bradrn’s scratchpad

[Vardelm]

Good stuff! It feels really natural & very different from English. The vocab & syntax work well together to accomplish that.

I might have had questions about the word order, but since you provided the "non-obvious" warning, it actually is quite obvious when you look for it.

[bradrn]

Good stuff! It feels really natural & very different from English. The vocab & syntax work well together to accomplish that.

I’m glad you like it! That effect was exactly what I was going for.

(This is also the first conlang where I’ve actually put some thought into the lexicon, so that helps too.)

[bradrn]

The noun phrase

The Proto-‘Savanna’ noun phrase has the following summary linear structure, where all components are optional (though at least one must be present):

posessor prenominal root noun modifiers relative.clause article

A near-maximal example:

Muna

Muna

ngen

ALN

ya-galab

VIS.PL-dog

debu

black

daqat

big

quse

three

maqa

what

ŋii

3s.PL

fawetl

speak.IMPF

feqe

cry

walha

go

tasek

moon

thaŋ

DEF.SG

those three big black dogs of Muna’s which howl at the moon

In practice, however, most NPs have no more than two or three words.

Starting from the left edge, the first item in the NP is the possessor, expressed as another NP placed before the head noun. Proto-‘Savanna’ distinguishes inalienable from alienable possession. Inalienable possession is used for body parts and kinship terms, and is used in the nominalisation construction, while alienable possession is used elsewhere. The former is marked by simple juxtaposition of the two NPs, while the latter uses an additional particle ngen after the first NP:

Muna

Muna

meŋul

eye

tleŋ

DEF.PL

Muna’s eyes (inalienable)

Muna

Muna

ngen

ALN

galab

dog

thaŋ

DEF.SG

Muna’s dog (alienable)

The construction is slightly different when the possessor is a personal pronoun. An inalienable pronominal possessor must be supplied as a prefix: bemeŋul tleŋ ‘my eyes’, not *bal meŋul tleŋ. However, both alternatives are acceptable for alienable pronominal possessors: bengen galab thaŋ / bal ngen galab thaŋ ‘my dog’.

Demonstrative modifiers also appear immediately before the head, as prefixes: siqangat thaŋ ‘this tree’, yaqangat tleŋ ‘yonder trees’, and so on. The indefinite/interrogative demonstratives appear in the same position, albeit as independent words: naaŋ qangat thaŋ ‘some/any/which tree’. Demonstratives and inalienable possessors alternate, in that only one may appear at a time: *besitlungse thaŋ ‘my this eye’. This syntactic position is what I mentioned in the above linear structure as the ‘prenominal’ slot. Note that alienable possessors and demonstratives may freely co-occur: bengen sigalab thaŋ ‘this dog of mine’.

The head noun immediately follows any prenominal or possessor which might be present, though more often it is the first word in the NP. Note that the head may be any nominal: a proper or common noun, a full pronoun, a nominal demonstrative or an adjective, though only common nouns are completely unrestricted in their combinatorial possibilities. Though nouns take no inflectional morphology apart from any prenominal element, in the interests of convenience I mention a few derivational affixes here:

-ge (n→n) ’negation / non-’
-(q)am (n→n) ‘collection’
pa- (n→n) ‘diminutive’
-ye (n→n) ‘instrument associated with’
-(w)u (n→adj) ‘quality / -like’
ga- (n→adj) ‘containing’

e.g.: phingis ‘palm tree’, phingisam ‘palm forest’, paphingis ‘little palm tree’ phingisu ‘palmlike’, gaphingis ‘having palm trees’.

Next comes various nominal modifiers. Immediately following the head is at most one nominal modifier, which together with the head forms a noun compound. Proto-‘Savanna’ uses noun compounds sparingly, certainly much less than English does: sequences of more than two nouns are highly discouraged, and even two-noun compounds are dispreferred to possessive constructions. Nonetheless, they do exist.

Next come adjectival modifiers. Unlike nominal modifiers, an unlimited number of these may in theory be supplied, though in practice at most two or three are usually given. After the adjectives at most one numeral (listed previously) may be supplied. This position may also be occupied by a quantifier such as pawtiq ‘all’, mengu ‘many’, or qitham ‘some’.

After the nominal modifiers, relative clauses may appear — normally no more than one, since multiple relative clauses can feel a bit ‘clunky’. All relative clauses are restrictive; non-restricted relatives, as found in English and many other languages, are absent, clause juxtaposition being preferred for these cases. Relative clauses are formed by removing the relativised-upon noun from the embedded clause to leave a gap, then adding the corresponding indefinite demonstrative at the beginning of the clause. Less commonly, the noun within the relative clause may be altered to a pronoun:

Sasay

home

be

1s

lhiisi

perceive.PFV

meŋul

eye

qaŋeth

person

thaŋ.

DEF.SG

I saw the person at home.

Siqa

this

nii

FOC

qaŋeth

person

[naaqa

who

sasay

home

be

1s

lhiisi

perceive.PFV

meŋul

eye

(ŋay)]

(3s)

thaŋ.

DEF.SG

This is the person who I saw at home.

Headless relatives are also possible, and even fairly common. In fact, most of the elements of the NP may occur headless: Danaq ngen thaŋ ‘that of Danaq’s’, daqat ‘a big thing’, naaqa sasay be lhiisi meŋul thaŋ ‘who I saw at home’ are all valid NPs.

Finally, the rightmost element of the NP is the article. This marks definiteness and number, as follows:

	Singular	Plural
Definite	thaŋ	tleŋ
Indefinite	no article	tlii

As seen in previous examples, articles — and in particular the definite articles — see far more use in Proto-‘Savanna’ than in a language like English. At the root of this is the fact that articles are generally less restricted in their combinatorial possibilities in the former than in the latter. In Proto-‘Savanna’, articles are permitted to co-occur with nearly all elements of the noun phrase: possessors, demonstratives, and most nominal modifiers. Thus, whereas in English we generally do not form NPs like *‘the my home’, in Proto-‘Savanna’ we must specify whether we’re talking about bengen sasay ‘a home of mine’ or bengen sasay thaŋ ‘the home of mine’. In fact it goes further: the prenominal slot is intrinsically definite, and so actively requires a definite article. (Something like betlungse (intended: ‘an eye of mine’) is questionably grammatical, but considered incoherent semantically.) A similar situation arises with quantifiers, though being plural these would require an overt article anyway. And even demonstratives usually take the corresponding definite article, though this is not obligatory.

On the other hand, co-occurrence is more restricted in some cases. The most obvious case is personal pronouns and proper nouns, which may not take articles— or indeed any other modifiers at all. Less obviously, there is the case of relative clauses such as the following:

qaŋeth

person

[naaqa

who

lhiisi

perceive.PFV

meŋul

eye

kasip

star

thaŋ]

DEF.SG

(?thaŋ)

DEF.SG

a/the person/people who saw the star

In this situation, a definite article would give two identical articles in a row at the end of the NP. Speakers generally feel free to omit the article in such cases, leaving the noun ambiguous as to both definiteness and number (though this of course can be alleviated through other means, e.g. adding a demonstrative, or including another personal pronoun in the relative clause). Article omission is also possible even when the relative clause does not end with an article, particularly when it is long (although such constructions are infrequent in natural discourse).

[Vardelm]

Good stuff as usual, but I found this particularly tickled my fancy. The use of marked vs non-marked is simple & elegant.

The former is marked by simple juxtaposition of the two NPs, while the latter uses an additional particle ngen after the first NP:

[bradrn]

Good stuff as usual, but I found this particularly tickled my fancy. The use of marked vs non-marked is simple & elegant.

The former is marked by simple juxtaposition of the two NPs, while the latter uses an additional particle ngen after the first NP:

That one’s not original, I’m sorry to say — Ewe and Lango (amongst others) have the same system.

[Vardelm]

That one’s not original, I’m sorry to say — Ewe and Lango (amongst others) have the same system.

Well, it's new to me. I may have seen such before, but sometimes it takes presentation in a conlang to you sit up and say "oh, that's COOL!" For example, I didn't really appreciate Salishan languages until I saw Ngolu/Iliaqu by Imralu.

[bradrn]

That one’s not original, I’m sorry to say — Ewe and Lango (amongst others) have the same system.

Well, it's new to me. I may have seen such before, but sometimes it takes presentation in a conlang to you sit up and say "oh, that's COOL!" For example, I didn't really appreciate Salishan languages until I saw Ngolu/Iliaqu by Imralu.

Wait, Iliaqu is inspired by Salishan? I honestly never noticed.

[Vardelm]

Wait, Iliaqu is inspired by Salishan? I honestly never noticed.

Only in terms of comparison of having "nounless" syntax. It's a main inspiration for Jin.

[bradrn]

More clause structure

In my last post on clause structure there were a few topics I glossed over a bit, mostly relating to peripheral elements. To start off, I already mentioned the existence of the adjunct word class. These are placed at the periphery of the clause, to make sentences like:

Sasay

home

be

1s

lhiisi

perceive.PFV

meŋul

eye

qaŋeth

person

thaŋ.

DEF.SG

I saw the person at home.

However, ‘clausal adjunct’ is really a syntactic position, not just a word class. In fact, any definite NP may appear in this position, usually to specify a place or time:

Si-maqu thaŋ

this-place DEF.SG

be

1s

lhiisi

perceive.PFV

meŋul

eye

qaŋeth

person

thaŋ.

DEF.SG

I saw the person in this place.

Qalit naatl be mah kuwi thaŋ,

sun when 1s do.PFV disease DEF.SG

be

1s

lhiisi

perceive.PFV

meŋul

eye

qaŋeth

person

thaŋ.

DEF.SG

I saw the person on the day I fell ill.

Relative clauses headed by words like maqu ‘place’ or qalit ‘sun/day’ are particularly common, as in the last example. In fact, removing the head gives a construction almost identical to the corresponding English adverbial clauses, except for the article:

[Naatl

when

be

1s

mah

do.PFV

kuwi

disease

thaŋ],

DEF.SG

be

1s

lhiisi

perceive.PFV

meŋul

eye

qaŋeth

person

thaŋ.

DEF.SG

[When I fell ill], I saw the person.

Superficially similar to adjuncts are topics. As is cross-linguistically usual, these occur before the relevant clause, with a pause following the topic; the topic is referenced by a pronoun within the clause. (The operation is more technically known as ‘left-dislocation’.) Often, this is all that is necessary, e.g.:

[Ŋa-pa-welus

that-DIM-man

naaqa

who

waq

do.IMPF

fetlhalh

IDEO.horrible

thaŋ]_TOP,

DEF.SG,

si-qalit

this-sun

thaŋ

DEF.SG

qi

3s

yusaye

come.PFV

be-ngen

1s-POSS

qalit

house

thaŋ

DEF.SG

waalhi.

go.PFV

[That awful little man], he stopped by my house today.

To which you could reply, say, Be daa siwe lhibuq siqa thaŋ ‘ah yes, I know the one’; since said horrible man has now been introduced as a discourse topic, for that point on there is no need for explicit mention of him using anything heavier than a pronoun.

Of course, left-dislocation is far from the only way to introduce a topic: there are numerous periphrastic constructions as well. Many of these involve verbs, along with a dummy subject:

Qi

3s

tlaquf

accompany

welus

man

thaŋ,

DEF.SG

qi

3s

yusaye

come.PFV

sasay

house

thaŋ.

DEF.SG

lit. With the man, he came to my house

Qi

3s

siwe

have

welus

man

thaŋ,

DEF.SG

qi

3s

yusaye

come.PFV

sasay

house

thaŋ.

DEF.SG

lit. Having the man, he came to my house.

(Compare English ‘as for X…’, ‘with regards to X’…, and so forth.)

The use of adjuncts is also reasonably common. (As noted above, left-dislocated topics seem rather similar to adjuncts anyway.) Relative clauses can be a bit bulky for this purpose, so possessive phrases — again, often headless — are generally used instead:

Welus

man

thaŋ

DEF.SG

ngen

ALN

(maqu)

place

thaŋ,

DEF.SG

qi

3s

yusaye

come.PFV

sasay

house

thaŋ.

DEF.SG

lit. The place of the man, he came to my house.

[Ares Land]

Looks neat. One bit that isn't terribly clear is why you mention an adjunct word class before noting that there is, in fact, no such word class.

I'm also not sure I understand how that last example works?

[bradrn]

Looks neat. One bit that isn't terribly clear is why you mention an adjunct word class before noting that there is, in fact, no such word class.

Sorry, I should have been clearer. There does exist a separate adjunct word class: it is defined by being able to go in adjunct position, without necessarily requiring a definite article. Furthermore, there exist words which can be used as adjuncts but not nouns, like gatlaq ‘certainly’ or qanta ‘now’. It’s basically equivalent to adverbs in other languages (except I use ‘adverb’ for a different word class, which actually modifies verbs).

I'm also not sure I understand how that last example works?

What part of it do you not understand? I honestly can’t see how it could be misinterpreted.

[Ares Land]

What part of it do you not understand? I honestly can’t see how it could be misinterpreted.

Oh, sorry. In fact, I don't really understand what the sentence means -- I think the literal translation might be too literal for me

[bradrn]

What part of it do you not understand? I honestly can’t see how it could be misinterpreted.

Oh, sorry. In fact, I don't really understand what the sentence means -- I think the literal translation might be too literal for me

Oh, it just means exactly the same as all the other sample sentences: As for the man, he came to my house. ‘The man’ is the topic.

[bradrn]

Verbal morphology and aspect

We turn now to one of the more unique parts of Proto-‘Savanna’ grammar: namely, its verbal aspect system. In outline, aspect is expressed using two separate yet interlocking subsystems of verbal morphology. A system of inflectional aspect signals perfectivity through stem change, while a separate system of derivational aspect can be used for more complex variation. Together, they allow the speaker to indicate subtle distinctions in temporal setting and focus, something which more than makes up for the deficiencies of the small verb system and lack of conjunctions. A good understanding of both verbal aspect systems is vital for translating anything to or from Proto-‘Savanna’, or even just using the language idiomatically at all.

Inflectional aspect

First, let’s get one thing out of the way: calling this subsystem ‘inflectional’ is somewhat of a misnomer, especially if it is considered in opposition to the ‘derivational’ subsystem, which is also something of a misnomer. I call this one ‘inflectional’ partly due to its phonological form, but mostly because of the aspect concord which exists in serial verb constructions for the inflectional subsystem, but not for the derivational subsystem. However, with regards to semantics the inflectional subsystem isn’t necessarily any less ‘derivational’ than the derivational one, if that makes sense. (Although then again, with regards to discourse there is a sense in which this is indeed true. See below.)

(So why call them ‘inflectional’ and ‘derivational’ at all? Because I needed to call it something, that’s why!)

The inflectional aspect system basically indicates perfectivity. That is, it signals the a distinction between imperfective and perfective aspect. In terms of form, this is accomplished through root alternation (as already mentioned): all verbs have separate imperfective and perfective roots, neither of which can be predicted from the other. However, there are similarities between verbs, and we can identify a number of different patterns through which the perfective and imperfective roots are related:

1. Identity: paatli~paatli ‘give’, qaathan~qaathan ‘fall’.
2. Addition of a suffix -i to the PFV form: fawetl~fawetli ‘speak’ fas~fasi ‘stand’.
3. Final vowel mutation: siwe~siwi ‘have’, lhiise~lhiisi ‘perceive’.
4. Irregular change at the right edge: yusaa~yusaye ‘come’, qisa~qisni ‘cut’, phage~phaŋi ‘break’, gadiq~gadye ‘absorb’
5. Synchronic suppletion: waq~mah ‘do’, tlaquf~lalef ‘come’.

Of these, most common are types 3 and 4. Type 2 is somewhat less common, and types 1 and 5 are very rare.

There are also a handful of defective verbs with only one root — most prominently, ndiyam ‘become’ exists only in perfective form. These are restricted to only one aspect, and cannot be used in the other aspect at all.

The function of the imperfective/perfective distinction is perhaps more interesting. Generally speaking, we can say that the imperfective aspect is used for events currently in progress, with some duration and/or internal structure, whereas the perfective aspect is used for events in the past, which have been completed and have resulted in some change of state. (Though notably, the latter does not require telicity or boundedness, i.e. it does not have ‘Slavic-style’ aspect: see Ephraim’s posts here and especially here for more.) In practice, however, both aspects are more broadly applicable. The imperfective aspect has a wider range of use than the perfective, and can be considered the functionally unmarked aspect: it is commonly used for stative or habitual events, hypotheticals, imperatives, conditionals, and other speech acts which are not straightforwardly perfective.

When we start looking at individual verbs, however, this neat picture starts to fragment a bit. In particular, aspectual variation highlights a clear distinction between active and stative subsets of verbs. With the former, the imperfective/perfective distinction basically occurs with the semantics outlined above. However, when a stative verb is put into the perfective aspect, there is an additional implication of inchoativity: that the subject is coming into the state. Thus, for instance, while siwe means ‘have’, the corresponding perfective root siwi means ‘take’, i.e. coming into the state of having something; similarly, while lhiise is to ‘perceive’ something, if you lhiisi it, you are instead beginning to see it. By contrast, for active verbs like waq~mah ‘do’, gadiq~gadye ‘absorb’ or qefaa~qefay ‘rise’, the single-word English gloss adequately covers both the imperfective and the perfective meaning.

Derivational aspect

We now come to the other aspectual subsystem, namely that of derivational aspect. As with ‘inflectional aspect’, the name is somewhat misleading. Certainly it does have some application to verbal derivation: compare e.g. gadiq~gadye ‘absorb/ingest’ with gadiqtsi ‘absorb in bits’, gadyeŋu ‘absorb completely’, gadiqŋu ‘have as a part’. But this is not always the case: in different contexts, these suffixes can cause little to no semantic change in the verb. In particular, they are heavily used to connect clauses in natural discourse (see below). This subsystem also has a number of formal properties more usually associated with inflectional than with derivational properties: its affixes are mutually exclusive, and usually they can appear with all verbs, or at least all verbs of a certain subclass.

In general, the derivational aspect suffixes act to alter the temporal focus of the clause — that is, which part of the event do we want to highlight? In some cases, it can also indicate the distribution of time — that is, when was the event being performed and when was there no event? Formally, the system consists of 5–8 suffixes, depending on how you want to count them:

Imperfective	Perfective
-tsi ‘iterative’ -ke ‘conative’ -ŋu ‘stative’ -li ‘continuative’	-tsi ‘iterative’ -ke ‘diminutive’ -ŋu ‘telic’ -me ‘ingressive’

As indicated by the table, the derivational aspect system does not operate entirely independently from the inflectional aspect system — but neither are they tied together particularly closely. While some suffixes may be used only with imperfective roots, or with perfective ones, others are available with both imperfective and perfective verb roots, with varying degrees of semantic difference between the two.

We will start by considering the iterative -tsi. At a high level, this aspect indicates that the action in question took place repeatedly in some way. While the iterative may be used with both kinds of verb roots, it has somewhat different connotations depending on what kind of root it is applied to. With a perfective root, the resulting word can only refer to repetition on a single occasion:

qisni ‘cut’ → qisnitsi ‘cut up’
phaŋi ‘break’ → phaŋitsi ‘break into many pieces, crumble’
siwi ‘have/take’ → siwitsi ‘take and take’
fasi ‘stand [up]’ → fasitsi ‘keep on getting up’

However, with an imperfective root, the iterative is more ambiguous, and can also refer to repetition over numerous different occasions:

qisa ‘cut’ → qisatsi ‘cut many times, chop’ or ‘cut often’
qaathan ‘fall’ → qaathantsi ‘fall several times’'
phage ‘break’ → phagetsi ‘break apart bit by bit’

Some verbs, especially stative ones, are rarely or never used in the imperfective iterative, purely due to semantic oddness: siwitsi ?‘have on many different occasions’, fastsi ?‘stand on many different occasions’. (Their perfective iterative forms are perfectly acceptable, though, as shown above.) Even some active verbs in the perfective iterative are fairly odd: segegtsi ?‘die and die’.

The suffix -ke is in some ways almost the opposite of the iterative. With perfective roots, it has a general diminutive sense: it reduces the intensity of the action. It is particularly often used as a semelfactive, to indicate an action done once only:

qisni ‘cut’ → qisnike ‘cut once’
phaŋi ‘break’ → phaŋike ‘crack open’
mah haŋuhaŋu ‘blow’ → mahke haŋuhaŋu ‘blow one gust’

But it can just as often appear with meanings like ‘just’ or ‘merely’, or as a general marker of reduced intensity. This is especially prominent with verbs which are inherently durative:

fasi ‘stand up’ → fasike ‘just stand up’
fawetli ‘speak’ → fawetlike ‘speak a bit’
gadye ‘absorb, ingest’ → gadyeke ‘eat a little’

-ke can also be used with imperfective roots, but with somewhat different semantics. In this case, it takes on a conative role, expressing actions which have not been completed:

qisa ‘cut’ → qisake ‘almost cut through’
segek ‘die’ → segeke ‘almost die’ (note the morphophonology!)
lhiise meŋul ‘see’ → lhiiseke meŋul ‘almost see’

[I’m not actually sure if ‘conative’ is the right word for this, but it doesn’t seem to have been used for anything else…]

The conative is also used with clause combining, to denote an action which occurred just as another action was finishing:

Be

1s

gadiq-ke

absorb-CON

tlaame,

food

qi

3s

lhiise

perceive

meŋul

eye

bal.

1s

I had almost finished eating when he saw me.

The last suffix available to both perfective and imperfective roots is -ŋu. As with -ke, its perfective and imperfective usages have rather different semantics. With perfective verbs, -ŋu has a basically telic meaning: it emphasises that an action has an ending point, which is achieved at some point. This can lead to rather subtle changes in meaning, many of which have no straightforward English equivalent. Some examples which do:

fasi ‘stand’ → fasiŋu ‘stand up’
waalhi ‘go’ → waalhiŋu ‘reach, arrive’
ndisi ‘place’ → ndisiŋu ‘set down’
mah kuwi ‘become ill’ → mahŋu kuwi ‘fall ill’

However, perfective -ŋu has a wider range of interpretations. It can also indicate that the action is totally completed, that the object of the clause is especially affected, or occasionally that the agent is particularly agentive:

segeg ‘die’ → segegŋu ‘die stone dead’
ŋandi ‘deform’ → ŋandiŋu ‘pull out, twist around, etc.’
siwi ‘have’ → siwiŋu ‘grab all’
siwi lhibuq ‘know’ → siwiŋu lhibuq ‘decide’
lhiisi meŋul ‘see’ → lhiisiŋu meŋul ‘look’

As an extension of this, the telic can sometimes take on a perfect meaning, or even that of a simple past. The essential commonality is that it acts to focus on the ending point of an event.

Perfective -ŋu is also heavily used in clause combining, where it indicates temporal sequencing:

Be

1s

ŋandi

pull.PFV

buthe

honey

thaŋ

DEF.SG

yusaye-ŋu,

come.PFV-TEL

be

1s

gadiq

absorb.IMPF

tlaame

food

thaŋ.

DEF.SG

I pulled out the honey, then ate it.

In the first clause here, note that the suffix could have just as well gone on the first verb ŋandi, with little to no semantic change; in such cases it is most common to place -ŋu on the last verb. We can also compare this to the corresponding sentence with imperfective -ke:

Be

1s

ŋandu

pull.IMPF

buthe

honey

thaŋ

DEF.SG

yusaa-ke,

come.IMPF-CON

be

1s

gadiq

absorb.IMPF

tlaame

food

thaŋ

DEF.SG

I pulled out the honey, and ate it.

Both constructions represent temporal sequencing, but with different connotations: -ŋu implies that the first action was completed, while -ke does not.

As already mentioned, -ŋu can also be used with imperfective roots. Here its distribution is more restricted: imperfective -ŋu can be used only with active verbs. With these it acts as a stative (or alternately resultative) aspect, denoting the state which results after performing the action:

gadiq ’absorb, ingest’ → gadiqŋu ‘have as a part’
qisa ‘cut’ → qisaŋu ‘be in shreds’
qefaa ‘rise’ → qefaaŋu ‘be elevated’

In addition to the three suffixes just covered, there are two more derivational aspects which are perfectivity-specific. Firstly, the continuative -li is restricted to imperfective roots. This aspect has the effect of focussing on the event itself, while de-emphasising the event boundaries (beginning and end). The effect with standalone verbs is difficult to translate in English, but constructions like ‘keep on V’ or ‘V for a long time’ convey some of the meaning. In natural discourse, the continuative is however more often used to indicate simultaneity in clause combining:

Baa

1p

fawetl-li,

speak.IMPF-CONT

naaqa

someone

yusaye-ŋu

come.PFV-TEL

waalhi

go

bal.

1s

As we were talking, someone came up to me.

(Incidentally, this example has another instance of the telic aspect, again to emphasise the completedness of the action.)

Finally, the ingressive -me is restricted to perfective roots; furthermore it is used only with active verbs. Simply put, the ingressive focusses on the starting point of an action, giving an inchoative sense for the resulting verb:

Qi

3s

yusaa-li

come.IMPF-CONT

be

1s

waalhi-me

go.PFV-INGR

tsagif.

running

As he approached I ran away.

As with some of the other aspectual suffixes, the ingressive has additional use in clause combining, in which it indicates that one event occurred before or near the start of another one. This is incidentally illustrated in the above example: the man approaching is coincident with my running away.

Aspect in clause-combining

For the most part, the uses of Proto-‘Savanna’ aspect in grammar and discourse are quite similarly to those of aspect in other languages. In this way both inflectional and derivational aspects are used to focus on different parts of the temporal setting of an event: the speaker can ‘zoom in’ on the beginning, middle or end of an event, or look at it as either a whole or a sequence of shorter events. Similarly aspect is used for derivation, as it allows an event to be constructed by slicing up other events in different ways. However, Proto-‘Savanna’ also uses aspect extensively in a third area: namely, clause combining. Of course all languages have interactions between aspect and clause combining in some ways, for instance preferring perfective aspect for event sequencing, but Proto-‘Savanna’ is unusual in the extent to which this has become systematised.

The key idea to note here is that Proto-‘Savanna’ has a strong preference for parataxis. It has few conjunctions, and few particles with conjunctive or disjunctive semantics; instead, clauses are simply placed next to each other. (This behaviour is cross-linguistically very common.) Often, aspect is the only indicator of the temporal relationship between clauses, and the derivational aspect system in particular has been co-opted to signal this relationship rather than temporal focus. As previously mentioned, this is especially common with the telic, continuative and ingressive aspects, but other aspects too are used in clause combining. Other usages may be (incompletely) listed:

• The iterative aspects can be used to denote two actions interspersed with each other, especially in SVCs
• The conative aspect may express an event interrupted by another
• The resultative aspect naturally gives the result of another action

And so on.

In natural discourse this tends to give some ‘separation of powers’ between the two aspectual subsystems. Inflectional aspect is used to give an outline of the temporal setting, as in other languages: event sequencing in the perfective, simultaneity in the imperfective, and so on. Derivational aspect then interacts with that to introduce extra expressivity: disambiguating temporal relationships, adjusting the nuances of events, altering the meanings of various verbs, etc. In this regard, the derivational aspect subsystem may be seen as having a more ‘derivational’ than ‘inflectional’ role role in discourse. Understanding this is essential to both understanding and speaking idiomatic Proto-‘Savanna’.

(I would have liked to do an analysed text to illustrate the principle, but it would take too long… maybe some other time.)

Note that it is important to distinguish between clause combining and serial verb constructions! Although some aspectual combinations may have similar semantics in both areas (as with the iterative), SVCs are more tightly bound and tend to have their own distinctive temporal interpretations. In particular, SVCs have aspect concord, requiring all their verbs to agree in inflectional aspect; this in turn reduces their availability to variations in derivational aspect. The latter in SVCs is most commonly used for true semantic derivation rather than anything else.

[bradrn]

Editorial note: the first post now has a table of contents.

[Ares Land]

A very neat approach to aspect. I think you'll have to work on that sample text it would make it a lot easier to understand.

[bradrn]

A very neat approach to aspect. I think you'll have to work on that sample text it would make it a lot easier to understand.

Sorry, which sample text?

[Ares Land]

That one:

(I would have liked to do an analysed text to illustrate the principle, but it would take too long… maybe some other time.)

[bradrn]

That one:

(I would have liked to do an analysed text to illustrate the principle, but it would take too long… maybe some other time.)

Ah yes, that one! Perhaps I can work on that once I’m done with the relay text. (Or perhaps not… I have other things on too.)

[bradrn]

More verbal morphology, and nominalisation

Most verbal morphology is part of the aspectual system, as outlined above. However, there are a handful of derivational elements which are worthy of mention.

The first is the passive participle suffix -(q)i. (Note that the initial glottal stop is omitted after a consonant, and the verb must be in the imperfective root. The latter is the case with the below-described elements too.) This converts a verb to an adjective describing the state after an event has occurred: paatli ‘help’ → qaŋeth paatliqi ‘helped person’, segek ‘die’ → qumukh segeki ‘dead body’, etc. This is the closest you can get to a passive in Proto-‘Savanna’, since it lets you say things like qaŋeth thaŋ nii paatliqi ‘the person was helped’ — but this isn’t really a true passive, since you can’t add in the agent.

The other two are the action and agent nominalising elements qan and naaŋ respectively. These are placed before the verb to yield the corresponding noun: qaathan danit ‘sit’, qan qaathan danit ‘sitting’, naaŋ qaathan danit ‘sitter’, and so forth. Patients may be specified the same way as usual (i.e. unmarked after the verb), while for action nominalisation, the agent may be specified as an inalienable possessor: be di lhiisi tluŋse qi qan fawetl feqe thaŋ ‘I could hear his crying’. The status of the nominalisers themselves is a bit difficult: they could be prefixes, but could just as well be independent particles, especially since they satisfy the minimal word requirements. Orthographically, I will write both them and the similarly ambiguous demonstrative modifiers as separate words.