bradrn’s scratchpad

[chris_notts]

I won’t be focussing much on phonology here, but it is worth noting that the preverb as a unit is somewhat distinct from the verb complex proper. For instance, hiatus avoidance does not apply at the end of a preverb: e.g. to-asan ‘it fell’. Similarly, when they contain more than a single subject marker, preverbs tend to get final stress independently of the main verb. On the other hand, processes like nasal assimilation (on which I’ll say more later) do apply across the boundary.

As you probably remember, Ch'ubmin (reworking in progress as Qummin) is also like this, and it definitely happens in the wild (see e.g. Marind). One thing I have struggled with a bit is the orthographic representation when some assimilatory processes cross the boundary. Some linguist and even orthographies mark the intermediate nature of such "words" by a dash, e.g. preverbs in the Ojibwe orthography, some write them as a separate words, some as one word. I believe that the one word solution is the only one which might regularly show the assimilatory changes. It looks like you chose to write the preverb and the following verb as one - was that specifically to reflect the pronunciation differences due to boundary effects in the orthography?

[bradrn]

It looks like you chose to write the preverb and the following verb as one - was that specifically to reflect the pronunciation differences due to boundary effects in the orthography?

Wordhood is a little bit hard to define for this language. I might write something more about it later. (Or you could try coming up with a definition yourself! I’d be interested to see how other people analyse this.)

But yes, it does considerably ease the writing of these kinds of effects.

[chris_notts]

It looks like you chose to write the preverb and the following verb as one - was that specifically to reflect the pronunciation differences due to boundary effects in the orthography?

Wordhood is a little bit hard to define for this language. I might write something more about it later. (Or you could try coming up with a definition yourself! I’d be interested to see how other people analyse this.)

But yes, it does considerably ease the writing of these kinds of effects.

In Qummin I've been vaguely following an approach I've seen elsewhere of defining two phonological domains, the word and the phrase, and saying that consonant lenition and assimilation are phrasal phenomena, while stress, apocope and syncope are word level phenomena. And outside of the phrase, assimilation and other changes are resisted. But whatever you call them, it seems clear that there need to be two phonological levels of integration to get the correct output.

[bradrn]

Verb stems and lexical organisation

[UPDATE: some of the verb roots listed here have been changed since this post, so don’t take the forms below as being authoritative. But the aspect markers and the composition of verb stems remain unchanged.]

After about six months of staring at paradigm tables, I believe I’ve found a verb system I’m actually happy with. It’s an interesting one… as far as I’m aware, no natlang has a verb system quite like it (though some Afroasiatic languages get close, notably Semitic and Berber).

(It also, unfortunately, mildly invalidates some of my earlier example sentences. I did say it might happen, though!)

In brief: Eŋes forms verb stems by placing verb roots in one of five aspects. Roots are in general non-concatenative and templating, though a large proportion of simple stems are purely suffixing. Though most roots cannot appear in all aspects, and their derived stems have been lexicalised, the system retains a high degree of transparency.

(‘Aspect’ may admittedly be a bad choice of term here. Really it’s more like Aktionsart: the intrinsic lexical aspect of an action. But I doubt it will cause confusion.)

I’ll start by setting out the forms of verb stems and their aspectual possibilities, for simple and compound stems. I’ll then discuss the functions of the five aspects in stem derivation. Finally, I’ll talk a bit about the more restricted system of auxiliaries used to derive predicative adjectives and causatives.

Simple roots

I’ll start with a summary table:

	√f-	√as-	√rw-	√wl-	√ndo-	√wgi-	√f-y-
Iterative	fes ‘~crumble’	ases ‘~tumble’	rwes ‘accumulate’		ndos ‘circle’		feys ‘climb’
Atelic	fe ‘~shed’	ase ‘stumble’		wle ‘go’	ndok ‘bypass’
Stative	foŋ ‘be ~broken’	asoŋ ‘be low’	rwoŋ ‘have’	wloŋ ‘be gone’	ndoŋ ‘passed’	wgiŋ ‘be’	foyŋ ‘be high’
Intensive	fen ‘~grind’			wlen ‘travel’	ndon ‘pass’
Punctual	fem ‘break’	asem ‘~trip’	rwem ‘take’	wlem ‘depart’		wgim ‘become’	feym ‘~lift’

Notational conventions:

• I’m using the tilde ‘~’ to highlight when the English gloss is a particularly poor approximation to the scope of the Eŋes verb (but I’ll talk more about the semantics later)
• The sign ‘√’ is used to indicate roots; I believe this is mostly a Hebraic convention, but it seemed appropriate here too

This should give a good impression of the overall ‘flavour’ of this system. Not every root exists in every aspect, and the resulting stems exist as separate lexemes, but they’re clearly and systematically related in both form and function.

The particular roots listed above are examples of what I’m calling ‘simple’ roots. Their distinguishing characteristic is that the aspect marker appears primarily as a suffix, without the root being continuous — and when it is discontinuous, the infixed portion is a single vowel. Additionally, they vary for aspect in a much simpler way than do compound roots.

Simple roots take one of three paradigms, all of which are represented in the table. Most simple roots end in a consonant, and take a -VC or -V suffix for aspect. Other roots end in a vowel, taking a consonantal suffix instead. And a few are discontinuous, requiring an intercalated vowel as well as a suffix. These three paradigms are closely related, as follows:

Aspect	Consonant-final	Vowel-final	Intercalating
Iterative	-es	-s	-e-s
Atelic	-e	-∅ / -k	-e-
Stative	-oŋ	-ŋ	-o-ŋ
Intensive	-en	-n	-e-n
Punctual	-em	-m	-e-m

Which root takes which paradigm is largely predictable, except for a handful of consonant-final stems which take the ostentably ‘vowel-final’ paradigm. Additionally, the vowel-final paradigm itself has a slight irregularity, namely that a few roots take -k in the atelic rather than a null suffix (as with ndok above, √ndo+k).


Complex roots

These deserve another summary table:

	√wl-naŋ	√rw-lon	√pasi-sef	√fw-s-feʔ	√w-rfin	√asan-rnir	√lis-mŋun
Iterative	wlesnaŋ ‘walk’		pasissef ‘flicker’	fwessfeʔ ‘cry’	wserfin ‘build’		lissimŋun ‘~search’
Atelic	wlenaŋ ‘stroll’			fwesfeʔ ‘~sob’	warfin ‘imagine’	asanarnir ‘squat’
Stative		rwoŋlon ‘facing’				asanŋornir ‘sitting’	lisŋumŋun ‘see’
Intensive	wlennaŋ ‘stride’	rwenlon ‘face’	pasinsef ‘shine’	fwesnfeʔ ‘wail’	wnerfin ‘make’		lisnimŋun ‘look’
Punctual			pasimsef ‘ignite’			asanmarnir ‘sit’

It should be clear now why I call these ‘complex roots’: they have at least two parts, with the aspect marker going between. In general it’s difficult to identify subparts of a complex stem with their own meanings, though certainly some components get repeated (e.g. √wl-naŋ ‘related to walking’ vs earlier √wl- ‘related to going’).

The alternations of complex roots are also more intricate than that of simple roots. They undergo three paradigms identical to those of simple roots, plus three extras:

• The ‘unmetathesised’ paradigm is identical to that of consonant-final simple stems.
• The ‘vowel-final’ paradigm is identical to that of vowel-final simple stems.
• The ‘intercalating’ paradigm is identical to that of intercalating simple stems.
• The three ‘metathesised’ paradigms have form -CV- (rather than the V+C form of the other paradigms seen so far). They are broadly similar to each other, but can be distinguished based on the vowel of the atelic and punctual aspects.

The three metathesised paradigms are as follows, with the unmetathesised paradigm for comparison:

Aspect	Unmetathesised	Metathesised-/a/	Metathesised-/e/	Metathesised-/o/
Iterative	-es-	-se-	-si-	-si-
Atelic	-e-	-a-	-e-	-o-
Stative	-oŋ-	-ŋo-	-ŋu-	-ŋu-
Intensive	-en-	-ne-	-ni-	-ni-
Punctual	-em-	-ma-	-me-	-mo-

Five out of these six paradigms are seen in the earlier summary table (only metathesised-/o/ is missing).

In general, the vowel-initial paradigm is found where the first component of the root ends in a vowel, the intercalating paradigm is used for tripartite roots, and one of the three metathesised components is used when the second component begins with two consonants. For all other roots, either the unmetathesised or a metathesised paradigm is used, though the choice is unpredictable.


Semantics

Of course, all these tables have probably already given a reasonable impression of the semantics associated with each of these stems. Nonetheless, it’s useful to properly review the function of each.

(The names are, unfortunately, a little inadequate. Nonetheless, I can’t think of better ones.)

Before giving details, one very important thing to note: none of these aspects change transitivity. For instance, both fem ‘break’ and foŋ ‘be broken’ are intransitive, and both rwoŋlon ‘facing’ and rwenlon ‘face’ are transitive.

The iterative aspect is probably the most straightforward: it refers to actions which, in some way, are associated with some form of iteration. This can be event-internal (e.g. wlesnaŋ ‘walk’, pasisef ‘flicker’), or less commonly refer to multiple situations (e.g. ndos ‘circle’). Notably, iterative verbs never seem to have a distributive interpretation — that is to say, the iteration is primarily in time, rather than in space. It also seems that iterative verbs are consistently atelic, never having an intrinsic ending point.

All this seems to make the iterative verbs rather difficult to gloss, at least in my experience. Thus, e.g. ases could best be defined as something like, ‘fall in a series of short steps’. I’ve glossed it ‘tumble’, but that seems inadequate: a slinky going down steps would be ases-ing, but ‘tumble’ isn’t really applicable there. Similarly, fes is really ‘break off in pieces’, having wider scope than the gloss of ‘crumble’ which I gave.

The atelic aspect is slightly more difficult. Often it has a specifically conative meaning: that is, not only does the action have no natural endpoint, but any endpoint it might have is actually averted in some way. Thus, if you started to fall but checked yourself, you have ase ‘stumbled’; or if you passed a place and did not visit, you have ndok ‘bypassed’ it. (As you may guess, this has been another difficult category to gloss.)

On occasion, it alternately takes a semelfactive meaning: that is, denoting an event which is punctual yet atelic. Often, this specifically denotes a single instance of the corresponding iterative action: thus, if you fwessfeʔ ‘cry’ a little bit, you give a fwesfeʔ ‘sob’. (Though note that this verb is narrower in scope than the English gloss, since you can ‘sob’ for a long time, but fwesfeʔ is only punctual.)

The stative aspect, as its name suggests, refers specifically to states. Very often, this state is the result of the corresponding punctual verb: fem ‘break’ yields foŋ ‘broken’, wlem ‘depart’ yields wloŋ ‘broken’, and so on. But just as often, it is some other state related to the meaning of the root: for instance, lisŋumŋun ‘see’ is not obviously derived from any of the other stems in √lis-mŋun.

The intensive aspect has probably the broadest usage of any of the aspects. I’ve named this form ‘intensive’ because sometimes it does seem to denote an intensive form of one of the other stems: e.g. fen ‘grind’ vs fes ‘crumble’, lisnimŋum ‘look’ vs lisŋumŋun ‘see’, fwesnfeʔ ‘wail’ vs fwessfeʔ ‘cry’. But, more generally, it seems to denote all kinds of accomplishments (durative telic events); as in ndon ‘pass’, or rwenlon ‘face’. Curiously, it is also the only non-stative verb form which can denote states, as long as they are intensive in meaning — I gave no examples of this in the summary tables above, but e.g. tuʔn ‘be friends’ is an example (cf. tuʔŋ ‘accompany’).

Finally, the punctual aspect is used with punctual events. In particular, it is used for telic punctual actions (achievements), since as already mentioned atelic punctual actions get the atelic aspect. Often, it has a specifically inchoative meaning: as in feym ‘be lifted, start rising’ vs foyŋ ‘be high’, or pasimsef ‘ignite’ vs pasinsef ‘shine’. Or, of course, wgim ‘become’.


The verbal auxiliary

[UPDATE: the use of the auxiliary with verbs has expanded in meaning somewhat since this was written, see viewtopic.php?p=81241#p81241.]

This accounts for the verbs proper. But it is important to note that adjectives can also be used as verbs, by prefixing them with what I call the ‘auxiliary’. This auxiliary cannot take all of the aspects mentioned above, but is restricted to only two: the stative and punctual (which as mentioned takes an inchoative meaning).

The form of the auxiliary is quite different to the suffixes seen earlier. There are three paradigms:

• Vowel-initial adjectives are prefixed with w- in the stative and m- in the punctual: e.g. w-enar ‘be hot’, m-enar ‘become hot’.
• Adjectives beginning with two consonants, and some others, are prefixed with wa- in the stative and ma- in the punctual: e.g. wa-snar ‘be wide’ ma-snar ‘widen’.
• Most consonant-initial adjectives get w- in the stative and ma- in the punctual: e.g. w-raw ‘be good’, ma-raw ‘become good’.

There is another use of the verbal auxiliary: prefixed to any verb stem, it derives a causative form. This prefix cannot vary for aspect at all (although the stem may). Curiously, this uses the stative form of the auxiliary, rather than the punctual: fem ‘break (intransitive)’ ⇒ wafem ‘break (transitive)’, asoŋ ‘be low’ ⇒ wasoŋ ‘lower’. As with the adjectival use, vowel-initial stems always receive w- and two-consonant–initial stems always receive wa-, while other consonant-initial stems unpredictably get one or the other.

(On reflection, for complex stems the choice of auxiliary may well be correlated with whether it takes a metathesised or unmetathesised paradigm. It’s something I should investigate, but haven’t yet.)

Of course, these two usages can combine, to get the causative form of an adjective. This uses the punctual auxiliary, then prefixes that further with w-, yielding stems like wmasnar ‘widen (transitive)’ and wmenar ‘make hot’. Naturally, these stems have no possibility of taking a different aspect, but doing so wouldn’t make very much sense anyway.

[Darren]

Those are some nice verbs. I don't really have anything else to say but they sure are nice.

[Glenn]

Does this verb system still include the preverbs that you described earlier? I was rereading your description of them, and found it extremely interesting.

[bradrn]

Does this verb system still include the preverbs that you described earlier? I was rereading your description of them, and found it extremely interesting.

Ah, I should have made this clearer. The preverb is just what I’m calling those formatives which come before the verb core, which includes the verb stem plus some surrounding modifiers (which I haven’t yet described properly). So in a sentence like, say, this one from earlier:

Simwaskun, bmbiwraw.
sim-waskun, b-mbi-wraw
little-pain, 1s-CS-good
It was hurting a bit, but I’m better now.

The verb stem is wraw, analysable as the auxiliary w- plus the adjective raw, and it’s preceded by the preverb bmbi-. (In fact, now that I look at it again, it might have been better to use the punctual here: bmbimaraw, with verb stem maraw ‘get better’.)

[Glenn]

Thank you for the clarification! I thought that this was the case, but I wanted to make sure, in case you had changed your mind since the earlier post.

[bradrn]

Thank you for the clarification! I thought that this was the case, but I wanted to make sure, in case you had changed your mind since the earlier post.

I may yet change my mind on things, but I’ll try to be clear about it if I do.

[bradrn]

The adverbial slot is peculiar in a couple of ways. Firstly, it’s the only slot which can in some cases admit more than one formative: e.g. sayseybwa ‘at home I do it like this’. However, this tends to be avoided.

In the spirit of ‘being clear about changing my mind’, I’m not sure if I still love this idea. There’s surely going to be other places in the clause to put adverbials, and using those makes rather more sense than trying to cram everything into this single verb slot. Taking that approach would also force the speaker to express information structure more clearly, but presumably that’s a relevant factor if they really feel the need to specify two adverbials.

Otherwise: this language is coming together remarkably quickly, by my standards. The next posts will probably cover phonetic rules, basic clause structure and the verb stem (not necessarily in that order).

[bradrn]

Phonetic rules

I’ve mentioned in passing a few of the phonetic rules which apply within Eŋes phrases. There are in fact quite a few of these rules and they play an important part in binding together different parts of the sentence, so let’s get them out of the way before moving on to more grammatical matters.

Allophony

Consonant allophony constitutes the clearest instances of phonetic processes in Eŋes. I already mentioned some of these processes when discussing phonology, but I omitted a lot of the detail there.

An interesting point about Eŋes is that a lot of the allophony is conditioned on syllable structure, rather than on any higher unit of language. Of course, the surface form therefore depends on how syllables are constructed. I will discuss this point a bit later in this post.

The consonant with the most extensive allophony is /r/, which exhibits no less than three distinct allophones, as follows:

• At the end of an unstressed syllable, it is realised as a tap [ɾ].
• Before voiced consonants except /j w/, and at the end of a stressed syllable, it is realised as a fricative [ʒ].
• Otherwise it is a trill [r].

Thus, for instance, basannirwmaraʔr ‘sitting I made it bigger’ conveniently shows all three allophones: [ba.san.niɾ.wma.raʔʒ].

A process of palatalisation affects the two phonemes /ɡ s/, when adjacent to /i/ or /j/ in the same syllable. It occurs to varying extents depending on the speaker:

• /ɡ/ goes for most speakers to [ɟ], though it’s not uncommon for it to go all the way to [d͡ʒ]
• /s/ generally palatalises to [sʲ] or [ɕ], more rarely to [ʃ].

So bmsiŋwawgimi tno ‘with care, I turned it into a rope’ could be [bə̆.mɕiŋ.waw.d͡ʒi.mi.tʰno]. (Though, editorial note: I haven’t fully decided on the syntax of indirect objects yet.)

Finally, stops (excluding the affricate /t͡ʃ/) are unreleased at the end of a syllable. Voiceless stops may additionally be aspirated before another consonant in a cluster, as shown in the previous example.

(Re)syllabification and schwa-insertion

As mentioned, pretty much all of this allophony depends on knowing the syllable structure of the phrase in question. This is determined through a process of resyllabification.

I call this resyllabification because it operates across morphemes and lexemes. For instance, moʔ kbuy ‘old place’ is [moʔk̚.buj] — that is, resyllabification operates across nouns and their associated adjectives. Within the verb complex, an initial adverbial does not participate in resyllabification, but it operates across everything else, with the exception of a final particle if there is one. (NB. final particles are one of the things I haven’t described yet: they’re elements like yaʔ ‘intensifier’ or nis ‘polite’ with mostly pragmatic impact.)

Within these bounds, resyllabification aims to construct, from left to right, as many CV(ʔ,y,w)(C) syllables as possible. Consonant clusters are distributed between syllables, unless this is phonotactically impossible. For instance, the earlier example of moʔ kbuy gets parsed as moʔk.buy irrespective of the lexeme boundary.

A more interesting case is a bit earlier in the post, b-msiŋ-wa-wgim-i. In this case the initial cluster /bms-/ is too large, but there is no way to resyllabify it. The solution is to insert an epenthetic schwa to make the initial syllable CV: [bə̆.ms-]. In practice, this only ever happens within the preverb, which contains the single-consonant morphemes /b- n- d- m-/. In other parts of the sentence, morphemes are at least CV or VC.

Hiatus avoidance

I’ve already mentioned a couple of times that Eŋes avoids vowel hiatus. This is not absolute, of course: for instance, the sentence rsa blisŋumŋun ‘I see the front’ has hiatus between the object and the verb. Nor is hiatus avoided between the preverb and the verb. Its most prominent domain of application is in the verb core, where it applies until after the object pronoun.

Hiatus avoidance is triggered when a morpheme ending in a vowel or a glottal stop is placed immediately before a morpheme beginning with a vowel. In this case, the second morpheme is modified, in a lexically determined way, to prevent the hiatus. There are three possibilities:

• Most morphemes add a glottal stop before their initial vowel: for instance, ase → -ʔase ‘stumble’, on → -ʔon ‘gap’, aŋgar → -ʔaŋgar ‘tree’.
• Some morphemes delete their initial vowel: for instance, iŋ- → -ŋ- ‘3p subject’, inab → -nab ‘dog’.
• A few morphemes beginning with /i u/ change their vowel to a semivowel: e.g. isen → -ysen ‘go’.

Thus, a phrase like isaʔ-iŋ-mo-isen becomes isaʔŋmoysen ‘certainly they do not go’. (And ‘they do not stumble’ would be iŋ-ma-ase → iŋmaʔase instead.)

Making this more predictable, for verbs the choice of hiatus solution depends on the verb root, not the stem: so e.g. all verbs in √as- insert a glottal stop to resolve hiatus.

Stress

Stress is not contrastive, and generally operates at the phrasal level (as in e.g. French). In Eŋes, stress has few phonetic correlates: mostly, it affects intonation, and can also result in a slightly longer vowel. The exception is when a stressed syllable ends with an /r/, in which case it is realised as [ʒ] (as detailed above) — though even then, this is not universal across speakers.

Generally speaking, stress tends to come at the end of a phrase. Within the verb complex, the most prominent stressed element is the final verb or inner formative. (Object pronouns and other such following elements can never receive stress.) The last element of the preverb can also receive stress, unless it’s a pronominal subject. Other parts of the sentence do not usually receive stress, unless they have focus.

[bradrn]

There is another use of the verbal auxiliary: prefixed to any verb stem, it derives a causative form.

Following my recent discussion with Ahzoh, I‘m considering changing this to make the verbal auxiliary a general transitiviser, rather than only a causative. With this change there would be few if any ambitransitive verbs, and most transitive verbs would be derived from intransitive ones. The auxiliary would retain its current causative use, but gain wider scope as an applicative (mostly instrumental and comitative): thus we might have wawlesnaŋ ‘walk with’, or wafeys ‘climb (tr.)’.

In general I like this idea; it feels very ‘Eŋes-like’ (for lack of a better word). But I need to properly work out the semantics before I can be really happy with it.

[bradrn]

The verbal auxiliary, revisited

I’ve decided to fully commit to the idea in the last post. No longer will wa- be a mere causative: now, it is a very general transitivity marker, which can signal instrumental and comitative applicatives as well as causatives. (Though, it’s not the only applicative — there will be a separate benefactive applicative, at least.)

Now, that doesn’t mean that there’s free variation between the various meanings. Quite on the contrary, wa- is bound quite tightly with the following verb stem — the result is listed in the lexicon, as is often the case with derivation. We can identify strong tendencies, though:

• With intransitive verbs where the subject is non-agentive (‘unaccusative’ verbs), including those derived from adjectives, only a causative interpretation is possible: asoŋ ‘be low’ ⇒ wasoŋ ‘lower’, menar ‘hotten’ ⇒ wmenar ‘heat’.
• With other intransitive verbs (‘unergative’), and with transitive verbs, the result is generally applicative: wlen ‘travel’ ⇒ wawlen ‘travel with’, asanmarnir ‘sit’ ⇒ wasanmarnir ‘sit on’, wnerfin ‘make’ ⇒ wawnerfin ‘make using’.
• A handful of verbs, mostly transitives, are ambiguous between readings: rwenlon ‘face sth.’ ⇒ warwenlon ‘make face / face alongside’.

Despite lexicalisation, the auxiliary is still productive, applying even to loans: e.g. bin ‘be clean’ ⇒ wabin ‘clean (tr.)’ (cf. Hlʉ̂ bid ‘clean’; the final -n comes from being slotted into the Eŋes intensive).

Arguably, some bipartite verb roots contain an obligatory auxiliary: e.g. √w-rfin ‘related to making’. Since the second part (here *√-rfin) is generally unattested by itself, this is somewhat of a moot point. But the transitivity certainly matches what we’d expect from a verbal auxiliary.

So much for the synchronic situation. But I’d like to talk a bit about the diachrony too. I haven’t made much mention of Eŋes diachronics before, but the evolution of the auxiliary goes some way towards explaining its synchronic distribution, so it’s worth discussing. (Besides, I find this a particularly interesting corner of linguistics.)

Firstly, we can observe that both Eŋes and its relatives make extensive use of serial verb clauses (though this is something I haven’t mentioned yet for Eŋes). Most of its relatives reflect a verb form *waq ‘make/do’, and so it seems reasonable to suggest that the Eŋes verbal auxiliary wa- ultimately derives from a causative SVC: ‘I make it low’ ⇒ ‘I make-low it’.

Simultaneously, another important development took place: a transition from ancestral SVO order, to predominantly SOV in Eŋes. It appears that the source of this development was an instrumental SVC, with the accusative casemarker in Eŋes being derived from a verb meaning ‘take’. That is to say, we have a development like ‘I push it’ ⇒ ‘I take it push’ ⇒ ‘I take-it push’. (For reference, the same development is currently happening in Mandarin, in the form of the ‘bǎ-construction’.)

This has the effect of making transitivity a somewhat vague category, since you can add an instrumental argument to a great many things. Thus, for instance, you could always say ‘I take-knife cut bread’ (trivalent, with an instrumental), but now you could also say ‘I take-bread cut’ (divalent, with an object). So there’s a clear functional motivation for an explicit marker of transitivity.

But why did a causative take on this role, and how did it become an applicative? The answer lies in the sociative category: a variety of causation which lies midway between ‘true’ causatives and applicatives. In essence, a sociative verb form denotes an action where the causer and the causee are both agentive, performing the same action at the same time. A nice example of this is a sentence like, ‘The nurse walked the patient every day’ (cf. Shibatani & Pardeshi 2002): the nurse is making the patient walk, but at the same time the patient is themselves walking. The unique aspect of sociatives is that they are as much applicative as causative — in this example, you could equally well say that the nurse walks with the patient (comitative), for the patient (benefactive), or arguably even using the patient (instrumental).

Now consider how such a sociative would be expressed in pre-Eŋes. It could be a causative: ‘nurse make patient walk’. Or it could be an instrumental: ‘nurse take patient walk’.

(Note, incidentally, that the sociative means the agent can be either argument: ‘patient walk’ and ‘nurse walk’ are both true. Syntactically speaking, this is what lets sociatives be intermediate between causatives, which demote the subject, and applicatives, which demote the object.)

But now, consider what happens when Eŋes switches to SOV: ‘nurse take-patient make-walk’ for the causative, vs ‘nurse take-patient walk’ for the applicative. The instrumental SVC is now homophonous with the object marker, so if you want to clearly mark this as being sociative, the only option is to use the ‘causative’ construction. But of course, this is no longer a clear-cut indirect causative: it is a sociative, and as such, is an applicative as much as a causative. And, since the instrumental construction is no longer available, the auxiliary ‘make’ ends up taking the whole of this functional load (if that phrasing makes sense).

From this point, analogy takes over, and it’s quite easy to generalise the auxiliary to all kinds of other applicative usecases. Most importantly, as the accusative casemarker becomes fully semantically bleached, the auxiliary comes to take over its function as the instrumental applicative. Sociatives are easily interpreted as comitatives, too, a meaning which is reinforced by the well-known syncretism between comitative and instrumental. The benefactive meaning is less easily taken over, because Eŋes still has a productive benefactive, but even so, I suspect the auxiliary will add a beneficiary to some verbs (e.g. ‘buy’ ⇒ ‘buy for’).

Thus, we get the current state of affairs: a tightly-bound verbal auxiliary marking direct causation for verbs where that makes sense, and the instrumental/comitative where that doesn’t. Of course, it can still mark the sociative, and in fact some of the verbs I quoted above are still most easily understood as sociatives. But, as this discussion should have made clear, thinking in terms of discrete categories isn’t really helpful here — really, it’s a single continuum from causative to sociative to applicative, and wa- simply marks an arbitrary point on this continuum.

(References: Shibatani & Pardeshi (2002) [who I briefly quoted already] introduced the idea of the ‘sociative’, in a generally thorough discussion of the different kinds of causation. Also very useful was Jerro’s (2017) discussion of the causative–applicative syncretism in Kinyarwanda. And, of course, I must thank Ahzoh for first bringing this topic to my attention.)

[bradrn]

I’ve been thinking about revisiting Wēchizaŋkəŋ, once I’ve gotten Eŋes where I want it (and I do actually have a definite plan for it, so keep on reading!). I’ve always liked the aesthetic, and rereading the grammar, there’s definitely interesting stuff in there I’d like to explore further.

The major challenge with Wēchizaŋkəŋ has always been the sheer complexity of its phonology: I can’t even reliably construct sentences myself. But I’ve been reading some grammars of polysynthetic natlangs recently, and it’s made me realise that, if anything, Wēchizaŋkəŋ is the only one of my languages which even begins to approach natlang levels of complexity. (See, for instance, Seneca.) So, for that reason alone, it’s definitely worth reconsidering.

What do I want to do with it? Well, firstly, the slot system as it stands is far too neat and tidy. It badly needs complexifying. I also feel I could experiment a lot more with noun incorporation and bound lexical items than I did previously — after all, that’s what makes a language classically ‘polysynthetic’. And I’m sure that as I think about it, I’ll discover other interesting things I could do to it.

(Along similar lines, I’d like to redo Hlʉ̂ some day. It has some good ideas, but requires much more intensive rethinking than Wēchizaŋkəŋ does.)

[bradrn]

Another phonetic rule

On the other hand, processes like nasal assimilation (on which I’ll say more later) do apply across the boundary.

I just realised I completely forgot to mention nasal assimilation in my last post on phonetic rules. So let’s fix that before it’s too late…

Anyway, there’s not much to it. The main rule is that nasals assimilate in place of articulation to a following stop. So, as mentioned in the quoted post, iŋ+pesef becomes impesef. It also happens when there are two nasals, alongside degemination: so iŋ+mbi+pesef becomes imbipesef.

In some dialects, assimilation goes a bit further: nasals assimilate to a following nasal, not just a following stop. So something like lisŋumŋun ‘see’ would be realised as lisŋuŋŋun.

As with all phonetic rules, there’s also the question of scope. As it happens, nasal assimilation can be applied anywhere within a sentence, at least in normal speech. So the distinction becomes about where it’s mandatory, versus where it’s optional.

In general, nasal assimilation mandatorily applies in the same span where resyllabification can occur. For instance, the quote above mentions that it applies across the boundary between the preverb and the verb (as indeed it does in impesef). The most important difference occurs with adverbials: resyllabification applies to a sentence-final adverbial, whereas nasal assimilation is optional in this position. (Of course, it’s also optional with other adverbials, but resyllabification doesn’t apply to those in any case.)

[bradrn]

Basic clause structure

Although the verbal complex is greatly elaborated in Eŋes, non-verbal constituents also show considerable complexity. Particularly important are focus constructions, but basic clause structure is generally relevant for understanding the language.

Constituent order

I’ve already mentioned that Eŋes is basically SOV:

Eŋes tam torwoŋ.
[eˌŋes.tam.toɾˈwoŋ]
eŋes tam to-rw·oŋ
person food DEF.SG-have
The person had food.

(Notation: in the morphological representation, I’ll use a middle dot to separate the components of the verb stem, but in English will gloss the whole stem as one unit. So √rw+oŋ is ‘have’, not ‘related.to.having-STV’.)

That ‘basically’, however, hides some level of complexity in precisely where everything goes.

The subject is the simplest case: it always precedes the verb. If the subject is pronominal, it gets incorporated into the preverb (as previously described). Full NPs, meanwhile, are placed at the beginning of the clause: outside the preverb and before an object NP, though after a focalised object or adverbial if there is one.

Objects are more difficult. Essentially, there are three slots where a (non-focalised) object can be placed:

1. Immediately before the preverb: this is the normal position for a full NP.
2. A bound marker after the verb: this is available for pronominal objects.
3. After the verb: this is the normal position for indirect objects, or for focussed direct objects.

Which of these slots is used for what, however, depends on the sentence.

For a divalent verb, the usual case involves placing the object in slot (2) if it is pronominal, or slot (1) otherwise. For a trivalent verb, the direct object always goes in slot (1), even if it is a pronoun, while the indirect object goes in slot (2) if it’s pronominal or in slot (3) if pronominal. Thus, as a comparison with the previous example, we have:

Sam bimoysasiwpasim ban?
[sam.bi.moj.sa.siw.paˈsim.ban]
s-tam bi-mo-isa-siw-pasi·m ban
ACC-food 1s.Q-NEG-come-can-give 1s
Can’t you get me the food?

(Note that the pronominal subject goes after the direct object, because it’s part of the preverb. One could consider this OSV constituent order.)

Nonetheless, it is not uncommon to see a direct object in slot (3). This has a focalising effect, and is very often used to introduce something new into the discourse. This construction is, however, somewhat restricted: it cannot be used when there is an indirect object, and it cannot be used with subjects (which as I said always appear before the verb).

Thus, one might say:

Saybisalisŋumŋun inab raʔr ub.
[saj.bi.sa.lis.ŋumˌŋun.i.nab̚ˈraʔʒ.ub̚]
say-b-isa-lis·ŋu·mŋun inab raʔr ub
home-1s-come-see dog big new.information
Coming home I saw a big dog.

And then proceed to talk about the dog, how big it was, how it was very friendly (or not, as the case may be), etc.

That last example displays another very common feature of Eŋes sentences: the presence of a final particle. There is a large set of such particles, distinguished by their placement exclusively as the last element of a sentence. Here, ub acts to highlight that this sentence contains new information. I’ll hopefully be able to describe some more of these particles in a future post.

Adverbials

In many languages, adverbials are more mobile than other constituents, and Eŋes is no exception. In particular, there is a large class of specifically adverbial morphemes, which can occur in at least three places in the sentence. (Adverbial phrases are more complicated and will probably get their own post.)

As it happens, we’ve already seen the most common adverbial position: as the first element of the preverb. In this position, it has the side-effect of moving a subject pronoun ahead of any following connectivity marker. On the other hand, I’ve already mentioned that it’s phonologically somewhat separate from the rest of the preverb, since it doesn’t participate in resyllabification or mandatory nasal assimilation.

The adverbial can be emphasised by moving it to the beginning of the sentence. If there are any full NPs, this yields an observable change in constituent order: NP adv-V ⇒ adv NP V. But even if there are no full NPs, this can yield observable changes, as phonetic rules no longer apply between the verb and the adverbial: compare isaʔŋisaʔase ‘they really came stumbling’ with isaʔ iŋisaʔase ‘certainly, they came stumbling’. Furthermore, the connectivity morpheme no longer precedes the subject pronoun in this case, e.g. compare:

Saymbiŋigiʔnsem
[saj.mbi.ŋi.ɡiʔnˈsem]
say-mbi-iŋ-igiʔ·n·sem
home-but-3p-eat
They’re eating at home now.

Say imbigiʔnsem.
[ˈsaj im.bi.ɡiʔnˌsem]
say iŋ-mbi-igiʔ·n·sem
home 3p-but-eat
At home, they’re eating now.

(Note that √igiʔ-sem is one of those roots which loses the initial vowel in hiatus.)

A final option is to put the adverbial at the end of the verb complex: here, resulting in imbigiʔnsemsay. This doesn’t make any substantial difference to the meaning, though sometimes it’s used for new information (like object movement). Otherwise, it can be used to fit an second adverbial when the preverbal slot is already taken, though then again this is rarely necessary.

Additionally, the set of adverbials is not homogeneous. Quite a few adverbials have a strong preference for initial position, for instance sik ‘with cracking’ or aŋan ‘there’. More rarely, some prefer final position, like an ‘now’.

Argument focus

Argument focus constructions play an important role in Eŋes grammar. Depending on the choice of focus marker, they support a wide range of functions, including not only ‘focus’ per se, but also quantification, temporal indication and clause linking:

Anles ni dantofwesmarfini.
[anˈles.ni dan.tof.wes.maɾˈfi.ni]
anles ni dan-to-fwes·ma·rfin-i
king FOC thus-DEF.SG-demand-3s
It was the king who demanded it.

Ndini sewŋulbuʔi, banib mawaʔi.
[ˈndi.ni sew.ŋulˈbuʔ.i | ˈba.nib̚ maˈwaʔ.i]
ndi-ni sew·ŋu·lbuʔ-i, ba-nib ma-waʔ-i
2s-FOC know-3s, 1s-CONTR.FOC NEG-do.so-3s
You know it, but I don’t.

Res ner bisiwwabin?
[ˈres.neɾ bi.siw.waˈbin]
res ner bi-siw-wa·bin
honey FOC.also 1s.Q-must-clean
Must I clean the honey, as well?

Oŋoyŋo isaʔmbirwoŋsesi.
[oˈŋoj.ŋo i.saʔm.biɾ.woŋˈse.si]
oŋoy-ŋo isaʔ-mbi-rw·oŋ·ses-i
3p-FOC.NEG certainly-ST.CHG-want-3s
But now it’s them who really don’t want it.

The primary argument focus construction, seen in the examples above, involves moving the focussed argument to the beginning of the sentence (losing any case-marking it might have), followed by a focus particle. Subjects, direct objects and indirect objects may all be focussed.

The available focus particles are neatly split into positive (beginning with a consonant) and negative (beginning with a vowel), as follows:

Particle	Gloss
ni	general focus
nib	contrastive focus
nyan	‘only’
ner	‘also’
nirob	‘even’
ymen	‘some’
(o)ŋo	negative focus
(o)ŋoyan	‘not only’
(o)nda	‘but not’

When focussing a full NP, the focus particle is placed after the NP. Some particles have an initial vowel, which is deleted following another vowel: thus anles oŋo … ‘it wasn’t the king …’, but anles ŋu ŋo … ‘it wasn’t the bad king …’.

Pronouns are similar, but take a special focussed form before the particle. Many pronouns have two forms, one used for positive focus and another with negative focus. Thus, for instance, we have aŋay-ni ‘it’s them’ vs oŋoy-ŋo ‘it’s not them’.

Some subtleties here:

• Naturally, only one argument can be focalised at a time. So if you want to express a sentence like, ‘Some people hate only dogs’, you’ll need to use periphrasis or some other method. (Of course, such sentences are pretty rare anyway.)
• The negative focus particles are, in general, not interchangeable with the negation of the corresponding positive focus particles: oŋoyŋo wle ‘none of them went’ vs aŋayni mawle ‘it was them who didn’t go’.
• More generally, negating quantifiers is always something to be careful about. Unlike English, this focus construction takes scope only over the focalised NP: Marwa nyan bamalisŋumŋu ‘it was only Marwa who we didn’t see’, not *‘we didn’t only see Marwa’ (which would be Marwa ŋoyan balisŋumŋu).

This is another one of those things I’ll need to address properly later on, especially that last bullet point. (I suspect I’ll need to learn some formal semantics if I want to properly work this out…)

Verbless clauses

Verbs are extremely important in Eŋes (as may have already become apparent). But there are a few contexts where non-verbal clauses are encountered too.

A common case involves responding to questions, where a single constituent is an acceptable response:

Na ni tanbifoŋ? / Inab.
[ˈna.ni tan.biˈfoŋ || iˈnab̚]
na ni tan-bi-f·oŋ? / inab
what FOC here-Q-stand? / dog
What is this? / A dog.

Nan biʔfoŋ? / Say.
[ˈnan biʔˈfoŋ || ˈsaj]
nan biʔ-f·oŋ? / say
where 3s.Q-stand? / at.home
Where are they? / At home.

As seen here, nouns and adverbials are acceptable in this construction. (As are verbs, but then of course it’s no longer a verbless clause!) Notably, adjectives are not acceptable alone: they must be verbalised with the auxiliary to serve as the answer to a question.

One can also add a focus particle to this construction — most easily with a noun, but also with an adverbial (the only situation in which an adverbial can occur with a focus particle). Thus, for the first question, one might instead answer inab nyan ‘it’s only a dog’ or inab oŋo ‘it’s not a dog’.

In one case, the focus particle is mandatory: namely with pronouns. Thus, one cannot answer *ba to a question, but only bani ‘me’ (or, say, banyan ‘only me’ or baner ‘me too’, etc.).

Noun+focus constructions have some wider applications too. They’re generally common for presentational sentences, where the speaker wishes to point out something. It’s usual to add an adverbial too: as in fgis ni tan ‘this here is a palm tree’.

Similarly, one can add another noun to the end of a focus construction. This has a basically copulative meaning, yielding statements of class-membership or identity: bani wnus ‘I am a man’, bani bradrn ‘I am bradrn’. (Contrast with the verbal equivalent, si-bradrn brwoŋloŋ ‘I am named bradrn’.)

(Note that this ‘copula’ has very restricted usage — it appears mostly when both things being equated are nouns, and for presentational sentences only when one is adverbial. It is not used for adjectival predication, which uses the verbal auxiliary, nor with locational expressions, which as seen in the last example use verbs like foŋ ‘stand’.)

[bradrn]

Pronouns

Personal pronouns

I’ve mentioned the personal pronouns a few times now. Here’s a full table:

	Focus (positive)	Focus (negative)	Subject	Interrogative	Object	Accusative
1s	ba-	bo-	b-	bi-	-ban	rban
1p	be-	bo-	ba-	babi-	-ben	rben
2s	ndi-	ndi-	n-	mbi-	-din	ndin
2p	nda-	ndo-	d-	mbi-	-den	nden
3s	aŋay-	oŋoy-	∅-	biʔ-	-i	rŋay
3p	eŋey-	oŋoy-	(i)ŋ-	(i)mbi-	-ey	rŋey

Reviewing where each of these series is used:

• The focussed series are used before a focus particle, at the beginning of a clause. There are different forms before positive (or consonant-initial) and negative (or vowel-initial) markers, with most vowels being neutralised to /o/ for the latter (with the exception of 2s, ndi-ŋo).
• The subject series is used when the subject of a verb complex is pronominal. It is placed at the beginning of the preverb, or after the connectivity marker if an adverbial is present.
• The interrogative series replaces the subject series to form questions. (Note: this series is unique in being mandatory even if the subject of the verb is not pronominal, in which case the usually 1s form bi- is used.)
• The object series is used at the end of the verb complex, to indicate the direct object of a transitive verb or the indirect object of a ditransitive one, when it is pronominal.
• The accusative series is indicates a pronominal direct object of a ditransitive verb, being placed in the usual position for accusative object NPs.

One very important fact to note here: none of these morphemes can occur free. That is, these are all bound morphemes which can only be used together with another morpheme (usually a verb). The closest thing Eŋes has to an English-like free pronoun, such as ‘me’ or ‘you’, is a focussed pronominal: bani, ndini, etc.

It’s worth noting the extensive syncretism that exists, mostly between singular and plural forms. Negative focus collapses both the 1s/1p and 3s/3p distintions, while the interrogative collapses the 2s/2p distinction. Additionally, the 1s interrogative form is extended to mark sentences where the subject is not pronominal at all.

Demonstrative pronouns

Eŋes has 28 demonstrative forms, which can be arranged into a classical ‘table of correlatives’ as follows:

	Proximate	Distal	Indefinite/interrogative	Anaphoric
Person	taʔ/saʔ	aŋaʔ/sŋaʔ	naʔ/naʔo-	naʔos
Thing	te/se	eŋe/sŋe	na/no-	nos
Amount	traw	aŋraw	nor	narus
Adnominal	naʔsin	naʔsŋan	naʔsnan	naʔsnos
Location	sin	aŋan	nan	non
Time	sis	aŋas	nas	nol
Manner	sey	aŋay	nay	nayos

The four sets of forms are as follows:

• The proximate and distal series are deictic (or ‘pointing’) forms: referring to an object or concept somewhere relative to the speaker. The proximate series is used when close to the speaker (metaphorically in the case of Amount or Time), and similarly the distal series is used when far from the speaker. (Note: the Person and Thing demonstratives have irregular accusative forms, shown with a slash.)
• The indefinite/interrogative series is used to form questions and some relative clauses. More generally they are used as indefinite demonstratives: parallel to English forms like ‘someone’, ‘anything’, etc. (Note: the Person and Thing forms are again irregular, this time in having a separate ‘negative focus’ form similar to that of the pronouns.)
• The anaphoric series is used to refer to something previously mentioned in speech: much like English ‘aforesaid’. (The Adnominal form naʔsnos is precisely parallel to English ‘the said’, though more common.) In many languages, the deictic forms take over this anaphoric function, but Eŋes uses these separate forms instead.

The various demonstratives do not all behave alike. Demonstratives referring to people and objects (taʔ, te etc.) are treated the same as nouns, excepting their irregular accusative and focus forms. Demonstratives of location, time and manner (sin, sis, sey etc.) are adverbial, having the same behaviour as all other adverbials. (Indeed, I’ve already given at least one example where sey is used in the preverb.)

The remaining demonstratives (Adnominal and Amount) are more unusual. Like adjectives, they can modify nouns directly. However, unlike adjectives, they cannot be verbalised using the auxiliary: that is, they cannot act as predicates. And they are certainly not nouns, because they cannot appear without a nominal head. (Indeed, the Adnominal series is what Eŋes uses for ‘this’, ‘that’, ‘which’, ‘said’ as nominal modifiers.) Thus, they appear to form their own class.

[quinterbeck]

Why the term Adnominal rather than Determiner for that particular series? Just curious

[bradrn]

Why the term Adnominal rather than Determiner for that particular series? Just curious

‘Determiner‘ in IE grammar generally refers to a syntactic slot which can be filled by a whole range of different items: articles, demonstratives, possessives, quantifiers, etc. Eŋes doesn’t have any single slot like that, so the term wouldn’t be accurate. ‘Adnominal’ is more precise.

[Travis B.]

Why the term Adnominal rather than Determiner for that particular series? Just curious

‘Determiner‘ in IE grammar generally refers to a syntactic slot which can be filled by a whole range of different items: articles, demonstratives, possessives, quantifiers, etc. Eŋes doesn’t have any single slot like that, so the term wouldn’t be accurate. ‘Adnominal’ is more precise.

I always interpreted "determiners" as being a figment of certain linguists' attempts to treat all languages as being dialects of English.