Unit 9. Sentence grammar

And this is the reason why every man should expend his chief thought and attention on the consideration of his first principles: – are they or are they not rightly laid down? – and when he has duly sifted them, all the rest will follow.

Caution! We have to think!

Halfway through this tutorial, it is time we had a closer look at sentence grammar.

Lemizh does not distinguish full sentences from incomplete sentences such as ellipses, clauses or phrases, and other kinds of speech. Every grammatical unit above the word is considered a sentence (or a sequence of sentences) and thus subject to the rules of sentence grammar. Indeed, even words are cited as one-word sentences – with a full stop at the end: sxnèz. –, as you have seen in previous units.

You already know three of the rules from unit 2. Here they are again:

One. Sentence structure

A word of level n is subordinate to the nearest word of level n−1 in front of it; the parole acts as a word of level zero.

Two. Definition of objects

An object of a word in a sentence is a word subordinate to the former, its predicate, plus all of its own objects.

There is no rule defining the word order of sibling objects. Possible reasons for deciding on an order include, but are not limited, to:

• Only the first object can be marked as agentive.
• The default order of plot cases is nom–acc–dat. (See the section on word order on the pragmatics pages.)
• A reversed order of plot cases (dat–acc–nom) focuses on the receptive viewpoint.
• Objects and words belonging together content-wise, such as coordinations, should normally be placed together.
• An object with a deep structure of sub-objects should be placed in final position to avoid climbing several levels in mid-sentence as this makes the sentence harder to understand (and climbing more than five levels is impossible anyway).
• Objects can be rearranged for stylistic reasons such as euphony, scansion, rhyme, etc.

Inversion, including bracket inversion, and desorption allow for even greater freedom of word order.

Three. Outer case

The outer case of the first word of an object defines its relation to its predicate’s stem via its descriptor; the outer case of a level 1 word is zero.

These are the remaining rules:

Four. Identity of action

An instance of a word stem designates a specific action.

When you instantiate (say or write) a word stem, for example the one in wmàb. ‘weave’, you are thinking of one certain action – one of weaving in our example. This may involve the action of several weavers (as in ‘We are weaving’), the use of several threads or the production of several cloths. It need not even be temporally or spatially connected (as in ‘We are weaving every Thursday morning’). In other words, you are thinking of a certain subset of all the weaving there is. Everything referring to your instance – i.e. the word itself, its objects and relative pronouns – refers to this same subset, to this action of weaving.

This set model has nothing to do with the one we encountered in the chapters on partitive cases.

Naming the same word stem (wmàb.) for a second time creates a second instance, which may or may not be identical to the first instance. Recall the difference between wáx wìe. speak-fact¹ PI_n−1-dat-nom^2a. ‘He is speaking to himself: the sender of speaking is the recipient of the same action of speaking’ and wáx wìxe. speak-fact¹ speak-dat-nom^2a. ‘The one being spoken to is speaking: the sender of an action of speaking is the recipient of an action of speaking, but not necessarily the same’.

Even the stem of a relative pronoun instantiates an action: one that is identical to the referred stem’s action (or to the parole).

We informally say that wmèb. instantiates a person or a group of people when we really mean its stem instantiates the action of weaving that defines the weaver(s).

The following two rules qualify or narrow down this subset; they provide additional information about it. They are the actual reason why wmàb. can refer to a subset of weaving (as opposed to all the weaving there is).

Five. Completeness of cases

A case characterises the action it refers to completely with regard to its case descriptor.

When you have instantiated a word stem, you can add objects to describe its associated subset more specifically, to restrict it. For example, you could add a nominative object. If you do so, it has to name the complete sender – meaning that a nominative object ‘father’ excludes from your instance of wmàb. all weaving that is not done by father. A temporal object ‘every Thursday morning’ excludes all other times he weaves. A dative ‘wool’ … but you get the idea. In this way you can say more and more precisely what subset of all weaving you meant when you said wmàb.. Likewise, if you add to your instance of wool a nominative object ‘goats’ (a word of level three), this restricts the set of wool, which in turn restricts the weaving. (And restricting xacgàzw. ‘twinkle’ with a dative object ‘Seamus’ excludes all the twinkling light that doesn’t reach Seamus’s eyes.)

By the way, restricting a set does not necessarily mean reducing it. If father is only weaving on Thursday mornings anyway, the temporal object will not reduce the size of the set, but it will still add meaning to the sentence. Rule Seven will make this clearer.

The stems of objects are in turn restricted by their predicate and its other objects: the mentioned wool can only be wool woven by father on Thursday mornings. This can be formally shown by an inversion.

The instance of weaving we have been talking about so far cannot include a group in which father is weaving, since naming the nominative object ‘father’ has already established him as the complete sender. We can, however, instantiate a new action of weaving (that is, say or write wmàb. for a second time) with the group as nominative, of which father can be a part: wmáb qmiè psrènbe. weave-fact¹ group-dat-nom^2a father-partnom-nom². ‘Grouped people, one of which is father, are weaving’. (The coordination has to equate father with part of a set of grouped people, not with a set of groups; therefore qmì. needs an inner dative.)

The same applies to inner case. Brackets (and coordinations) are important applications of Rules Four and Five. In the phrase Ìx wèxy. male-acc¹ speak-nom-acc²., the man is the same as the speaker because both words cover the content of the same instance (Rule Four) of man-making completely (Rule Five); the predicate via its inner case, the object via its outer case. Finally, the inner cases of relative pronouns are also covered by this rule: using pronouns to refer to things that have already been introduced ensures that we are referring to the same things, not to new ones of the same kind; recall these two examples.

Partitive revisited

The mechanism described above implies that Lemizh objects are inherently restrictive in the sense of English restrictive clauses and phrases (‘the man that is speaking, the speaking man [as opposed to the silent one]; Nechwatal the mechanic [as opposed to Nechwatal the cardiologist]’): these help identify their referents, or, in Lemizh terms, the instances of ‘man’ and ‘Nechwatal’, respectively. Contrast this with non-restrictive clauses and phrases (‘the man, which is speaking; the speaking man; Nechwatal, the mechanic [the only Nechwatal I know]’): these provide additional information about their referents, which are already identifiable beforehand.

In Lemizh, all objects are restrictive with the exception of cumulative brackets and coordinations. Brackets cannot restrict their predicate because the restriction has already been done by the predicate’s inner case. (The man is the same as the speaker.) And this is the reason why we need partitive brackets to express restrictiveness: technically, they aren’t brackets because their outer case does not match their predicate’s inner case, and so this argument does not apply. The same case can be made for coordinations. However, we do not need to translate all English restrictive constructions with partitives, as we will see in a moment.

The same affair, seen from a different viewpoint: xacgàzw meqxèn hemÌsi. twinkle-fact¹ star-nom-partnom² Seamus-acc-dat². ‘Some [or possibly all] of the stars twinkle at Seamus’ differs from the non-partitive phrasing in that the stars that twinkle at Seamus are not necessarily all of the stars instantiated (Rule Four) by the word mèqx.. This becomes important whenever this instance of stars – but now the whole instance – is referenced a second time, as by an object or a pronoun. The reference to the whole instance need not necessarily be grammatically overt, as in the third of the following examples.

While omitting the partitive in the first two examples would definitely change their meaning, this is not so for the speaking man example: as Ìx. is the only instantiation of ‘man’ here, it can instantiate any set of men, notably that of speaking men (or a subset of it). This renders the partitive unnecessary in most situations, even those that are restrictive in the above sense. Conversely, the partitive case implies that the speaker has a larger set of men in mind, perhaps the ones he is pointing at, or the ones present in the room, or the ones he will refer to in the next sentence. Sometimes, though, we want to mark a clause or phrase expressly as non-restrictive, i.e. as not necessary for identifying the referent; strategies for doing so will be discussed in unit 16.

Now let’s have a look at inner partitives. The first of the following examples refers to the weavers twice (once completely and once via a partitive). In the second the reference to the whole set of walkers is not overt, as there is no accusative object without an inner partitive, but it is implied to consist of the beaver and the butcher.

The fact that an object has to name its complete descriptor poses a problem with a number of cases, notably instrumental, benefactive, causative and consecutive, which are often used to describe just one of several tools, beneficiaries, causes, and consequences, respectively (or more precisely: just part of the set which constitutes the tool etc.). If the wind-caus opens the door, there are normally other causes such as the fact that the door was ajar. The mechanic’s-ben coat may have seen other owners and thus beneficiaries of coat-making. The literal, purely grammatical solution would be to construct such objects with inner partitives. The practical solution is to ‘talk loosely’ by omitting the partitive. This results in a sentence that is literally false but usually good enough, just like saying ‘There are cubes of Turkish Delight’ is false but good enough if there are roughly cube-shaped bits of Turkish Delight. The same is true for words with an inner instrumental, i.e. tool nouns, which would need an outer partitive if we were to speak literally because at least the wind is also part of the means of sailing. (You cannot just avoid Rule Five by omitting partitives at will. See Noninstantiability narrowing and Widening with a partitive case on the pragmatics pages for more on this.)

Completeness of cases is also the reason for the somewhat weird descriptors for partitive cases. The sender of an action could be taken from any number of sets, so the partitive nominative could never be a complete characterisation – save for the addition of ‘thought to be taken’ to its descriptor, which uniquely pins down the set the speaker is thinking of in the current context. (I have often omitted this part of the descriptor, or paraphrased it, in example sentences because it is just a technicality and adds nothing to clarity.)

Six. Missing objects

A missing object is equivalent to the absence of information about its descriptor.

Normally, you will not add objects in all existing cases to your instance of wmàb.. If you leave out, say, the locative object, Rule Five cannot restrict the weaving with respect to its location. But thanks to Rule Six this does not mean you refer to all locations where father is weaving wool every Thursday morning. It rather means that you omitted any spatial information about your instance of weaving. Leaving out the nominative object of ‘wool’ does not mean this instance of wool refers to wool by all woolly animals. It might still be only about goat’s wool, but this information has not been included in the sentence.

Typical reasons for omitting objects are:

• The speaker does not think the information important.
• The speaker does not know the information.
• The speaker is unwilling to share the information.
• The information has already been communicated earlier.

An example that may not be obvious: jnÌ sklÌxtyn. 1/1-acc¹ room-acc-partacc². ‘all rooms’ does not necessarily describe all rooms there are, because ‘room’ could have a missing accusative object ‘green’. If so, the object of the main predicate would only be the ‘green rooms’ (which is, incidentally, a good example of the importance of Rule Two). But of course the information that the rooms are green can only be omitted if it is clear from context anyway. If this isn’t the case, the addressee can safely conclude that the mentioned phrase is really intended to mean ‘all rooms there are’. (We will further discuss this issue in the chapter on real-time interpretation on the pragmatics pages.)

Rules Five and Six imply that every instance of a word has exactly one action, one sender, and so on: Five excludes additional senders if one nominative object is already present, and Six gives meaning to missing objects, establishing them as an integral part of Lemizh sentence grammar.

Seven. Degree of reality

Given an object and its predicate, the predicate is considered more real and the object more hypothetical.

The sentence làxt xOàjy sokrateÌse. ‘I want¹ to hear² Socrates³’ contains the information that I want¹ something (i.e. to hear Socrates), but not that I actually hear² someone (i.e. Socrates). The main predicate ‘want’, so to speak, lives in the world the sentence is talking about (more formally, the world of the parole), which is the more real, while its object ‘hear Socrates’ lives in the world of my wish, which is the more hypothetical of these two worlds. The parole, having level zero, acts as the predicate to the sentence as a whole and is therefore still more real. This reflects the fact that the parole is part of the real world; it is as real as anything linguistic can be. Turning this around, we see that the sentence is more hypothetical than reality: it can be a metaphor or some other figure of speech, a statement about a fictional or otherwise imagined world, an error, a lie, a linguistic example sentence, etc. We call the main predicate’s kind of reality, the one that is just one level more hypothetical than the parole and the real world, grammatical reality. Every sentence claims grammatical reality of its main predicate, or, loosely speaking, claims its main predicate.

The sentence qàzg làxty xOàjy sokrateÌse. ‘I think¹ I want² to hear³ Socrates⁴’ claims that I think something, while the other two verbs, so to say, are pushed down one degree of reality: ‘want’ becomes more hypothetical, and ‘hear’ even more so.

In the sentence dmàt mÌse lÌbvy. ‘I see¹ white mice²’, the object ‘white mice’ is hypothetical as well: there is no claim that the mice actually exist. But, depending on the situation, the addressee might still justifiably conclude that the mice do exist, if there are no other good explanations such as a hallucination; but this is a question of meaning and context, not grammar. In ‘Lucy² gets¹ a bottle² from Father Christmas²’, Lucy is only as real as the mice: her existence is a logical consequence of the fact that an inexistent person cannot be given a bottle. The same is true, of course, of Father Christmas and the bottle. Such conclusions, which are not licensed by Rule Seven but by logic and/or context, are called pragmatic reality. Pragmatic reality even works for missing objects: ‘Lucy gets a bottle’ implies the existence of a giver even if the nominative object is not overt. Whether and to what extent an object is pragmatically real can depend on many factors: in ‘She eats¹ it because of the vitamins²’, the existence of the vitamins is questionable because she might only think there are vitamins. In ‘The beaver makes¹ lace²’, the lace at first only exists in the hypothetical world of the beaver’s lace-making intentions, but when it is finished it becomes pragmatically real.

Here are some consequences of Rule Seven:

• Inversion. The degree of reality can be changed using an inversion. Examples will follow in the chapter on modal adverbs in unit 13, and others.

• Noun phrases. Until now we have translated phrases having a main predicate with an inner non-factive as noun phrases: wèx nàgcy. speak-nom¹ war-fact-acc². ‘the teller of a war’. As this phrase claims grammatical reality of the teller, the more accurate translation is ‘A teller of a war exists’ or ‘There is a teller of a war’. Lemizh does not know stand-alone phrases (or clauses) but only complete sentences.

  Note that Rule Seven is about the degree of reality of the predicate, not of the predicate’s stem. While the teller of the war is grammatically real, the underlying action of speaking is only pragmatically so. Technically, this is pragmatic reality of the missing factive object. By contrast, no pragmatic reality for the action follows from predicates such as wòx. ‘the intention of speaking’, wÒlx. ‘the reason for speaking’, or Ìv. ‘the content of eating, food’.

  We have encountered several other noun phrases that are equivalent to whole sentences, not necessarily with inner non-factives: ‘the behaviour, the strange one’ = ‘He behaves strangely’, ‘sneezing, two individuals’ = ‘She sneezes twice’, ‘the second meeting’ = ‘They met for the second time’, etc.

• Conferring reality. Because of Rules Five and Seven, the predicate of a bracket confers its degree of reality on the object: Ìx wèxy. male-acc¹ speak-nom-acc². identifies the man with the speaker, as we have seen; so the speaker must be as real as the man. To expressly claim that the mice you have seen are real, you can say dmèt mÌse lÌbvy. see-nom¹ mouse-acc-nom² white-acc-acc³. ‘(There are) seen things, white mice’. Conferring reality also works for partitive brackets: if a speaker is taken from a set of men, there must be men.

• Speech acts, statements that aren’t about the world but are actions in the world (‘I promise to come’, ‘You are hereby promoted’), seem to contradict the notion of the parole being one level more real than the main predicate. But then, we can say wxàjg vaà —. promise-fact¹ PI_n−2-fact-fact² …., conferring reality on the main predicate. And then, following Rule Six, we will usually omit the pronoun, and all is well.

• Inversion ban. A negator must be the predicate of the negated action or thing:
  • nà xOájy axileÌsi. ⇒ xOajná axileÌsi. not-fact¹ hear-fact-acc² Achilles-acc-dat^3a. ⇒ hear-fact-not-fact¹ Achilles-acc-dat^2a. ‘Achilles isn’t listening’, but not **xOáj axileysì nÌa. hear-fact¹ Achilles-acc-dat^2a not-acc-fact². ‘Achilles is listening, which he isn’t doing’;
  • nà xOàRjy. ⇒ xOaRjnà. not-fact¹ hear-temp-acc². ⇒ hear-temp-not-fact¹. ‘He never listens’, but not **xOàj nÌaR. hear-fact¹ not-acc-temp². ‘He is listening, which he is never doing’;
  • nà xOÒljy. ⇒ xOOljnà. not-fact¹ hear-psu-acc². ⇒ hear-psu-not-fact¹. ‘He has no reason to listen’. The inverted form does make sense here: xOàj nÌOl. hear-fact¹ not-acc-psu². ‘He is listening, for which he has no reason. He is listening for no reason’ is no self-contradiction.

Compounding Rule Two revisited

Per Rule Two of compounding, ‘in the relationship between the original predicate and object (modifier and head of the compound, respectively), the rules of sentence grammar are retained as far as applicable’.

• (One. Sentence structure. Not applicable because the structure of the original sentence is lost by the compounding process.)
• (Two. Definition of objects. Unimportant because the head has no objects of its own in the original sentence.)
• (Three. Outer case. Not applicable because the head’s outer case is lost.)
• Four. Identity of action. Both modifier and head are instantiations of specific actions in the original sentence, which however do not necessarily match the instantiation of the compound. For example, the object in tilcdrìR djÌty. more-cons-one-egr¹ day-acc-acc². ‘next daytime = tomorrow’ must be discrete for the ordinal numeral to make sense; but the compound djyttilcdrÌ. day-acc-more-cons-one-acc¹. can also be seen as continuous.
• Five. Completeness of cases. The epenthetic case characterises the head completely with regard to its descriptor.
• Six. Missing objects. The only overt object of the modifier is the head. All other objects are missing and thus indicate the absence of information about their descriptors.
• Seven. Degree of reality. The phrase avkmàr mlÌvy. eat-fact-allow-loc¹ sweet-acc-acc². ‘a place where one may eat sweets, a place for eating sweets’ does not claim the existence of a place where sweets are actually eaten, as ‘allow’ has a higher degree of reality than ‘eat’.

Punctuation and other symbols

Here is an overview of the puctuation marks. We already know the pauses of speech, the separators and the straight quotes.

Draft of a Lemizh keyboard. The labelling is incomplete.

Exercises

	How many different actions occur in the following sentence?
	gané ganìe.
	sing-nom1 sing-dat-nom2a.
	Which actions do the cases in the following sentence characterise completely?
	wáx wiè dmÌaR.
	speak-fact1 PIn−1-dat-nom2a 3/4-acc-temp2.
	What can you conclude from the presence of the word lÌxw. ‘green’ in the following sentence, apart from that we are talking about all the green rooms?
	jnÌ sklÌxtyn lÌxwy.
	1/1-acc1 room-acc-partacc2 green-acc-acc3.
	Can you tell whether the beavers in the following sentences exist, and by which kind of reality?
	dmàt trÌxki.
	see-fact1 beaver-acc-dat2.
	dmàt trÌxke.
	see-fact1 beaver-acc-nom2.
	dmèt trÌxke.
	see-nom1 beaver-acc-nom2.
	ià veè dmèty trÌxke.
	love-fact1 PIn−2-nom-nom2 see-nom-acc2 beaver-acc-nom3.
	Translate and compound, and invert the translations unless inversion ban applies:
	He eats nothing-acc.
	The sound of a trumpet came from nowhere-ela.

Unit 10