gotta be honest I haven't even tried to read your whole post because it is a gigantic wall of text, you perhaps could have done with splitting it u across several posts.

Same, and it is a shame because it looks really interesting
Can you make a TL;DR summary?

[chris_notts]

I've gotta be honest I haven't even tried to read your whole post because it is a gigantic wall of text, you perhaps could have done with splitting it u across several posts.

Same, and it is a shame because it looks really interesting
Can you make a TL;DR summary?

I was trying to present an overview of how the different 'slots' in the verbal word work together and interact with each other. I could try to summarise the individual areas, I guess... see below for part 1, which summarises how agreement works.

Verbal Template

[NEGATIVE-AGR1-PREVERB-AGR2]-[INCORP]-[ROOT-VOICE-ORIENTATION/ASSOCIATED.MOTION-MOOD-AGR3-PLURAL]

Agreement

The verb agrees with up to 2 core arguments. There are 4 agreement slots (AGR1, AGR2, AGR3, PLURAL), but (a) AGR2 and PLURAL are always controlled by the same argument, and (b) not all are used. The ones which are used depend on relative topicality of the arguments. In a transitive direct clause, the actor (A) outranks the patient (P), in an inverse clause the opposite is true.

Intransitive: AGR2 + PLURAL controlled by lone argument

Transitive Direct: AGR2 + PLURAL controlled by A, AGR3 controlled by P

Transitive Inverse: AGR1 controlled by A,
P controls either: AGR2+PLURAL if a speech-act-participant (SAP), otherwise AGR3

Anti-Agreement

If an argument which would control agreement is fronted to pre-verbal position for focus, then agreement with it is suppressed.

If the verb is intransitive, then a focus marker occurs in AGR1 and none of the other agreement slots are used.

If the focused argument of a transitive verb is the A, then the verb is formally inverse: a focus marker appears in AGR1, and the location of agreement with the P follows the same rules as for any other inverse clause

If the focused argument of a transitive verb is the P, then the verb inflects like an intransitive verb (there is no marking related to the P at all).

Impersonal

AGR1 can also contain an impersonal subject marker ("someone..."). If the verb is transitive, then P agreement follows the inverse pattern.

[mèþru]

Thanks!

[chris_notts]

I've been thinking about the deictic vowel alternation a bit more, how it originated, and also reworking it a bit. There are some things I didn't like about the previous version, including mutation of non-final vowels in preverbs.

My thinking about where the alternation might come from is as follows. The basic pattern is a +/- high vowel distinction within the pre-stem, with +high indicating action close to the deictic centre, and -high action further away. The vowel system, ignoring epenthetic vowels, is basically a box with higher /i, o/ and lower /e, a/, and the alternations I currently have are i ~ e, o ~ a, and e ~ a. I vaguely assumed that a patterning two ways would be explainable by the historical vowel mergers which produced the four vowel system in the first place, e.g. via an o - a merger followed by lowering of u.

I had originally thought of vowel coalescence or merger as a possible source, but that is problematic because most plausible patterns of productive vowel merger wouldn't produce the right patterns. In particular, in most languages where vowels merge +round tends to be preserved as a feature at least sometimes, which messes things up.

So now I'm thinking that the historical source of the pattern might be dissimilatory raising before a historical "here" morpheme *-a-. Productive vowel raising before a is common in Basque, where -a is the definite article. Most dialects have a-a -> e-a:

neska-a -> neskea "the girl"

Many also have raising of mid-vowels in this context, e-a -> ia, o-a -> ua. A few dialects then have deletion of -a when non-final, so the definite article is partly marked by height shifts in vowels.

If we assume a similar process for Sint, then we would have:

1. Vowel final preverbs whose final vowel isn't already high are raised by following *-a-, "here"
2. In this post-vowel context, *-a- is then deleted
3. Preverbs which already ended in a high vowel are modified to follow this pattern by analogy
4. Vowel-final preverbs now alternate to mark the here/there contrast

Funnily enough, an underlying low vowel morpheme has then turned into a +high vowel feature.

This then means we have:

1. For vowel final preverbs, a +/- high alternation to signal the here/there distinction
2. For consonant final preverbs and when no preverb is present, a linking vowel is inserted after them which shows the same +/- high pattern by analogy

It also means, in the case of multiple stacked preverbs, that preverbs have an underlying +/- high preference which is overridden when they occur finally in the chain, where *-a used to occur.

EDIT: or maybe -a survives in the case where there is no vowel hiatus. So that means that, if the final preverb doesn't end in a vowel, you have an -a vs zero/epenthetic ɨ contrast, with +/- low being the opposite way round.

[chris_notts]

A written up version of the new deictic vowel rules:

The Pre-Stem

The pre-verb serves two core, obligatory functions. The first is to host two of the three verbal agreement slots, AGR1 and AGR2, and the second is to indicate very fine grained distinctions of deixis, direction, path, and relative location of the action described by the verb. In addition to these functions, it also hosts optional markers of negation. Its structure is as shown in figure \ref{prestem-template.

AGR2-(V)
Minimal Preverb Structure Template

(NEG)-(AGR1)-(PREVERB)-(V)-AGR2
Prestem Structure if Additional Prefixes are Present

The form of the preverb depends on whether it only consists of an AGR2 agreement marker and a vowel, or whether other prefixes are present. The element V represents either the deictic vowel, if present, or otherwise an epenthetic /ı/ if necessary to create a legal syllable structure. See the following section for more detail about when a deictic vowel is present.

The pre-stem can contain up to two agreement markers, depending on the person and pragmatic status of the verb's arguments. A basic summary is that AGR2 agrees with the highest ranking argument, and then agreement with the other argument of a transitive verb occurs in AGR1 if AGR2 is the verb's patient.

The Deictic Vowel

Most verbs are marked for action at/towards the deictic centre (`here') or action elsewhere. This distinction is underlying marked by the presence or absence of proximate -a- in the pre-stem. In the case where no preverbs are present, -a- occurs at the end of the pre-stem, and alternates with zero / an epenthetic vowel. An epenthetic vowel will surface if necessary to create legal syllables, since Sint doesn't permit onset clusters:

naa
0-naa
3F-go
`she went'

kınaa
k-naa
3M-go
`he went'

kanaa
k-a-naa
3M-P-go
`he came'

This distinction is not marked if first person agreement is present in AGR2 and the verb is positive indicative:

nınaa
n-naa
1EXC-go
`I came/went'

If the prestem contains a preverb, negator, or AGR1 agreement marker which ends in a consonant or diphthong, then the deictic vowel occurs between these elements and the AGR2 marker. Compare:

isnaal
is-naa-l
NEG.OPT-go-OPT
`let her not go!'

isanaal
is-a-naa-l
NEG.OPT-P-go-OPT
`let her not come!'

The marking of deixis becomes more complex if -a- follows a prefix ending in a monophthongal vowel. First, there are some prefixes which drop their own vowel when followed by another vowel. v(i)- `not' is in this class, in this case producing a vi ~ va alternation. Of those which don't, proximate action is marked by raising of the final preverb vowel, and distal action by lowering. The possible vowel alternation patterns are listed in the following table.

Pattern & Underlying & Distal & Proximal
1,i,e,i
2,e,e,i
3,a,a,e
4,a,a,o
5,o,a,o
Preverb Vowel Alternations

As the table shows, for three of the four underlying vowels which can occur, only one alternation is possible. The only underlying vowel which show unpredictable behaviour is a, which may raise either to a front vowel or a back vowel to form the proximate. Note that in all cases either the distal or the proximal vowel is identical to the underlying vowel of the preverb.

This raises the question of how the underlying vowel can be determined. The underlying vowel of a preverb occurs in two contexts: firstly, when there are multiple preverbs within the prestem, and secondly when the preverb is followed by a first person agreement marker and the verb is positive indicative. The following examples both show the underlying form of ši `up':

šitlaknaa
ši-tla-k-naa
UP-ALONG.D-3M-go
`He went up along (there)'

šinaa
ši-(n)-naa
UP(.P)-(1EXC)-go
`I went up'
`She came up hither'

The example above is in fact ambiguous, due to degemination of identical consonants, between a reading where the first person -n is present and ši is in its default form, and a third person subject with proximal ši. With the distal form še, this ambiguity is not possible:

šenaa
še-Ø-naa
UP.D-3F-go
`she went up'

This means there are two alternations which are embedded in the middle of lexicalise preverb-verb pairs: AGR2 person agreement and deixis marking.