Universality of grammar and grammatical universals
von August Dauses
Inhaltsverzeichnis:
Introduction
Chapter 1: Isolating versus inflecting and agglutinating languages
Chapter 2: Grammatical morphemes as relative indicators and concomitant phenomena
Chapter 3: Grammatical marking, classification and word formation
Chapter 4: Grammar and linguistic history
Chapter 5: Schematic, pleonastic and secondary usage
Chapter 6: New categories - new redundancies
Conclusion
In the following we will give a summary of the general language theory we have developed during the last few years (Dauses, A., Einführung in die allgemeine Sprachwissenschaft 1997; Englisch und Französisch. Zwei indogermanische Sprachen im Vergleich, 1998; Ökonomie und Kybernetik natürlicher Sprachen, 2000; all published by Franz Steiner) together with a number of further thoughts developed in the meantime and produced as a consequence of our theory. This enables us to define grammar universally and to delimit it from vocabulary and at the same time we are in the position to explain where exactly in a sentence grammatical markings and grammatical morphemes can occur in the various languages of the world. In a first step, we will examine what kinds of grammatical markings which we are more or less familiar with from our Indo-European languages are missing in other languages (or also in some of the Indo-European languages).
If we imagine a fictitious language, in which all kinds of grammatical markings are missing or used to be missing, we come to the principles of an isolating language, of a language then which can do with a minimum of markings.
After that we will inquire into the nature of grammar and into its delimitation from vocabulary and word formation. If grammar cannot be determined onomasiologically, then what use is the unity of grammatical categories that are so different among themselves? Could it be its merely relative character?
Finally, we will ask the question to what extent grammatical morphemes can be characterized as 'weak' linguistic signs with also many analogous usages and secondary usages (respectively secondary meaning) and why languages which have a lot of grammar never cease to produce new morphemes and new categories that nonetheless are still found in that framework we instinctively consider to be grammar.
1. Isolating versus inflecting and agglutinating languages
When talking about grammar, we first and foremost think of morphology and morphosyntax, that is of indications of case and number as far as the noun is concerned, the marking of person, tense, mode and aspect regarding the verb; we think of the usage of prepositions, articles, pronouns, of the formation of the adverb from the adjective, of the usage of some conjunctions, etc. So it seems to us quite natural that these grammatical categories exist and also that the corresponding grammatical morphemes have to be learnt alongside fixed rules concerning their usage. These rules range from more or less force of habit to the so-called 'obligatory' usage. Occasionally, these grammatical morphemes are so closely linked with the lexeme that the lexeme on its own does not occur any more (cf. Latin canto) and in many cases the marking is such a force of habit that an 'ellipsis' at least seems very disconcerting, cf. the fictitious answer *Yes, see. to the question Do you see the house? The answer without subject and object pronouns, which appears strange to us, includes however all relevant pieces of information, because it becomes quite clear from the context who sees what, and in Japanese this answer would be the only possible one. And when a Japanese says (in imitation) *See house., this could mean I see a house, the house, houses, the houses. Personal pronouns are rarely used in the subjective case; there are no articles and there is no marking of number. Therefore we cannot speak in terms of number or article in the Japanese grammar, and the personal pronouns are not integral parts of the verb either as it is the case with Indo-European languages (English I see him) so that it remains to be reflected upon whether in the Japanese language they have to be regarded as grammar in the stricter sense. Not every economical and economically used (!) linguistic sign belongs to grammar, for grammar according to our understanding also means usage according to set rules and therefore not only in those cases when the relevant piece of information is necessary for understanding! We will come back to this point later.
Grammatical categories and markings, which we take for granted, are therefore by no means universal, and this is most evident when we compare different languages, language stages and language types. The way this becomes most explicit is to consider isolating languages (for example Chinese, Vietnamese), which, in simplified terms, can be called languages without morphology.
Such languages operate with syntax in order to mark the relations in a sentence and otherwise use only full words (with a few exceptions). As a consequence, they have no grammatical rules in our sense either (or only very few).
Yet, we do by no means have to learn Chinese in order to understand the principles of an isolating language, a language which therefore is poor in grammar and economical (!). Instead, we construct a fictitious language which can do without morphological markings and ask ourselves what kinds of alternatives are needed to respect the necessary distinctions in order not to put at risk communication. By doing this, it will soon be clear what we intuitively understand by grammar and morphology and by which means they differ from vocabulary, but at the same time it will also be clear that many phenomena of our Indo-European grammars are not necessary per se, but that they are also due to a background of linguistic history.
As a consequence, we do at first without the constant marking of the person as far as the verb is concerned (by means of the personal ending or subject pronouns) and we use this kind of marking only in case the person is newly introduced or a change takes place (*I went to town, had breakfast, smoked a cigarette, went back. Met a friend. He said…). The Japanese or also isolating languages show us that this is perfectly possible. We would only have to pay attention to avoid confusion between a noun and a homonymous verb (fish – I fish; fire – I fire etc.). This can most easily be achieved by avoiding to use the same root of a noun and a verb that are homonymous but that have a completely different meaning. Instead of taking this verb, we take a completely different word or use a periphrastic construction (for example *to catch fish for to fish); at least this is the way the Japanese and the Chinese languages work. This kind of economy (the elimination of the personal marking) is detrimental to the equally economical possibility of word formation. However, we would also like to qualify that by saying that this kind of word formation in the Indo-European languages has from the first day onwards by no means been used systematically (that is as far as all nouns are concerned) and that it is also economical in a restricted sense, because the verb just includes one of thousands of possibilities of dealing with an object (you cannot only catch fish; you can also kill, fry, eat, sell, cut, feed it, etc.). This means that we always have to rely on further lexematic means respectively 'periphrastic constructions'.
Consequently, the obligatory marking of the person as far as the verb in the Indo-European languages is concerned goes back to word formation. The personal sign becomes an intrinsic part of the verb and remains that way, also in cases where the noun and the verb have grown apart from each other to such an extent that it is impossible to confound them any longer. (cf. German Gang – gehen; Wille – wollen; Gesang – singen, etc.). This results in a marked redundancy, which has a further consequence: languages that normally mark the subject with the verb by means of a personal ending or by a personal pronoun, analogously make use of an object pronoun with clockwork regularity (Ich sehe ihn; er sieht mich; wir sehen es, etc.). In this case the Japanese is much more economical and more sensitive as far as the context is concerned, as well as the Chinese as an isolating language, which does not know the obligatory usage of object pronouns; it is only more frequent there, because otherwise the verb would be understood in a passive sense (Ich sehe = ich sehe/ ich werde gesehen).
In a next step, we also eliminate the tenses, which help us in the Indo-European languages to recognize the verbs, and use instead time adverbs and a neutral verb form (present tense): *Yesterday I go to London, meet my friend.
For less specific indications we use for example once or some time ago in order to make a positioning in the past possible (it works similarly in the Vietnamese language), and naturally we use these indications just once at the beginning of the text and not in every single sentence. We would not repeat the adverb yesterday continuously either.
We can also do without further past tenses: it is by no means necessary to distinguish the ‘aorist’ from the perfect and it is by no means that all languages have this distinction either (cf. Latin, spoken French, Japanese, etc.). We even less need a pluperfect: the sequence of events usually results from the context or can be expressed by an adverb or a conjunction (after he came…). Sequence versus simultaneity can finally also be distinguished by marking the progress expressing simultaneity (consequently: He had breakfast. I came. – He was having breakfast. I came.).
The future tense is unknown to many languages or only rarely used, too (Old English, vulgar Latin, spoken German, Japanese). Adverbs together with the present tense are sufficient, all the more the future event often expresses a wish, a hope or an obligation so that you can also use the corresponding verbs.
Later on, such verbs developed a morphological future in some Indo-European languages, too (I will go; French j'irai < ire habeo).
We do not need a morphological category of aspect that would be necessary to express that a verbal action is in progress either. To express that, theoretically also adverbs (for example English just, German gerade) or periphrastic constructions (French être en train de…), consequently lexematic means are sufficient, which we only use when the emphasis of the progress seems especially important to us. This makes it possible to avoid the redundant marking of the aspect, which is used in English.
The morphological category of mode respectively subjunctive is even less needed: we express desire, hope or doubt with verbs or modal adverbs (to want, to hope, to doubt respectively probably, perhaps, etc.). We just need a particle for the imperative as a sentence mode, possibly also for the question in case intonation is not sufficient.
Sentences in the passive voice on the other hand can frequently be substituted by sentences in the active voice (Der Hund wird vom Vater geschlagen = Der Vater schlägt den Hund), or they can be substituted in a sentence consisting of a subject and a verb simply by omitting the object in the sentence (consequently: *we love = we are loved); as it is the case in Chinese; cf. also English the office opens at 5 o'clock = 'the office is/ will be opened at 5 o'clock'). The Chinese language, however, is more sensitive as far as the context is concerned and more flexible than Indo-European languages with their binary decisions (on whether to use the active or passive voice). Consequently, a fictitious *Lange nicht sehen (for example after a greeting) can in theory be interpreted as we have not been seen for a long time or simply as we have not seen each other for a long time. On the whole, isolating languages are by far more flexible than those having morphological markings.
However, we can not only eliminate markings concerning the verb, but also those concerning the noun. We take it for granted that most nouns are marked as far as the number is concerned. But instead of this schematic and less specific marking (singular = 'at least one object', plural = 'more than one object') we could also use flexible indications of quantity ('some', 'more', 'many' etc.) and otherwise leave it to the context how the noun should be interpreted (fictitious: *I see the house = the house/the houses; cf. English I bought the fish!).
By comparison, if we use our plural, we leave it open, too, whether it is a question of two, three or more objects. Isolating languages, but also the Japanese language perfectly show that this system functions as well.
We can also do without the marking of the case functions by using morphemes as modern English and French show, simply by marking syntactically the distinction between a subject and an object or the one between a direct or an indirect object.
The genitive as well could be substituted by a syntactic means: father's love = *father love; this is the way the genitive is marked in Chinese (also without the attributive particle de), and likewise even further attributes (to our understanding the adjective, the relative clause, the adverb as an attribute of the verb) can be marked by the position on the left of the word it refers to.
We do not need case or prepositions for indications of place either, but we can use nominal concepts alternatively (cf. for example an der Seite, am Rande, im Inneren), consequently that which we have understood by lexemes so far, which we only use in case they are necessary for understanding, but not according to set rules and which therefore are redundant once again (especially as governed cases or prepositions). In Japanese, Turkish or Chinese we are not allowed to count such nominal concepts (instead of prepositions) as grammar either, but we have to consider them as a part of vocabulary. A grammatical luxury is furthermore the continuous use of the definite and indefinite article in the Romance and the Germanic languages. They are missing in many languages of the world (Japanese, Turkish, Chinese) and they had to be developed in the Romance and Germanic languages. The indication that something has already been mentioned is rarely useful. As a consequence, we just need an auxiliary sign enabling us to point out that it is not about an already mentioned, but a completely new object, for which the Latin language uses the pronoun quidam 'any', which is only used very rarely as well.
Other functions on the other hand (the existence of alternatives/ the non-existence of alternatives: er ging zum Schrank und holte eine Socke/ das Gewehr), are only weakly marked in so far as the context and our knowledge of the world (more often than not you have several socks, but rarely several guns in a normal cupboard) are normally sufficient in order to understand the meaning so that a classification becomes necessary in only very special situations.
And you will not use the possessive pronoun analogously and redundantly either in languages in which it is not marked continuously that there is no alternative to the object already mentioned (definite article) (cf. English I wash my hair; he nodded his head; cf. also the German 'dativus commodi' of the personal pronoun: ich wasche mir die Haare). This pronoun only becomes useful when it refers to an unknown person.
The adverbial function does not have to be marked morphologically, if adjective and adverb can be distinguished syntactically (English the honest father works honestly; French le père honnête travaille honnêtement) and in English we also have morphologically unmarked adverbs (think different).
We do not need a copula that is free of redundancies in our language either (my father is a teacher/ tired), because we do not use a tense there either, but time adverbs and a neutral form of the verb. If the Chinese language uses an old demonstrative pronoun instead of a copula between two nouns (in imitation: *father, that teacher, 'my father, that is a teacher') then only because in Chinese the juxtaposition of two nouns can also be understood as a genitive syntactic construction ('father's teacher'). Sometimes economical (isolating) languages reach their limits, too and have to use auxiliary signs in order to avoid ambiguities!
We can also do without further 'copulative' elements; we especially think of certain conjunctions like 'und' or 'that'. A sequence of events is mostly understandable due to the context (cf. Er kam am Morgen und ging am Nachmittag; er kam und erklärte mir…), cf. therefore also Latin veni, vidi, vici. Even the simultaneity of two events or states of mind does not have to be marked by an (actually ambiguous or polysemous) conjunction, cf. er schrie und weinte. Similarly we do not need a conjunction 'daß' that can be missing in English (He said he was tired) either. Such a conjunction does by no means exist in all languages and can be explained historically by the redundant usage of a neutral object pronoun (He is tired. I see that. > I see that: he is tired. > I see that he is tired). The redundant pronoun could be explained as the redundant marking of the object pronoun. Consequently, redundancy generates new redundancy, and finally the conjunction goes back to the Indo-European splitting of a word stem into a noun and a verb, which made personal endings necessary and which led to the fact that analogously to the marking of the subject (by means of the verb ending) also the object became more frequent. The conjunction 'to' introducing the objective infinitive (He asked me to come) is dispensable as well and it would be possible to simplify the whole construction: *He ask (without ending or tense) I come. It is this way we are more and more approaching the isolating language structure. 'Copulative' prepositions like English to start to, French commencer à…etc. would be dispensable, too.
Now we can understand better that grammatical morphemes or what we consider them to be are mainly characterised by being used highly schematically. In fact, the speaker does not always subtly distinguish whether the morpheme in the context is more or less useful or whether it would also be dispensable there so that you can also speak of a redundant usage. We only have to think of the continuous use of the tenses or of the continuous use of personal ending or pronouns together with the verb etc. Consequently, grammatical morphemes are by their nature characterised by being used redundantly - may it also be to a varying extent. This is also meant implicitly whenever you are speaking of rules applied to the usage of grammatical morphemes whenever you are speaking of the obligatory usage of grammatical morphemes. The term 'obligatory' however has to be slightly qualified, for it first and foremost derives from the normative or didactic grammar: nothing is completely obligatory. Therefore it is perhaps more adequate to speak of habits being fixed to a varying extent, according to which grammatical morphemes are used in a specific context. In fact, we generally prefer to speak of pleonastics instead of redundancy because of various reasons:
- There is only redundancy in the sense of the definition if the relevant piece of information is given several times in a sentence or syntactic construction (drei Häuser; nous chantons); in this sense many grammatical morphemes are not redundant, but only hyper-explicit and thereby pleonastic, in so far as they provide the kind of information the listener can deduce from the context without them; we only have to think of anaphoric pronouns or anaphoric tenses (gestern kam Hans; er ging in die Stadt. Dort kaufte er ein Buch...; the subject pronoun and the past tense do not provide new pieces of information).
- Some grammatical morphemes are to be expected with high probability in a certain context, but they are not obligatory or redundant in so far as there are also alternative formations or markings so that among other things also stylistic and connotative considerations can play a role when they are selected by the speaker.
In this case we only have to think for example of the choice between several forms of the past or the future (What have you done? – What did you do? The sun will shine – will be shining) of several facultative usages of the Romance subjunctive or of the effects that can result from the omission of special conjunctions ('und', 'daß'). Some markings are also pleonastic and not only redundant in so far as they also express additional meanings or connotations.
However, a decisive criterion for grammatical morphemes remains that they are mostly used according to certain norms so that their occurrence in the context can be calculated with high probability, and furthermore that they provide explicitly the kind of information that can also be deduced from the context (consequently implicitly), which is often closely linked with that.
It is true that also ‘full words’ can be used pleonastically-redundantly (Did you see the house? – Yes, I saw the house), but this is entirely up to the speaker and is restricted to occasional cases. Otherwise, we think of such redundancies as nuisances, cf. *I saw the house. The house was cheap. I bought the house. Now, the house is my property and I like the house; cf. for example also the fictitious redundancy of a time adverb (vs. tense): *Yesterday I went to town: I met a friend of mine yesterday. He told me yesterday that… I answered yesterday…
To summarise, we can now say that isolating languages are characterised by the fact that they mark syntactically the functions in the sentence and the relations and that they use auxiliary signs or particles only to a very limited extent. Grammar in these languages is first and foremost syntax so that there is neither a redundant marking of case, number, tense, nor a redundant usage of pronouns and conjunctions either.
By contrast, the grammar of inflecting languages is highly redundant. The functions and relations in the sentence (you only have to think of case, adverb, object clause, relative clause) are often marked by grammatical morphemes additionally to the syntax, and furthermore there are redundant usages of pronouns, articles, tenses, aspectual markings as well as governments (governed cases, governed modes) and congruencies (number) and much more. Closely linked with that are also numerous additional usages and sub-norms, which results from the schematic use of grammatical morphemes and which makes the learning of grammar of an inflecting language so difficult. We will come back to this point later. We would also like to point out a possible misunderstanding by saying that, whenever we talk of redundancy or pleonastics in the context of grammar, this must not be understood as a recommendation in order to abolish grammar or to simplify it to such an extent that it becomes the grammar of an isolating language. Isolating languages have norms as well, but others than inflecting languages, and those who want to 'abolish' the past tenses in our languages would first have to introduce alternative time adverbs (once, shortly before…).
And those who should ‘abolish’ the subject pronouns in English would have to see to it that there would not be any disturbing homophonies between a verb and a noun (to go – a go; to house – the house); furthermore they would have to introduce a new morpheme distinguishing the imperative from indicative forms (cf. *go = I go or 'go!' etc.). In highly redundant languages the omission of a marking can also fulfil a grammatical function. In a somewhat paradoxical way we could say that it is just the complexity of our grammars which opposes a simplification!
English has sometimes been compared with Chinese and has been considered an isolating language. This applies in the first place to the cases in which the syntax adopts the functions in the sentence (subject – object, possibly also adverb: think different). In other cases you have to take care not to equate analytical formations and an isolating language structure: consequently, it is true that the subject and object pronouns are analytical signs (consequently no bound flexives), but that they are also used highly schematically and therefore also redundantly so that we consider them as part of the verb (I go, you go…, I see him, you see me…).
Consequently, such and further analytical signs and formations remain in the tradition of the inflecting languages, we only have to think of analytical tenses: they are used far more than they are needed and therefore are used redundantly as well as pleonastically. This is exactly why we instinctively count them as grammar.
We have derived our (new) understanding of grammar principally starting out from the comparison of isolating languages with agglutinating and inflecting languages. This has got a big advantage, but also its limits, which we want to illustrate briefly: the advantage consists in the fact that we avoid mistakes due to interference when describing other languages and it consists in the fact that we can graduate things in foreign as well as in our languages: a subject pronoun that is used by far more rarely in Japanese or Chinese than in English or French seems to us also to be less grammatical or even seems to be considered as an integral part of vocabulary. By comparison, we would probably not count dieser Mann or Euer Ehren for 'er' respectively 'you' as grammatical morphemes either. Similarly, nominal concepts of indications of place in Japanese or Chinese (or Turkish) which, corresponding to our Indo-European cases and prepositions, we would count as grammar, would have to be considered as lexemes. In any case, these are not used pleonastically-redundantly here, but only when they are necessary (by contrast, cf. Latin in templo, with the juxtaposition of case and preposition, or English on Monday, with the pleonastic preposition for example in: He will come on Monday or also He goes to London vs. He goes home!) On the other hand, an (aspectual) marking of the progressive form of the verbal action in Chinese (corresponding more or less to the English to be about to…), which is only used when necessary, would also be less grammatical or simply a lexematic periphrastic construction in contrast to the expanded form of the English language, which is mainly used according to set rules and thereby in an obligatory way, consequently also in case the context would be sufficient for the understanding of meaning (I have no time now; I’m going to school). Compare also the expanded form of the English language with the lexematic periphrastic construction in French (être en train de faire…), which is only used in special cases.
By using our method to define grammar, we also avoid mistakes resulting from an onomasiological consideration: not everything expressing time, place, person, etc. can therefore be regarded as grammar, neither in inflecting nor in other languages; and only if we graduate things like this by considering pleonastics as a measure for grammaticality, it is possible for us to separate the tense from indications of time or to make plausible why we prefer to count case as grammar rather than prepositions, which we tend to count as nominal indications (Inneres, Zentrum, Mitte, Vorderseite, etc.), cf. also the following chapter. In contrast, another question to be reflected upon is what a school grammar treats as grammar and by what right. A Latin school grammar for example deals with the different noun and verb classes (cantare, habere, dormire…), so with the allomorphic forms of certain morphemes (for example of the infinitive), which could perhaps also be treated in the chapter about word formation; and a Chinese grammar for Germans will possibly treat in the chapter 'prepositions' that which in Chinese comes close to our prepositions, even if the relevant words (!) have a completely different status there.
Moreover, our method makes it possible to detach ourselves from a mechanical determination of grammatical morphemes and to treat for example all flexives schematically in the same way. The circumstance that a morpheme is closely linked with a lexeme (it is synthesised) must not be a reason to consider it already as a grammatical morpheme, except when we equate grammar with flexion in general. But in this case there would be, apart from flexives, no grammar any longer! The school grammar treats the comparison (for example Latin fortis - fortior) as grammar, perhaps also because you have to learn certain forms. However, it is not pleonastic or redundant per se yet. Similar considerations also apply to the conditional Latin cantarem ('I would sing'): it is used because it is necessary in order to express a condition, and only such constructions as si tacuisses, philosophus mansisses, with a triple marking of the condition are pleonastic-redundant: conjunction + subjunctive. In an isolating language the construction is the following: *si tacere, philosophus manere, even without the marking of the time level, which in this case must derive from the context (cf. also Old French si j'eus = si j'avais or si j'avais eu!).
These considerations should only serve to show that our understanding of grammar is not only identical with a traditional or orthodox concept of grammar, but that it is in some ways more flexible and more adequate for typological and comparative approaches. Not every flexive belongs automatically to grammar, and with regard to the Latin comparison fortis - fortior we would perhaps have to speak of an incorporating morpheme or of a morpheme relative to word formation.
2. Grammatical morphemes as relative indicators and concomitant phenomena
As we have seen, there are languages with many grammatical categories and markings and such with only very few markings (isolating languages). This raises the question as to what the unity of this grammar consists of, the unity of this grammar which despite all the differences relative to the languages of the world also has a lot of characteristics in common with other grammars so that we may suppose a universal principle behind. It is noticeable that - apart from several interference mistakes - we can distinguish grammar from vocabulary in foreign languages quite easily, too, even if this happens rather intuitively.
However, from a linguistic point of view the difference between grammar and vocabulary is by no means that easy and cannot be founded on an onomasiological basis either: apart from abstract relations and functional indications, grammatical categories and markings can at every time be replaced by (more specific) lexemes, consequently a tense by a time adverb, an aspectual marking by a corresponding adverb or a periphrastic construction (for example French être en train de...), the number by an independent indication of quantity, the mode by a modal adverb, a 'weak' conjunction by a more specific lexeme (dann, darauf for and, folgendes for daß). As we can see, there are also more specific lexemes for all grammatical indications.
However, the different grammatical categories among themselves do not form a unity from an onomasiological point of view either, and therefore it is not possible to find a generic term which includes what is common to the different categories.
We could also ask the question as to where the common field lies between the indication of person (as far as the verb is concerned), tense, aspectual marking, case, number, article, copula and certain conjunctions (und, daß). The grammatical categories are far too disparate to be treated as a kind of 'semantic field' or 'morphematic field'.
Consequently, there is no barrier between vocabulary and grammar from an onomasiological point of view and there is no unity of the different grammatical categories from an onomasiological point of view either. So this unity is based on a common behaviour rather than on common contents: grammatical morphemes are used very schematically. They are therefore also highly frequent and convey information that could even without them be deduced from the context so that it is possible to speak of pleonastics or redundancy.
This is also due to the circumstance that grammar represents a system of rules: we do not only learn what a grammatical morpheme means, but also where and when it is used. The speaker of an agglutinating or an inflecting language takes this for granted so that he, in turn, infers grammar from the usage rules. These rules, however, can by no means be taken for granted, but are, at first sight, even paradoxical, because it seems as if the language was dictated to the speaker so that the speaker would only be the executive organ of this language. Such an idea would even have mentalistic implications: in this case the speaker would be the slave of his linguistic system and even of the sentences and texts produced by himself!
As a consequence, we have overlooked something that distinguishes grammatical morphemes from lexematic signs or 'full words'. As far as these 'full words' are concerned, there are no rules dictating to the speaker when he has to use them either.
Obviously, it is not possible to dictate to the speaker about which objects he has to talk and how specifically he has to describe these objects, also because the speaker does perhaps not have this specific knowledge. He may know that his neighbour owns a car, but he does not know whether it is old or new, green or red, big or small or how fast it goes. A rule applying to the usage of such classifications in certain contexts would sooner or later inevitably lead to a nonsensical usage.
Grammatical categories are therefore by no means already to be considered as the basic categories of our thinking and experiencing and cannot be such; and this is the reason why there are for example no grammatical morphemes for size, form, weight, age, speed, rhythm, etc. either.
Consequently, it would be superficial to describe grammatical morphemes only in categories of frequency and less specific meaning: on the one hand, there are less frequent morphemes and on the other hand, there are frequent lexemes being at the same time also less specific; we only have to think of the adjectives for 'good' and 'bad'. Their use is exclusively up to the estimation of the speaker and is not dependent on a rule that would also hardly be possible because of the reasons we have just talked about: The speaker cannot be forced to give a judgement of quality; and similarly he cannot always know whether an object is good or bad (or old or new, etc.). However, if an adjective with the meaning 'good' was ever to become a grammatical morpheme, it would have to lose its original meaning first and would have to fulfil a new (grammatical) function.
The Japanese prefix o expressing politeness, which in o-genki ('esteemed health' = 'your health') has become a pronoun used to address somebody in a polite way, could, in a similar way, be understood like that. Consequently, such a possessive pronoun can again be used redundantly-pleonastically, like a subject pronoun in English or German for example.
Grammar often means something like a grammar of rules, and a grammar of rules necessarily means pleonastics-redundancy. A rule cannot be the following: do not use the morpheme X in the context Y, although it would be necessary, but only the other way round: use the morpheme X even where it would not be absolutely necessary. Such a rule can, as we have shown before, neither refer to the characteristics of concrete objects, nor to the description of the external world.
In a further step, it can be derived that such rules and usage norms can only refer to phenomena that the speaker himself has to be aware of at any time so that the (pleonastic) marking in certain contexts does not cause him further efforts either. Consequently, he only expresses explicitly that which he must already know or that which he can know when planning his speech act (sentence, text), irrespective of the question whether the listener needs this information or whether he can also deduce it from the context.
This would then be the trite reason why our grammatical markings seem so natural to us; in fact, they seem so natural to us that we are no longer aware of them so that not the marking but, on the other hand, the omission of them becomes noticeable. We do not only produce grammatical morphemes in order to convey a certain piece of information to the listener, but we also produce them as an expression and a mirror of these functions that we have to be aware of whenever we plan a speech act or a text! Only this way it becomes understandable that grammatical morphemes frequently represent concomitant signs of certain categories (cf. case and number as concomitant signs of the noun, person and tense as concomitant signs of the verb, conjunctions as concomitant signs of the subordinate clause, etc.).
To sum up, we can say: grammatical morphemes do not refer to characteristics of the real world, but in the first place to such functions inside the text that the speaker has to be aware of whenever he plans a sentence or a text so that he can also mark them at the 'surface structure' at any time. Of course, this does not at all mean that he has to mark them continuously (and for example in all languages)! We will come back to the historical background of languages rich in markings vs. those poor in markings later.
In the following, we would like to illustrate briefly our thoughts: we find a minimum of grammar in isolating languages that mark syntactically the functions and relations in a sentence. It may remain an open question whether this fixed syntax is really always, that is in every context, necessary, but it does not require from the speaker additional considerations or decisions either, because when planning a sentence he always has to know which elements figure as a subject, object or predicate or which element represents an attribute (genitive, adjective, relative clause, adverb referring to a verb). If he is not aware of these basic relations, he cannot plan a sensible sentence at all, at best only utter incoherent words. This is exactly the place where other languages can produce highly schematic grammatical morphemes, for example a nominative, an accusative and a genitive, in addition to a more or less fixed syntax (for example S-V-O), perhaps even a relative pronoun introducing a relative clause, an adverbial suffix (even if the adverb can perhaps be recognised additionally by the syntax). In this case grammar refers to the relations of the elements of the sentence among themselves and does not have anything to do with the characteristics of the real world.
In our inflecting languages we use quite schematically (in an obligatory manner) a personal ending or a subject pronoun with the predicative verb, whereas isolating languages proceed far more economically. Naturally, even the speakers of isolating languages (or also of the Japanese) always have to know where the predicative word or the predicative verb refers to, even if they do not mark it. Without this knowledge the cohesion of the text would soon disappear. The same applies to the object pronouns: the pronouns on their own are by no means always necessary, but naturally the speaker always has to know where the (transitive) verb refers to. The explicit marking does not force the speaker himself into making a new decision. We often speak of an anaphoric function, especially of the personal pronouns of the third and of the sixth person, but this simply means a redundant (pleonastic) use, if there are no ambiguities.
In our inflecting languages we use quite schematically tenses together with the predicative verb. They indicate the kind of relationship between the event and the time of the speaker and the relationship of the single events among each other in terms of time. This knowledge is an absolute condition for the understanding of a sentence or a text, otherwise the timeless sentence is dangling in the air or the text loses every cohesion and coherence. The speaker has to be aware of these relations at the moment he is planning his speech so that the varying frequent marking does not force himself into making a new decision either. The continuous use of these tenses, which are mostly anaphoric (!) again, leads to redundancies, which are, however, not felt to be disturbing either, because the speaker and the listener have to be aware of the temporal relationships of every speech act. This is why these tenses are so unspecified, too: the speaker and the listener must know whether two events take place approximately at the same time or whether they represent a sequence of time (in the past or in the sense of a future event), but they do not need to know anything about the distance that lies between the events (gestern, vor drei Tagen, vor einem Monat; in zwei Wochen, in einem Jahr).
Similar considerations can also apply to the grammatical marking of the aspect, in the first place of the action in progress compared with another action or with reference to a certain point in time (He was sleeping when I came/ at nine o'clock.) The speaker, who himself brings both verbal actions into play, also has to know about their approximately temporal relationship among themselves (simultaneity/ sequence).
As a consequence, the corresponding marking does not demand a new decision or a more specific knowledge of him. He must by no means know in detail which phase the verbal action in progress is going through, whether it has started just a moment ago or whether it is already almost finished. A periphrastic construction like Spanish estar + gerund (está trabajando) only indicates the progress of action at a specific point in time, but it does not say anything about which phase this action is going through, whether it has started just a moment ago or whether it is almost finished, etc.. By contrast, Spanish estar para indicates that the relevant action is going to start in an immediate future, and Spanish empezar a ('to start doing something') indicates that an action starts, consequently that it goes through an 'inceptive' phase. In order to be able to evaluate this, we already need specific knowledge about the relevant situation, knowledge that goes beyond the basic temporal relationship. This is exactly that which distinguishes lexematic periphrastic constructions from real grammatical periphrastic constructions: the former demand more specific knowledge, while the latter demand only the knowledge of the temporal relationship of one action with reference to another. And this is exactly why 'real' grammatical periphrastic constructions can also become obligatory (cf. in English the progressive form), but why they can never become lexematic periphrastic constructions, because the speaker does not at all always know which phase a specific action is going through. A grammatical periphrastic construction of the aspect only reflects the kind of minimal knowledge of the speaker that is absolutely necessary, if it is about two actions or events. This is the reason why a grammatical periphrastic construction can also become pleonastic-redundant in contrast to a lexematic one: this minimal knowledge about the temporal relationship of two actions or of a state with reference to a point in time can often be deduced from the context by the listener so that he does not need the corresponding markings any more (What are you doing? – I’m going to school; but in French Je vais à l'école; German Ich gehe zur Schule). Every language has to be in a position to distinguish the progress of an action from the beginning of an action, but not every language has a grammatical (!) periphrastic construction at its disposal. By not accepting our distinction between grammatical and lexematic periphrastic construction, we soon get into the difficulty of letting ourselves be guided by similarities as regards the form and the content when defining grammatical periphrastic constructions.
Furthermore, from an onomasiological point of view there will soon be no limits any longer, cf. he was about to…, he started to…, he wanted to…, he began to…, at the beginning…, etc. Grammaticality must neither be made dependent on form, nor on content (onomasiologically). In the first place, Grammar refers to basic relationships, in this case to relationships between two temporal events.
This also applies to an 'aspectual' tense, for instance to a resultative perfect, for example (I have bought the book/ discovered a new isle). Such a perfect indicates in the first place that there has not taken place anything new with reference to the described so that the result is to be considered as valid up to now. Such an indication does not demand a new decision of the speaker because he himself knows best whether he wants to let the event follow another one (*but I lost it an hour later/ and left it a week later) or not. Consequently, the listener can deduce the relevant meaning from the context respectively from the text structure even without an explicit indication so that the marking, which in many languages does not exist in this form either, once again can be called pleonastic-redundant (cf. languages with only one tense of the simple past).
Grammar, in the first place, indicates basic relations and does not refer to objective characteristics of the external world. Now it would be possible to raise the objection that, as far as the marking of the plural is concerned, it refers to external characteristics, not only to relations. This objection, however, has to be qualified: the plural is not characteristic of an object, but only indicates that it is about several objects disposing among themselves of similarities so that we can treat them as one single category. As a consequence, the plural is principally to be treated the same way as the variety of different objects, and generally the speaker also has to know whether he is talking about an apple or about an apple, a pear or a plum so that even the marking of the plural does not force him into making a new decision. Finally, the decision between singular and plural is made easier also by the fact that even the singular does not orientate itself towards the real world but towards the foreknowledge of the speaker. If I say Hans hat eine goldene Uhr, I do not exclude the fact that he owns several such watches. It is just the existence of one such specimen that is familiar to me. Consequently, the singular does not simply exclude the plural in the real world!
Consequently, the number is not a relational element, but it is based on the basic knowledge of the speaker: he has to know whether he is talking about a single object or about more objects. Therefore the number has not to be realised in all languages!
The definite article on the other hand, indicating that the object X has already been mentioned, can be used schematically in so far as the speaker actually has always to know whether he has already mentioned an object in the same text a moment ago or whether it is about a newly introduced object (which he himself is going to introduce newly!). So he has not to make a new decision, and the relevant information does not refer to the external world either, but only to the internal text structure the speaker himself is also responsible for. In the sentence: He went to the baker's and bought three rolls on the other hand, the article only implies that the rolls were sold by the person in charge of this action (at the baker's and not at the carpenter's) and it can therefore be deduced from our general knowledge: everybody knows that rolls are sold at the baker's. The decision, here for the definite article, does not confront the speaker with any problems.
Other grammatical phenomena can be considered as relational indications or simply as concomitant phenomena of the text structure, where the external world does not play a role either. Here we think of the conjunctions 'and' and 'that' that only indicate that two sentences belong together, in the sense of a sequence of sentences or in the sense of a verb with an object clause (Er schrie und brüllte; er sagte, daß…). These conjunctions do not per se convey any new information either; it is at best possible to achieve various stylistic effects by their different usage or their omission. (German Er aß und aß und aß…, Latin veni, vedi, vici; German Er sagt, daß er nicht will – er sagt, er will nicht). On the whole, such conjunctions represent however redundancies, in contrast to by far more specific conjunctions like 'because', 'though', 'thereby' etc. For their usage the speaker needs specific knowledge referring to the external world (and not only to the text structure). Therefore such conjunctions can only be rarely redundant. They are at best redundant as conjunctions themselves, indicating that there is a connection between both clauses. This connection could also be expressed by using two main clauses (cf. Er machte es, obwohl es verboten war - er machte es. Es war aber verboten).
Even the copula 'to be', which itself does not convey any new information, has a concomitant function as far as nominal predicates are concerned (N ist müde/ Lehrer) so that the speaker himself does not need to make a new decision at all. However, even the verb 'to be' in the sense of an indication of place (er ist im Wald) often has just a concomitant function, especially if the indication of place with a preposition still follows. That which is characteristic of most Indo-European languages is the fact that these 'empty' verbs are at the same time the most frequent ones: frequency, redundancy and grammar are very closely linked with each other!
Finally, so-called governed phenomena are mere concomitant signs; we only have to think of such cases that are dependent on prepositions (ex urbe, in silva) or of the governed subjunctive of the Romance languages (French je veux que tu viennes). Apart from some variations in the usage and apart from the possibilities of choice, they are mere redundancies produced only highly schematically by the speaker so that he himself does not have to make a decision referring to the characteristics of the object world. In this sense, the governed mode could also be interpreted as an allomorphic form of the indicative after certain verbs or conjunctions.
In principle, we could examine like this all grammatical phenomena of all the different languages of the world with the intention of finding out to what extent they represent attributions and functions (in a sentence or in a complex sentence), to what extent they express temporal or aspectual relationships, and to what extent they refer anaphorically to the text itself or to what extent they only represent concomitant phenomena; but they will always have one thing in common: they do not refer to the characteristics of the external world or just in so far as these (tense) are once again to be seen in relation to the speaker (or in relation to themselves). And even the number is not a characteristic of an object either; the plural only indicates that it is about different objects: mere concomitant phenomena on the other hand have nothing to do at all with the real world (governments).
Only a schematic or an obligatory marking is possible, which the speaker has to be aware of at every point in time when creating a sentence or a text, which corresponds in the first place to the internal relationships or functions in the sentence or text.
By contrast, the knowledge of the external world and of the characteristics of the objects differs from one speaker to another, and their linguistic realisation depends decisively on the communicative intentions of the speaker himself, consequently on his interests and aims. It follows that such characteristics cannot be used in an obligatory way nor highly schematically either.
In this new sense it is now possible to speak of the universality of grammar so that we can establish definitely for all the languages of the world that which can be grammar and that which is always going to be left to vocabulary, even if we were not be in the position to enumerate in an exhaustive list all potentially grammatical phenomena and sub-categories of all languages of the world. The sum of all usages and sub-usages is gigantic and yet the most different grammars of the world always resemble each other and there continue to come into existence also new categories and usages, which are very similar to the old phenomena or which even substitute them.
Furthermore, with our flexible method we especially avoid mistakes or weaknesses that derive from a one-sided definition on formal conditions (for example flexives) or the other way round from an onomasiological consideration (for example tense and indications of time; the locative and further indications of place) and we do not have to attribute certain phenomena or even paradigms (for example of the prepositions) at a flat rate to vocabulary or grammar either. Certain phenomena are more or less grammatical in so far as they are more or less used schematically-pleonastically and in so far as they thereby orientate themselves more or less towards the internal text structure or towards the external world. Consequently, from our point of view the prepositions of the Indo-European languages would have to be treated in a subtly differentiated manner, whereas they are treated at a flat rate in a school grammar. Prepositions can fulfil relational functions (case functions): they can be used pleonastically (cf. I'm going to London, he comes on Monday), but they can also be used as specific indications of place (he is under the tree). Moreover there are further usages, which could also be considered to be part of the word formation (to think of, to look at, cf. also the following chapter).
In any case our model resulting from a comparison of isolating with inflecting languages allows us to treat the question of the grammatical status and the degree of grammaticality flexibly, which we want to illustrate briefly with the example of the sentence modes.
Of course, the speaker always has to know whether he makes a statement, whether he asks a question or whether he wants to formulate an invitation, and he can mark these intentions highly schematically with the aid of a particle for a question or for an imperative.
Such basic intentions are not characteristic of the external world, but they do not concern functions and attributions inside the sentence or mere relationships (like the tenses) either, and the decision for the sentence mode is completely up to the speaker himself. In principle, he has the same freedom here as when choosing a lexeme (cf. Komm! vs. Ich wünsche, dass du kommst; Kommst du? vs. Ich will wissen, ob Du kommst.). We can speak of redundancy if apart from the intonation morphological means are used, too: we only have to think of the pleonastic question in English and French (Do you go home? – Est-ce que vous retournez?).
One thing, however, should have become quite clear now: in a scientific debate we have to ask the question of grammaticality of a phenomenon in a far more subtle way than in a school grammar, which only wants to give rules for learning a language.
3. Grammatical marking, classification and word formation
As we have seen, it is neither always so easy to understand highly schematic markings following the rules by using the concept of 'grammar', nor to even specify this concept of grammar. Indications of relations and indications of function inside the sentence and inside the text cannot simply be equated with the basic distinctions between different meanings as they are for example represented by the sentence modes. It is therefore absolutely legitimate to speak of a certain plurality as far as the determination of grammar is concerned just as it is legitimate to speak of a flexible and gradual determination of the degree of grammaticality. However, it is especially a gradual determination that makes the philologist determine which kind of criteria he wants to use. If this is not case or only on the basis of an intuitive foreknowledge, the points of view concerning the form and the content are easily mixed up so that it becomes soon impossible to delimit definitely 'real' grammar from vocabulary any longer.
In a theory concerning the Romance languages about the so-called pre-determination of grammatical morphemes in modern French, analogously to the Latin comparative (fortior) and to the elative (fortissimus), also augmentative and diminutive suffixes have been considered to be part of grammar, in addition to that analogously adjectives referring to size (for example French petit and grand). This very soon leads to a kind of chaos, because it is obvious that the distinction between grammar and vocabulary disappears immediately, if we try to reach an analogous treatment according to onomasiological points of view (cf. chapter 1 and 2). This does not only apply to the comparison of two language levels (Latin-French), but also to the comparison of different languages among themselves. Analogously to the Indo-European grammar we treat too easily in non-Indo-European languages as grammar that which is used neither schematically nor redundantly nor in an obligatory way, but that which is used as freely as the forms considered as lexemes (for example nominal concepts of case or prepositions with a meaning of place).
Consequently, an inadequate definition easily leads to an incoherent theory and to an incoherent treatment of the phenomena so that classifications of a concept may be counted as grammar for analogous reasons. Adjectives expressing the idea of something ‘big’ or ‘small’ (in the literal or figurative sense) are, however, such classifications that orientate themselves directly by the external world and by its assessment so that they cannot simply be counted as grammar, just as little as further adjectives or, more generally speaking, attributes.
In order to distinguish classification and thereby also word formation on the one hand from grammar on the other hand, our criterion of pleonastics - redundancy is well suited and at the same time it is in accord with the conceptions of grammar as an obligatory marking: a classification orientating itself by the external world cannot become redundant or obligatory because
- it is up to the speaker himself whether and to which degree he wants to specify something (for example according to size, weight, colour, quality, etc.) and because
- such a classification also demands specific knowledge (as far as size, weight, colour and quality are concerned).
This specific knowledge is however by no means always available so that an obligatory indication with reference to certain characteristics or attributes necessarily leads to a kind of misleading information because in this case the speaker indicates something he does not know in reality!
Every reader can find this out by using the collocates 'big' or 'small' for every noun in every sentence. Let us only take a sentence like Das Wetter in Spanien begünstigt die Landwirtschaft with three nouns, which will lead him to desperation or to a completely arbitrary usage of these adjectives, which would immediately be deprived of any meaning. They soon would become the emptied prefixes of the nouns.
Real classifications do not go together with real redundancy, otherwise they are no classifications any longer, but empty 'appendices'!
In connection with the delimitation of grammar from word formation it means the following: classifications that become obligatory turn into emptied appendices of the words in question or let us put it differently: classifications that are to become grammatical, re-disappear in the 'black hole' of word formation!
In fact, such phenomena exist in the natural languages: in Suaheli, nouns and similarly also connected adjectives referring to living things are connected with a prefix m-, which is said to have had the meaning of something 'big', whereas things connected with a prefix ki-, are said to have had the meaning of something 'small'. These are completely emptied prefixes that are learnt together with the words of a class. Further noun prefixes in Suaheli are to be assessed in a similar way.
We can find something similar in the Indo-European languages: the gender used to indicate the sex, but even if it is about living things the speaker cannot always know which sex it is about so that this gender was always used schematically before it was more and more 'emptied'. As a consequence, it was gradually used in a fragmented way from one analogous usage to another (also as far as plants and objects were concerned), and was finally memorised one by one. Consequently, a classification turned into an emptied suffix and finally into a morpho-syntactical behaviour (congruencies of gender) that is to be learnt together with the single noun, because the morphological form does not permit an unambiguous piece of information about the gender (cf. already classical Latin or even modern French: la maison, la main, le pain).
In the Indo-European language gender is basically grammar that disappears in the 'black hole' of word formation, before even this word formation became not recognisable any longer so that just a morpho-syntactic behaviour was preserved. As far as this gender is concerned, it is not about a redundant marking any longer, already because of the fact that the grammatical gender does not have a content any more. It would at best be possible to speak of redundancy as far as the sex is concerned and this because
- it is marked regularly and because
- it is often used additionally to lexematic information (cf. pater - mater; vir - femina; bos - vacca).
We could therefore, referring to gender, also speak of allomorphic forms or even better of an allomorphic behaviour.
By doing this, we have already reached an area that is typical of the Indo-European languages and that lies between grammar and word formation: the allomorphic forms are the different realisations of a grammatical morpheme and they develop among others from the crossing of two morphemes that used to have different functions (Latin curro – cucurri; dico – dixi, the perfect and the 'aorist' coincided as far as their functions were concerned) or from the emptying of a word formation morpheme being preserved as part of a grammatical ending. So different word classes are generated as it is the case with the nouns that differ according to the gender or as it is the case with the verbs, Latin cantare, monere, whose stem vowels go back to an old case form respectively to a causative.
Naturally, we have extremely simplified the highly complex connections between word formation, stem formation and the formation of allomorphic forms in the Indo-European language. However, this allows us just to illustrate that word formation and grammar in the Indo-European languages are closely linked with each other and that we can also count the different formal realisations of grammatical morphemes, that is the allomorphic forms, as grammar and that we thereby also consider the allomorphic forms and allomorphic distributions to be part of this grammar. Such an understanding of grammar has however nothing to do any longer with the fact that the usage of a morpheme is highly schematic and redundant, but only with the different realisations of these morphemes, consequently with the arbitrariness of the 'signifiers' when compared among themselves.
And this arbitrariness is therefore not universal either in the sense that it is necessarily part of the nature of grammar, but it is a part of a single language and a product of linguistic history. We simply learn this allomorphic form together with the lexemes (for example verbs) and certain grammatical functions (cf. tense, infinitive, etc.). In any case the allomorphic form does neither refer to contents nor to the structure of the sentence or the text, but it is an arbitrary distribution of forms, whose memorisation suggests a regularity as far as the lexemes are concerned (Latin cantare, habere, dormire; murus, casa, templum). This regularity however only consists in the memorised arbitrariness and the high frequency of these allomorphic forms that can simply be combined with many lexemes.
Allomorphic forms develop themselves among other things by the fact that original elements of the word formation are not understood any longer so that they are understood and memorised as a unity together with the ending.
There are also further connections between grammar, word formation and vocabulary, which in turn are characteristic of our Indo-European languages and which are connected with their history. The Indo-European has a category of its own of prepositions that represent neither nominal concepts nor verbal concepts and that represent so to speak the second generation of the cases referring to a place. At first, they can be used as free additives to these cases so that they can also become pleonastic-redundant (Latin *in, templo stat; ex, templo venit). In classical Latin we find as a relic the 'tmesis' (obstat, but ob mihi stat!). Being free additives they are also syntactically free and can therefore stand next to the verb and form a unity with it (ex, it > exit; in, it> init, etc.), also in the nominal area (introitus, abitus, exitus). They then form a new concept together with the lexeme, that is they specify the lexeme in question.
Now this can have as a consequence that we have to do with a very redundant marking of the indication of place or origin, if word formation and 'real' preposition are involved at the same time: de nave desilire, ex urbe ire, de provincia exire. Case, preposition and word formation prefix work together in order to mark the origin. This would not be possible in a different language, in which nominal concepts correspond to these deictic prepositions (for example 'inside, front, back').
Such nominal concepts are not free syntactically and would be understood as a subject or as an object of the verb in the 'wrong' place (cf. in imitation: Latin ineunt vs. *interiora eunt; interiora would then be the subject referring to eunt). The flexibility of our Indo-European prepositions has therefore also a syntactic background: they used to be freely movable additives with a deictic character and in fact no nominal concepts so that they could also be connected with the different word classes. And since these prepositions that are free and dispensable additives which, in a pleonastic way, make for example the indication of direction clear (desilire de, profisci ad, inire in), are also used a bit schematically, it is by far easier to use them analogously once again, cf. to look at, German schauen auf, to strive at, German streben nach; to hope for, German hoffen auf, etc. They can now be regarded as redundant indicators of the object (grammar) or even as integral parts of the corresponding verbs which are followed by an object.
This hybrid character also becomes distinct if such pleonastic additives assume for their part a grammatical or partial function once again. The development of a prepositional or ‘personal’ accusative in some Romance languages has to be imagined in a similar way (Spanish Maria ve a Juan 'John sees Mary', with the object at the first place in the sentence). An originally certainly pleonastic preposition used together with verbs indicating a direction (cf. English to look at) can be used to distinguish the object as aim from the subject, and since this is in general relevant only to persons but not to a person and an inanimate object (cf. A Maria ve Juan vs. El muro ve Juan: the wall cannot see), it was only the so-called 'personal accusative' that came to exist. Since this accusative however fulfils only in few cases an important function, it was not really taken into consideration any more and was used more and more schematically so that it became a pleonastic companion of the accusative referred to persons, therefore as far as orthodox syntax is concerned, too (Juan ve a Maria).
Grammar and word formation on the one hand and word formation and the formation of allomorphic forms on the other hand are therefore closely linked with each other in the Indo-European languages: the more pleonastic a grammatical marking, which is used for example together with a verb, becomes, the more often analogous treatments can occur.
To the extent, however, to which such prepositions are not used systematically – in order to mark the accusative, even if pleonastically, they are not memorised with this same function either, but as an integral part of the verbs. This is why they become 'governed' particles (compare consequently to see + O, but to look at, to strive for, etc.)
Perhaps it is possible to reduce these considerations to a simple formula: numerous allomorphic occurrences are often an indicator of highly pleonastic grammars, because it is only that which fulfils a very little function which can also be extended analogously or crossed within themselves and such crossings again lead to arbitrary allomorphic forms (= total synonymy as far as complementary distribution is concerned). Consequently, word formation and grammar are closely linked with each other in our Indo-European languages, even though we could but outline this connection only as an example. In another context we have already indicated this in chapter 1, in fact with reference to the marking of the person as far as the predicative verb is concerned (Latin canto, cantas, cantat, English I sing, you sing, he sings). This marking is part of the word formation because it is possible to deduce a verb from a noun (with a different meaning) out of the same root, and at the same time it is part of grammar, because the subject to which the predicative verb is referred is (pleonastically or also redundantly) indicated by it.
In this case grammatical redundancy is the consequence of word formation as shown by languages that are poor in redundancy. Such languages do not know such products of word formation and create periphrastic constructions (for example *Fisch fangen for fischen) or use completely different words, which has a far more rare marking of the person as a consequence.
Furthermore, such languages use the object pronouns in a more economical way so that further usages of personal pronouns (for example as relative pronouns) and sub-usages (for example as dativus commodi, etc.) become far less probable there. Thus, word formation indirectly has still further consequences for the grammar of a language and its development.
But even in our languages where we do not deduce verbs from nouns with the same root (for example German Bein, Nase, Stuhl) and even in such languages that generally do not know such kind of word formation (Japanese), the distinction between noun and verb seems to be a fundamental, perhaps even necessary distinction, which we therefore consider to be universal. This has probably in principle to do with the fact that in the Indo-European languages we create predicates that are no identity predicates (Vater ist Lehrer; das ist ein Stein) almost always with verbs so that we equate the function in the sentence (predicate) with the word class (verb). Strictly speaking, the function in the sentence is to be separated from the word class and we can obtain this for example by using analogously to the isolating languages words capable of forming a predicate (consequently especially verbs to our understanding) without a difference in a nominal or predicative function (consequently fictitious *people go; I like go; go is good for health; *I love you; we speak about love; love is good). We just must not commit the mistake of transposing this 'verb' according to the syntactic tradition of the English language into a 'real' verb and noun again!
Consequently, we have to start from the assumption that as far as both usages are concerned (the nominal one or the verbal predicative one) exactly the same concept is expressed. In this sense we have to do with polymorphous forms in the Indo-European; the same concept can occur as a noun as well as a verb and this verb can have predicative and nominal (non-predicative) forms (I go – to go/ going respectively Latin eo-ire).
This joining-up of certain forms and word classes with the function in the sentence is, however, not universal yet. It is only the function in the sentence that is universal and certainly also the circumstance that this function is recognisable in the sentence. This is possible by using own forms and markings or by using syntax and meaning. Isolating languages (for example Chinese) prefer the latter method, that is they use certain words ('verbs' to our understanding) in the same form in a nominal as well as in a predicative way. If these – always unchanged – words are to be considered 'verbs' remains an open question.
On the other hand, there are certainly in all languages such words that transmit a concept only neutrally, consequently that only 'name' it (compare the nominal function!) and which, as nominal concepts, can therefore only have the function of a subject or object. If they are used predicatively, the consequence is a confrontation of two nominal concepts that in general is interpreted as an identity predicate (N ist Bäcker; das (ist) ein Haus).
Such nominal concepts are necessary to name concrete objects for example. The determination of a predicative usage with a specific meaning (which does not only indicate the existence of exactly this object), for example a fictitious *to tree = 'to cut a tree') would have as a consequence that we could not call the object in question isolated any longer.
Such nominal concepts are, however, also useful respectively economical if they name concrete objects and illustrate complex facts and situations, which in turn are to be found in most different contexts. They then are specified by a word creating a predicate ('verb'), compare for example 'art, war, education, science': something can be real art, but it is also possible to study arts; you can be at war or wage war; and there is also a war industry; it is possible to provide education or to spread education; it is possible to promote science and there is a scientific theory and a criticism of science.
We cannot go into detail here, but we consider it to be universal for reasons of economy (!) that there are also mere nominal concepts that can only be used as a subject or as an object in a sentence, or only in nominal predicates (identity predicates) (N ist Bäcker, das ist Kunst). If these are to be called nouns, may remain an open question (in Indo-European not all nominal forms are nouns, let us only think of the nominal forms of the verb!).
As a consequence, word classes in the sense of our Indo-European grammar are not necessary or universal, but the functions in the sentence are universal and it is also universal that in all languages of the world there will be words that are only used in a nominal function. This has in the first place economical reasons, but it can also be justified analogously (modern Chinese has a compound ('love-sympathy') for the concept of 'love' that is not used in a predicative way any longer).
This digression about the word classes (word categories) shows on the one hand that in the Indo-European languages word formation and grammar are closely linked with each other; at the same time it qualifies all considerations on the 'nature' of these word categories (consequently also potential ontological ideas). On the other hand, this digression refers to the essential role of linguistic history for the formation and the further development of our Indo-European grammars, which means that it refers to the following chapter.
4. Grammar and linguistic history
So far, we have shown that there are languages with only very few grammatical morphemes and markings, and others which have a lot of grammatical categories and markings. In addition, we have tried to find out which functions and relative indicators can be called grammtical at all, and how grammar is different from vocabulary. This enables us to describe theoretically and universally any possible kind of morphological marking in any given language without claiming that these markings actually exist.
The differences between the actually existing languages of the world (some of which have a lot of grammatical categories and others only very few) must therefore also be accounted for on a historical basis. Yet, in the following, we do not intend to give an outline of Indo-European grammar, but rather do we intend to bring out the general principles which were responsible for the genesis of complex or also complicated grammars, taking the Indo-European languages as an example.
In the Indo-European languages, the splitting of the root of a word into noun and verb (with different meanings), and thus word formation, is responsible for the fact that the verb in question is always used with a personal marking (Latin canto, cantas…, English I sing, you sing…). As a consequence, free morphemes (that is the personal pronouns) were to emerge, in addition to the personal endings which were used highly schematically. Later on, these new free morphemes were to replace in turn partly the old personal endings. It is an open question whether the first development, the splitting into 'weak', that is schematically used morphemes (endings) and 'strong', that is free pronouns which were only used when required, was absolutely necessary. But this development became possible and it was also encouraged. And it is no coincidence that those languages which do not know this kind of word formation (Japanese, isolating languages) do not distinguish personal endings from personal pronouns either, but that they only use pronouns, that is free signs. And such languages do not distinguish either between clitic and non-clitic pronouns, that is between those pronouns which are used highly schematically so that they are considered as integral part of the verb and others which appear in isolation or in a stressed position. This splitting into strong and weak signs, however, which our example of word formation (the derivation of a verb from a noun) has illustrated, is a generally valid principle which enables us to distinguish grammatical morphemes from lexemes or more grammatical elements from less grammatical elements! We come back to this point in chapter five. Grammatical morphemes are weak signs to the extent that they are used more or less schematically and not only in those cases in which they are absolutely necessary in the context in question in order to convey a certain piece of information. Therefore, they also appear often in those cases in which they don't provide any new information nor any relevant new information; and this is exactly what we mean when we talk about rules or the obligatory use of grammatical morphemes.
Personal endings and (more or less) obligatory personal pronouns (subject pronouns or object pronouns) are therefore weak and grammatical morphemes, but those free pronouns the use of which the speaker decides upon himself can not be defined in the same way. Consequently, we could ask ourselves by what right we count these free pronouns among grammar and to what extent we let ourselves be guided by semantic similarity (onomasiological point of view).
In any case, this splitting into strong vs. weak (grammatical, more grammatical) signs is symptomatic especially for the Indo-European languages with all their redundancies, which we would like to illustrate briefly:
A past tense is a weak, unspecific, grammatical morpheme which is used highly schematically and which is not only used when it is necessary, so that it may even seem redundant next to a strong sign with more specific content (Yesterday I went to town). In Chinese, there is a particle which indicates the accomplishment of an action (le), which, in fact, is never used if there is a more specific adverb which indicates the past in the same syntactic construction; comparatively no-one would say in German: *Gestern damals ging ich in die Stadt! Two signs of roughly the same value may even result in a disturbing redundancy.
In isolating languages, however, particles which indicate accomplishment are obviously felt to be equal to other adverbs of time, there is no relation between strong and weak signs nor between lexeme and grammatical morpheme, or this relationship is far less distinct.
Therefore, it seems reasonable to put forward the hypothesis that, in isolating languages, more specific adverbs and less specific particles (for instance those indicating accomplishment) have always been equal in value, so that it never crossed one's mind to use them together in one syntactic construction, and therefore the concept of 'tenses' in our sense doesn't exist in these languages. Conversely, for languages which have a past tense, this hypothesis means that less specific markings were used with a certain regularity at the beginning, even before the more specific adverbs and also the lexemes emerged. The synchronic juxtaposition (Yesterday I went to town) would then be the result of a diachronic sequence: unspecific older markings are preserved when the more specific markings emerge and their use increases. Again, there would be a splitting into strong and weak signs which we have also postulated for the relationship between personal ending and personal pronoun. The weak sign is at the same time the older one, which is used more and more schematically and can also combine with a word class (verb); the stronger sign, however, is the younger one, which the speaker only uses when he wants to use it.
This hypothesis seems plausible to us especially as far as a past tense is concerned, and it is no coincidence either that almost all languages have such a tense with the exception of the isolating languages: lexematic words like yesterday, last year (literally-etymologically: 'the other day' respectively 'tired, late') haven't had this temporal meaning right from the beginning or were ambiguous as far as past tense or future meaning is concerned ('the other day' = 'yesterday' or 'tomorrow'?), so that they still needed an explicit reference to the time level. It became redundant only later when the words or collocations in question were used exclusively in one sense (for instance the past tense). If the vocabulary of a language is less developed it needs auxiliary signs which clarify the temporal relations so that we would have to ask ourselves once again why this was not the case in the development of the isolating languages: have they started anew on a more highly developed level?
Irrespective of these hypothetical reflections on the subject of the isolating languages, our hypothesis concerning the inflecting languages is as follows: weak temporal auxiliary signs emerged very early, even before the vocabulary was much developed, and their use remained a regular habit even later when more specific indicators of time had developed. This was all the more easily possible as even some of the more specific indicators used to be ambiguous as far as temporality is concerned ('the other day'), so that they were accompanied by auxiliary signs right from the beginning. And when, later-on, redundancies developed, the marking by means of a temporal sign had already become a regular habit which was not noticed anymore and which would not be revised (by force).
Therefore, such temporal markings for past tense are to be expected in most languages and they are more relevant than any markings for the future: firstly, there will always be more to report about things past than about an insecure future which, most of the time, only consists of few plans and secondly, the future meaning can be deduced from the non-marked form, supposing that the past tense is already marked. In most cases, he distinction between future and present tense does not pose a problem because of the context and the situation, as is illustrated by many languages which use the present tense also to express the future.
When, now and in the following, we talk about grammatical morphemes as 'weak' signs this is done first and foremost from a synchronic and functional point of view: we talk about morphemes which are used highly schematically and not only when required so that there will also be redundancies. Regarded on its own, a particle (e.g. a deictic expression) which is used to express past tense is not weaker than an adverb which gives a reference to the past. It is only weaker and less noticed in so far as the unspecific sign appears more and more frequently side-by-side with a more specific one in the same syntactic construction or that it appears totally schematically, always linked to the verb, even in those cases when the context is explicit as far as the time level is concerned (cf. a continuous narrative in the past tense: yesterday I went … I bought … I returned). The 'weakness' is only a result of redundant usage!
Therefore, flexives are not already per se weak grammatical signs, there is only a high probability that such very old markings, which are also little specific, will appear later with more specific signs side-by-side in one syntactic construction, so that then, they will be less noticed, considered as concomitant signs of these more specific signs and therefore become redundant.
This is especially the case as far as the so-called government is concerned, a special feature of the Indo-European languages. The old cases for place, for example, were restrained by more specific prepositions (initially free adverbs) so that they became schematically used concomitant forms (governed cases) (Latin in templo, ex horto). Had both categories emerged at the same time, the speakers would have decided between the alternatives (in the same way as the speakers of isolating languages decide between the alternatives of an adverb or a particle of the past); it would not have happened that the more specific sign is first used optionally side-by-side with the more unspecific one (fictitious: *ex, eo, templo), and that then it becomes more and more frequent until it finally degrades the older sign (case) to a mere concomitant form. The pattern is the same in this case as in the relationship between tense and adverb of time.
But there is also one significant difference between government and redundancy (yesterday I went…): government always also means a semantic weakening of the governed phenomenon which gives up more and more of its own meaning. In the Latin in templo, ex silva, de muro, sub muro, cum copiis, we can hardly explain today which is the meaning of the ablative case, so much has it become a mere (empty) concomitant form of the preposition. A redundant tense of the past, however, always retains its 'past' meaning and therefore remains semantically stable.
Government is therefore a special case of redundancy and by no means universal: it is based on the fact that a more specific sign can occur side-by-side with a less specific one in the same syntactic construction instead of replacing it. In the case of the Indo-European prepositions, this is only possible because, being adverbs, they were syntactically mobile and did not constitute an integral part of the sentence construction. In those languages which use noun forms instead of prepositions the nominal form cannot be used as a free component, cf. for example: *Er befindet sich an der Vorderseite des Hauses (= 'in front of the house'), but certainly never: ***Er befindet sich an der Vorderseite des Hauses im Hause! In this case, redundancy could only be produced by force. Word classes or rather the fact of not belonging to a word class, and syntax play a distinctive role in the emergence of such governments.
But the syntactic mobility also plays a distinctive role in another case of government, if we think of the governed use of the subjunctive mode in Latin and the Romance languages. At first, the phrase rogo te ut venias! is just as redundant as the phrase *ich bitte dich, du sollst kommen! The subjunctive is not as specific as a lexematic sign, but this does not make it per se a weak 'morpheme' which would have to become redundant for this reason. It only becomes a 'weak' morpheme when, after a rearrangement of the sentence structure, it is regarded as part of a construction and becomes dependent on a more specific lexeme (verb): ut venias! rogo te. => rogo te, ut venias! So, the syntactic mobility of the final clause led to a juxtaposition of a more specific and a less specific sign within a complex sentence, so that the attention of the speaker/the listener was always directed towards the more specific sign and the mode was interpreted as concomitant form to the more specific sign (cf. also governed cases and prepositions). This development was by no means inevitable as is shown by most other Indo-European languages. In Latin and the Romance languages, it was only very pronounced because there were also further 'illogical' analogies (cf. Ne veniat! timeo 'He'd better not come! But I still fear it' => timeo ne veniat!), so that the mode itself was less and less noticed and only memorised schematically together with certain lexemes. And with every further analogous use this mode was further weakened.
The (governed) subjunctive, however, is not in the centre of our reflections, we are more concerned with the question whether grammatical morphemes or what we consider to be such, are weak signs per se. They are considered to be such only to the extent that further circumstances were added which then have led to a more and more schematic use of these little specific and old morphemes. There are only general probabilities, which have to do with the genesis of the languages (cf. tenses) or are based on further characteristics of these languages (syntactic liberties, cf. the emergence of governed cases, of a governed mode).
There is one further phenomenon which seems to be typical for Indo-European languages and which is, at first sight, similar to the types of government. It is the congruencies we think of, which are for example responsible for the fact that, in Latin, the number is always marked as far as nouns are concerned even in those cases when there is a more specific indicator of quantity in the same syntactic construction (tres homines) and also for the fact that attributive adjectives, too, always occur with a (congruent) marking of number (tres homines boni). Number, of course, is also a simple and very little specific morpheme, but this does, by no means, justify these high redundancies. Unlike it is the case with the past tense, they can hardly be explained by the fact that this morpheme used to have an especially relevant function, so that the marking was preserved even though this function lost more and more of its relevance. In this case, it is primarily the fusion of case and number which is responsible for the redundancies.
Such a morphological fusion can for instance be made by a phonetic fusion between the marking of number and case, if both markings occur regularly juxtaposed in a periphrastic construction (for example case –a + singular e > e; case a + plural u > o). It can however also be only a product of an interpretation. This is the case if we have to do with a 'non-symmetrical' marking, where one time only the case, another time only the number is marked so that in any case the number respectively the case is to be 'completed'. Consequently, if an -s only marks the nominative, but an -i the plural, and if this –i can only be the (unmarked) nominative, then the singular with reference to the plural does not behave as a neutral form with reference to an additional kind of marking: by using the form of the nominative, the speaker at the same time makes a decision for the singular, and if he uses the plural form he makes at the same time a decision for the plural (and would in this case perhaps use *-ei for the dative, -bhos for the dative in the plural. Even if this *-bhos (Latin turribus) consisted of two parts (of the dative and the plural), it would - in comparison with all the other plural markings - only be memorised as a unity of the dative and the plural!). By contrast, in Turkish, there is an additional form (-ler) for the plural, which can therefore also be omitted, if the context is already explicit enough as far as the number is concerned (in imitation: *three house/ three new house). Congruencies and the ensuing redundancies are by no means universal, but different in any single language and dependent on specific factors. Neither is the number, or rather the plural, per se a weak grammatical morpheme; it only becomes one under certain circumstances.
High redundancy and grammaticality are not merely founded on the 'weak' character of old morphemes or inflections, but they only become 'weak' morphemes in many languages because of further developments.
Even a weak conjunction with the meaning 'and' (formerly 'there, then') is not per se weak and could also be used much more economically (cf. the Chinese language!). But it is weakened to the extent that more specific words are used side-by-side with it in the same syntactic construction ('and then', 'and thereupon', 'and three hours later'). The original 'then, thereupon' (= and!) is restrained and degraded to a mere concomitant form. One can only ask oneself why this does not happen in all languages and, as it were, inevitably and therefore constitute a universal phenomenon (so that one has to assume a similar development sooner or later also for the Chinese language). One reason for this could be the fact that our inflecting languages have used right from the beginning many deictic signs and auxiliary signs (cf. the cases); thus, the increase in frequency could be explained by analogy. In addition, there might also possibly be a syntactic factor. Syntactic factors play an essential role in the development of the Indo-European languages, which we have already seen when talking about the development of the prepositions (from originally free adverbs). The conjunction 'und' can occur with more specific adverbs and adverbial phrases, and, in principle, such adverbs are syntactically free. At the beginning, they possibly didn't even occur right next to the conjunction 'and', but in some distance so that the redundancy was also far less obvious (…and he talked with her on politics, an hour later…). Such an 'improvement in retrospect' at any place in the sentence is only possible, of course, if the syntax has no strict rules, which smoothes the way for a later juxtaposition (…and an hour later he talked…), which, otherwise, would be felt to be a disturbing redundancy. This would be a hypothesis for an indirect way of developing redundancies.
Morphemes which we consider grammatical or to which we ascribe a high degree of grammaticality, have not been weak signs right from the beginning, but have become such only in the course of the history of a language. But there is a high probability that languages which use many unspecific signs or auxiliary signs right from the beginning restrain those signs later by using more specific signs, so that the older signs become concomitant signs or weak and schematically used auxiliary signs. For the isolating languages, this raises the question why they didn't need any such auxiliary signs which would have developed into redundant markings later on. However, it is not our task here to answer this question.
The 'weak' grammatical morphemes have not always been used highly schematically and redundantly, and this is certainly also true for the purely relative cases (nominative - accusative - genitive), which don't have any meaning of place (any more). Their increase up to obligatory use is connected with the fact that the syntactic rules of inflecting languages are not as strict as those of isolating languages and that the final position of the verb (pater videt/patrem videt) makes it more difficult to distinguish between subject and object.
In chapter six, we will treat the question why even newly developed grammatical morphemes and markings can become just as schematic and pleonastic-redundant as the old ones.
5. Schematic, pleonastic and secondary usage
The development of 'weak' grammatical morphemes which are used highly schematically and therefore also pleonastically-redundantly, has also always to be regarded against the background of the history of a language and is therefore by no means universal. The higher the number of such grammatical morphemes and markings, the more complex the grammar in question becomes, and this complexity increases considerably by the fact that these morphemes develop in turn a number of uses and norms, and the more intricate the net of markings, the more complex become the mutual delimitations and subnorms.
Such complex rules of usage and mutual delimitations derive paradoxically from the fact that grammatical marking is to a certain degree schematic, which can go as far as obligatory use, and this schematic way of using them is at the same time the expression of low regard for these weak morphemes which are often lacking in information and which are therefore used according to certain rules. We could actually expect that such morphemes follow very simple rules so that their performance is in direct proportion to complexity (low performance, simple rules). But especially in our inflecting languages, such an economic behaviour cannot be detected. We notice their high complexity at the latest when we start learning another Indo-European language. Even after years of intensive training, we don't manage to reach perfection.
The complexity of the grammar of inflecting languages is then paradoxically linked to the schematic use of grammatical morphemes, but, once more we cannot establish a direct proportional link between schematic and idiomatic use and therefore schematically used morphemes develop just as many sub-uses as those which are less schematically used. In other words: our grammars always remain highly complex and when, in some cases, they seem to simplify, they develop new and complex rules of usage elsewhere. We would like to illustrate this point briefly in the following:
We mark (schematically) the predicative verb with a personal ending or a subject pronoun (Latin canto, English I sing), which leads to the fact that such a marking is also used analogously in those cases in which the verb does not even refer to a person nor to an object (Latin pluvit, English it rains), so that it can't be an anaphoric usage either. And the (cataphoric?) pronoun in English it seems that…, it is good for you to… seems also to be totally without function. Analogously to the marking of the subject as far as the verb is concerned, the object pronoun became also more frequent, and, in turn, a conjunction developed from its abstract anaphoric use (I say that: he comes > I say that he comes), which might have been encouraged by a free syntax (He comes. I say that. > I say that: he comes).
Virtually a prime example of highly schematic and, at the same time, highly idiomatic usage is the (definite and indefinite) article: in the different Germanic and Romance languages, it has developed a number of usages which cannot be deduced anymore from the anaphoric function it used to have (cf. French l'amour - English love; French aller à l'école - English to go to church; French retourner pour le déjeuner - English to go home for lunch; French le lundi - English Monday; French le parlement décide - English Parlament decides etc.). The complexity of the usage of such a grammatical morpheme which carries only little information cannot be explained positively from its functions, but conversely, it has to be explained from the little attention which the speaker devotes to these weak signs, and which he therefore also uses analogously in cases where they don't convey any further new information. From one analogy to another, new norms and subnorms emerge, which, on the whole, leads to a complexity which a native speaker is not even aware of. The speaker compensates for this by memorising the collocations in question; and this exactly is the reason why he would have huge problems if asked to explain this kind of usage to another person. In this way, an increasingly schematic usage makes analogous transition easier and encourages the development of new idiomatic uses! An increasingly schematic usage leads also to the speaker manoeuvreing himself into a dilemma (which, however, is not experienced as a real problem): in the Romance languages, the imperfect is fairly easily and schematically delimited from the perfect by the fact that it expresses a state or an action in progress at a certain point in time (in contrast to the starting point of an action). This means, however, that there isn't any neutral form anymore, there is no past tense anymore which marks neither the starting point nor the state of an action. Nonetheless, there are contexts and situations, which cannot easily be forced into this basic pattern, so that the question arises which tense the speaker is supposed to decide on. Modal auxiliaries ('want, should, must') are such cases in which the distinction between the starting point of an action and the state is difficult to make, so that, here again, specific norms for the usage develop, though to a varying degree, even within the Romance languages.
The characterisation of a person doesn't fit very well either into the pattern: starting point of an action vs. state at a certain point of time, and therefore, the usage varies in this case in the Romance languages (cf. French N. fut/était un génie - Spanish Fue/era hermosa). Schematic usage on the one hand leads to idiomatic usage on the other hand: there is no neutral form anymore, so that, inevitably, there are also neutralising effects: both forms are, at least potentially, equal in value.
In many languages, a present perfect differs from a simple past tense roughly in that it describes an event, which - at least from the point of view of the speaker - is not followed by a further one, so that, in many cases, it is only the result which is to the fore (Kolumbus hat Amerika entdeckt; ich habe das Brot gekauft; er ist weggegangen). But not in all cases, this distinction can be drawn without problems, especially if a point of time in the past is mentioned or implied, therefore we ask in English: When did you buy the bread? whereas in Spanish we can also use the perfect tense: Cuando has comprado el pan?, cf. English I bought the bread in London (the place implies an indication of time) vs. Spanish He comprado el pan en Londres. And between the more or less fixed and obligatory usages of the perfect tense and the preterite, there is also a number of fluctuating usages and varying affinities in these two and other languages, which could also be treated as a matter of stylistics.
In French, we could try to distinguish a simple future and a periphrastic future (j'irai - je vais aller) on the grounds of temporal distance (distant future - near future), all the more considering that the periphrastic future is derived from an action which has already begun (je vais…). But such a pattern is soon to be dissolved again because the criterion of proximity and distance can't be determined objectively. What, for someone, lies within reach, might for others lie in the far distance. Again, more complex subnorms and varying affinities between certain situations and the usage of tense forms develop, including also (modal) nuances.
In the English language, the progressive form indicates schematically that an action is in progress at a certain point in time, especially when used for actions and events which can be interrupted (I saw him when he was going home). So, it is by no means only used to mark the progress of an action which otherwise wouldn't be discernible, but also in those cases in which the context in itself would be sufficient to convey the meaning (cf. other languages without any such marking!). This pleonastic and redundant use (cf. chapter six for its development), in turn, makes analogous extensions easier: pleonastics makes easier further usages: I am speaking the truth! You are telling lies! The child is constantly crying! or: The United States are saying that… And if certain subnorms become increasingly frequent we also talk about a linguistic development: in this sense, the grammars of inflecting languages are constantly in the process of progress or development; at least, many times more than the isolating languages, in so far as one can speak of grammar at all in those languages. The weak grammatical morphemes of our languages are often used schematically, approximately and analogously and not only in those cases when a certain indicator is indispensable in the context!
Schematic usages are per se analogous usages and this is why the concept of analogy plays such an important role in the inflecting languages. But also such grammatical morphemes which are used less schematically, that is not according to a more or less simple pattern, have analogous rules of usage including numerous secondary usages and subnorms. First and foremost, this is true for all categories which are still in the process of development (cf. chapter six) and which, therefore, haven't found a fixed place in the structure of the language, which are, so to speak, not yet firmly established. So, in most cases, they cannot be used schematically yet, according to a simple usage rule. This simple usage rule (with all its exceptions) is only the result of a settling process, of a gradual approximation of different uses, among which the speakers find a common denominator. And, as we will see later, this again has to do with the fact that new markings are not produced in an act of creation to express a certain signification or a distinction, but that, conversely, they are interpreted from the context, so that, right from the beginning, they only express pleonastically that which exists already as signification in the context.
A new tense, a new marking of aspect or an emerging article have by no means a fixed place in the 'system' of grammar right from the beginning, so that we had better speak of fragile norms and subnorms, and some markings also remain permanently fragile in the sense that they don't develop any fixed usage rules nor even obligatory usages. In contrast to the progressive form in English, the analogous forms in Spanish and Italian are still optional today (estoy cantando / sto cantando - 'I am singing'), and in contrast to the so-called partitive article in French (which has also taken over the function of the plural), the partitive article in Italian is only an optional variant. And just as, for all obligatory uses, one has to know especially all the exceptions and special usages, in this case, one has to know the norms and subnorms for all usages. This applies likewise also to the segmentation in modern French (les fleurs, c'est joli; moi, je travaille; je l'ai vu hier, ton frère). Grammatical morphemes and constructions are used approximately, schematically and analogously, and if they have once become obligatory, this does only mean that analogous usages will increase drastically so that it becomes more and more difficult to distinguish subnorms in this intricate network of usages. That is why the rule for the different uses becomes broader and more schematic, and this exactly leads to the development of obligatory usage. The actual question would now only be to know under which circumstances such increases in frequency have occurred.
But not every obligatory usage of a grammatical form can be understood as a mere balancing between numerous sub-usages and subnorms which goes hand in hand with an increase in frequency. Especially in the Indo-European languages, there are also fragmented and, at the same time, obligatory usages which contradict any idea of economic and simple language structures; we especially think of the so-called governed forms, that is governed cases, governed prepositions or the governed subjunctive in the Romance languages (French je veux que tu viennes). Such phenomena cannot easily be explained rationally either, because, in many cases, they occur by means of irrational and sporadic analogies: and these, in turn, are possible in those cases where the morpheme in question loses its separate meaning or passes it on to the governed sign. The governed morpheme becomes then the, so to speak, allomorphic concomitant form of the governing morpheme with which it is more or less exclusively memorised. A governed subjunctive is a mode which is still kept in use although it doesn't provide any new information in itself and has become redundant (Latin rogo te ut venias), this is why it is more and more frequently memorised with the governing sign, so that it can still be used even though the original sense has been completely lost (timeo ne veniat > non timeo ne veniat). And this is why, on the one hand, analogies occur in the history of a language to some extent schematically, if, for example, verbs with similar meaning (of desire, of fear, of hope) behave in some similar way (that is they also 'request' the subjunctive), on the other hand, however, they occur also sporadically and in some rather bizarre way, if no longer the signification, but the construction in itself becomes the trigger of the mode: in Latin, there used to be a subjunctive of anteriority which expressed the potentiality of the event in question (antequam de re publica dicam - 'before I can talk about the state'), but in the Romance languages, it was generalised as a concomitant form of this anteriority (French avant qu'il soit venu), so that the original condition (potentiality) was lost. The subjunctive became therefore associated with anteriority or in even more general terms: associated with temporal relations, so that it could later also be used analogously to express posteriority (après qu'il soit venu). While the fragmentation of uses still increases, the mode in itself loses more and more of its original meaning and is only memorised as a concomitant form of certain verbs and constructions. And exactly because of this demotivation and semantic emptying, it is very difficult to cut back its usage: the analogies are no longer linked with meaning (in contrast to the usage of the subjunctive in old English, for example, which was still more 'motivated'), they don't adopt the original meaning of this mode anymore but they develop sporadically and in a fragmented way, and if, in one place, the subjunctive is replaced by the indicative, the indicative will be superseded by the subjunctive in another place. For the speaker, both forms have long since become equivalent, just as allomorphic forms.
Likewise, this is also true for governed cases and governed prepositions (Latin in templo, ex silva, ante portas; English to think of, to worry about, to strive for, to aim at etc.). The more 'empty' such morphemes become, the less we can still speak of secondary usages. Consequently, the schematic usage of a morpheme which carries meaning turns into a schematic memorising with certain trigger signs; analogies are no longer 'systematic', but sporadic and mutually independent.
Grammatical morphemes are weak signs to the extent that they are used schematically, approximately and analogously, and not only when they are absolutely necessary. But to the extent that they are not only used where they are absolutely indispensable, the listener gets used to them and qualifies them in turn by devoting less attention to them. In many cases, they are only dispensable concomitant signs and often, they have to be interpreted anew with regard to the context (cf. e.g.: Meine Eltern haben mir neue Möbel gekauft. Sie sind sehr lieb.). In view of the constant usage of subject pronouns in English or in German, for example, metaphoric usages also attract less attention. If a nurse says to a patient: Wir gehen jetzt ins Bett!, we instantly grasp the meaning of this imperative. In the Japanese language, which only uses these pronouns when they are absolutely necessary, a literal translation of this sentence would lead to a grave misunderstanding! Conversely, redundant markings and constructions can also generate secondary usages if they are used without redundancies and have new meanings. A modal future (French il sera dans le jardin) develops, for example, if there is no explicit or implicit point in time in the future in the context which the future form could anaphorically refer to. What remains is the interpretation in the sense of a certain uncertainty or probability. Similarly, this is also true for the use of the future to express an imperative meaning in French (tu ne tuera pas! 'Thou shalt not kill!'). Such secondary usages are possible only in those languages which have a morphological and therefore pleonastic future tense, which can also be used side-by-side with a more specific adverb of future (tomorrow, I will go…), so that an 'elliptic' construction can again be used for other significations. In a language which only uses an adverb (e.g. 'later') instead of a future (that is for example, an isolating language) this adverb will never be used together with a further and more specific indicator of the future (*später morgen, *später in drei Tagen), so that in cannot be interpreted as anaphoric and neither as redundant! Though we expect a more specific indication in languages which have a morphological future, we don't expect the same in isolating languages. And this again has to do with the history of the inflecting languages, as we will see later on (chapter six): the future was not created by an act of creation to express future events and then used accordingly, but it developed from a construction, which, in a future context, was interpreted as a concomitant form of this indicator of future, so that the construction was pleonastic right from the beginning!
A modal usage of the conditional (ich würde sagen/ vorschlagen, wir bleiben zu Hause) is possible again, because the protasis, which we would expect, is missing (wenn es möglich wäre…). In an isolating language which only knows one conjunction expressing condition ('if'), this would not be possible at all (cf. in imitation: wenn Geld haben, dir geben - 'if I had money I would give it to you').
In languages with many redundancies, 'elliptic' constructions can be used in turn for new significations and stylistic effects. Thus, the lack of a copula is characteristic of aphorisms (omnia praeclara rara) or expresses astonishment (he tired - she tired), the missing of an object pronoun can mark a polysemy (he drinks).
Even the usage of an historical infinitive or an epic present tense instead of a past tense is a stylistic means which is only possible because the marking of the past would be pleonastic-redundant in the context. In isolating languages without any past tense, this would not be possible at all.
The French imperfect tense indicates that an action was in process when a new action set in (il sortait, quand je suis venu). But it can also be used literally if the context doesn't permit this usage (il sortait à 9 heures). This use becomes possible because of a number of 'secondary usages' of this imperfect tense (French Le trois mars, nous faisions une promenade. Nous sommes allés chez…), which were established before.
The grammars of our languages are full of examples of figurative or metaphoric uses. It seems to us that they have, to a great extent, a common root: pleonastic or redundant usages make it easy for analogous secondary usages to develop or, conversely, they give new meaning to 'elliptic' constructions.
All the factors which we have mentioned act in combination to make our grammar so complex and idiomatic and lead to numerous rules and exceptions which we don't have to expect in isolating languages. The many usages and sub-usages of grammatical morphemes in inflecting languages are characteristically different in any single language, but, in the end, they go back to the same elementary (grammatical) principle: pleonastic and schematic usage produces necessarily subnorms and sub-usages, and from such sub-usages new grammatical categories or morphemes can derive, which we will see in the following chapter.
6. New categories - new redundancies
There are languages which only have very few grammatical categories and markings, that is especially isolating languages, which are also very stable from a diachronic point of view, and other languages, especially inflecting ones, which have many grammatical categories and markings, which are also unstable from a diachronic point of view, because they generate new categories while, at the same time, losing others (e.g. cases or endings in general).
A lot of markings which are already very old are in many ways redundant, we only have to think of the cases or the prepositions, the number, the tenses and the personal ending as far as the verb is concerned, and besides, the Indo-European grammar has developed for thousands of years in so far as it has all the necessary markings at its disposal, which shows us also the comparison with other languages of the world which can do with few markings. If, within the framework of Indo-European grammar or grammars, we talk about development, this must not be understood in the sense of optimization: a language which has all the necessary at its disposal is already optimally equipped. Therefore, these developments were either shifts (from the cases to the prepositions or the syntax, from the personal ending to the pronouns), or substitutions (e.g. a synthetic tense by an analytical one) or the emergence of new categories and markings which had not been necessary previously, so that such developments could also be regarded as a kind of 'luxury' of our languages (new and additional tenses of the past or the future, aspectual markings, articles etc.). But, of course, 'luxury' does not mean that such categories could be eliminated easily later again: when they get established, they replace in turn other, older means and also develop more subtle new significations. But this does not mean that they were necessary, and therefore we cannot calculate whether they will develop and even less when they will develop, and we cannot predict precisely either how they will develop and thus get established in the language, apart from general probabilities which we will deal with later.
Up to now, we have also taken for granted that new grammatical morphemes and markings emerge at all, and that is especially those which are used with a certain regularity up to the so-called obligatory usage, and we have probably been rather surprised that they turn up more or less within the framework of that which already exists (new tenses, aspectual markings, pronouns).
We have already described the framework within which grammar moves universally in chapter 2: it is above all a matter of classifications and relations (also of temporal manner) which are conditions for the formation of a comprehensible sentence or text, that is for the cohesion in the sentence or in the text.
But we still have to explain why new morphemes and markings are used pleonastically again in a similar way, so that they, too, are not only used in those cases when they are absolutely necessary or when they are supposed to express emphasis, but again with differing regularity. This is the reason why we talk about grammatical morphemes in contrast to lexematic signs.
This new pleonastic usage is by no means obvious, and if we explain the development of an old tense by the fact that old auxiliary signs were simply necessary when, for example, there were not enough adverbs of time yet (e.g. gestern, morgen), and that these old auxiliary signs were still used when they had become redundant (gestern ging ich…), this explanation is no longer true for such tenses which have developed much later when the adverbs of time had already fully developed, and similarly, this is also true for markings of aspect (e.g. the English progressive form). One could expect especially of newly developing tenses that they are only used when they are really necessary or useful.
But since this is not the case and since the new tenses are used again schematically in a way that corresponds to the use of the old tenses, there is only one explanation: these new tenses are pleonastic right from the beginning, i.e. they mirror the contexts in which the signs occurred which were interpreted as a new tense. Thus, the new tense is not created to express something, for example a result (resultative perfect), but a sign which occurs in a context which describes the result of an action is interpreted as a concomitant form of this resultative action (*I have the bought book respectively I have bought the book). This resultative sense can, of course, also be expressed in a language with only one past tense, and that is by means of the context and the structure of the text: if no point in time is mentioned and no further action follows we can assume that the statement is still valid or that the action has consequences for the present (cf. Latin librum empsi, or Portuguese comprei o livro).
Thus, the new (analytical) perfect has not been created to distinguish actions which have consequences for the present from such which have no relationship to the present anymore. This was possible for a long time by means of the context. Perhaps it rather occurred in those contexts which already suggested the interpretation in the sense of a result, so that it was understood as an expression and, at the same time, as a concomitant form (!) of this signification. This development was encouraged, of course, by the present tense form have which, together with a past participle, refers to a past event.
The new tense is not used to mark a difference which otherwise would not have been possible to express, but it derives its meaning from the context in which it occurs, and therefore it becomes a concomitant form of such contexts, before, later on, it probably develops new usages (and 'emancipates itself').
That is why the pleonastic or redundant usage exists right from the beginning: the morpheme was not placed in a context to give it a new signification but, conversely, it was interpreted within the context. We will illustrate this principle with the example of a new future.
A second consideration follows: why do such new tenses not emerge in all languages in the world, or rather what circumstances favour the Indo-European languages? The answer is first and foremost of syntactic nature: the analytical perfect requires the existence of a syntactic construction consisting of a noun and an attributive past participle (*I have the bought book). If this syntactic construction is preserved while the contents change and become less prototypical (object + action executed upon the object) (*I have something planned), the attribute becomes the predicate as far as content is concerned (planned) and the original predicative verb (have, respectively I have) empties itself, but is preserved as part of an orthodox formation of a sentence with a finite verb.
Thus, this complex development also requires a complex construction which is preserved although it has long lost its original right (as regards content). The original predicative verb has only the (traditional) function to mark tense (present) and person, and this is why it is interpreted as mere concomitant form to the past participle of the verb (I have planned something).
This development could also be described as a process of semantic emptying, but it is no coincidence that, when a new tense emerges, it is always only the finite verb, that is the original predicative verb, which is emptied. But it has nonetheless still a function, because it marks the person and the tense and consequently, it preserves the tradition of Indo-European sentence constructions. There will never be a semantically empty lexeme! A very similar process is to be observed as far as the development of a new and analytical future is concerned (French. je vais chanter). The finite verb becomes the concomitant morpheme of the verb in the infinitive (chanter).
Again, specific syntactic conditions and developments are involved in this development. One condition is the fact that a finite verb is followed by a further verb in the infinitive. This cannot be taken for granted and it can only be understood in analogy to constructions with modal verbs (ich will/ soll/ muß gehen, or French je veux chanter > je vais chanter). A construction formed by verb + nominal form of a verb (ich gehe zum Singen/ Arbeiten/ Einkaufen), which we find in German, for example, can never become a new future, simply because the indication of the aim (… zum Einkaufen) makes sure that the verb keeps its meaning 'movement directed towards an aim'. Besides, otherwise the predicative verb would be missing. In our languages, a finite verb cannot become the future of a noun.
But thus, we only need a smaller 'mis-construction', that is once more a syntactic development, to deprive the verb 'to go' of its meaning 'movement directed towards a goal', to generate an 'empty' verb, cf. fictitiously: French. je vais en ville faire quelque chose > je vais faire quelque chose en ville. A little syntactic rearrangement is sufficient to turn the indication of direction into an indication of a place where something happens. The verb with the meaning 'to go' will lose more and more of its concrete meaning in such and analogous constructions, it will become the auxiliary morpheme of a new construction consisting of an 'empty' finite verb + verb in the infinitive. Consequently, complex syntactic conditions and developments are necessary to generate a new future. A development on a direct way, on the other hand, is difficult to imagine: a verb with the meaning 'to go', which expresses a movement directed towards an aim, cannot simply collocate nonsensically with other verbs (like 'to laugh', 'to cough' etc.).
Now we can see at the same time the reason why this new future also becomes a grammatical form, in the sense that it is used pleonastically-redundantly (je vais travailler demain): it was not created to mark the future tense in contexts which otherwise would be interpreted as present tense. It rather developed in contexts which already had a future meaning (je vais faire cela demain, je vais travailler en ville demain), so that the semantically emptied finite verb was more and more interpreted as mere concomitant form of such future events. And because this verb 'to go' originally indicated the beginning of an action ('I go, I am on my way'), this new future is associated especially with actions which will take place in the near future. Thus, it is the future context which is responsible for the interpretation of the morpheme as future, which therefore becomes a concomitant form of such contexts, before, later on, it can even be used without any other indicators of future (je vais le faire).
Thus, new grammatical markings are 'weak' right from the beginning in the sense that they adapt to the context, so that they become concomitant signs of this context.
The preservation of a construction in spite of the emptying of one of its elements is responsible for the creation of an adverbial suffix (French honnêtement). To the extent that, after the gradual decline of the cases, the ablative (honesta mente) was no longer understood as such, the adjective was interpreted as forming a unity with the little specific noun. And since the old noun (in the ablative case) had been pleonastic right from the beginning (laeta mente 'in a cheerful mind': one can only be cheerful in one's mind!), it could easily be used analogously (lentamente), whereby it was weakened even more until it was only understood as (mostly redundant) marking of the adverbial function. The decline of a marking on the one hand (case) encouraged the emergence of a new marking on the other hand.
Connected with the decline of the cases is the fact that the Latin bibere de vino (French boire du vin) was more and more interpreted as mere expression of an indefinite quantity. The partitive construction with de is interpreted as a variant of the accusative. The usage of this 'partitive article', however, is often already pleonastic-redundant insofar as, in most cases, the situation clarifies already that someone drinks only part of the wine which is available and not all of it.
Thus, often one could also say: N. trinkt Wein, and analogously this is also true for the plural (N. ißt Brötchen, French N. mange des petits-pains). Because of this weakness of the partitive article, an analogous usage is hardly noticeable (N. a visité des villes), even though, in such contexts, it is exactly not intended that someone would like to refer to a certain number of objects which actually exist (that is cities in this case). The partitive article and the plural have now become approximately equivalent, and, in French, this partitive article is later to replace the (indefinite) plural (une ville - des villes). A marking which had been slightly pleonastic right from the beginning expands (des villes) and ends up carrying on the tradition of the marking of the plural.
We have seen, so far, how new markings emerge when constructions are preserved formally, but have to be newly interpreted when one of the phenomena which are involved is emptied semantically so that it is understood as part of a signification, which, in turn, derives from the context in most cases, so that the new marking becomes pleonastic-redundant, and consequently, we speak of new grammatical categories!
But, the other way round, it may also happen that a pleonastic element is removed from a construction and newly interpreted. This requires, above all, a syntax which is not entirely rigid. This is how a new conjunction daß derived from an object pronoun: Du bist krank. Ich sehe das. > Ich sehe das: du bist krank. > Ich sehe, daß du krank bist. If the schematically used, anaphoric pronoun which does no longer refer to a concrete object at all, is used analogously also cataphorically, where it is even less useful, it can also be understood as a connecting link between both sentences and therefore as a conjunction, and eventually, it can be understood as the opening signal of the object clause, especially if there are further elements between verb and object pronoun (ich höre oft, daß er anständig ist). The distance from the verb encourages the sense of belonging to the following object clause.
In a similar way, the stressed and free subject pronoun in French could have developed from a usage of the dativus commodi, for instance: old French pense moi > pense souvent moi > pense souvent, moi. With increasing distance to the verb, the pronoun in the dative case, which isn't a 'real' dative at all, so that the pronoun can only be interpreted in connection with the verb (as 'intensification'), is more and more understood as a free additional element, and this free additional element is then felt as subject analogously to the personal ending (stressed, additional subject pronoun). Thus, it expresses, in a redundant way the function of the subject, which, in turn, is already expressed by the personal ending, so that the pronoun takes over an expressive and an emphasising function at the same time.
A conjunction to, which introduces the infinitive (he asked me to go), can be explained in simplified terms by the fact that an indication of an aim, which is originally linked to the verb, and even pleonastically with such verbs which already imply an aim (to ask to), is used more and more schematically (I want to go). Thus, they are weakened even more and finally they are more and more felt to be concomitant forms of certain verbs with a following infinitive. Their information tends towards zero, they become redundant concomitant forms. It is the free syntax which is finally responsible for the fact that anacoluthic constructions are less noticeable, too, so that the remaining constituents in the sentence are interpreted as a new syntactic unity. This is also a way in which a new grammatical category develops, cf. schematically: He was in the church, singing and praying. > He was, singing and praying, in the church > He was singing and praying and asking God… The emphasis is only on the action itself, the place of the action moves to the background. If such expressive constructions increase, a progressive form develops, which emphasises the action and its progress. The originally local verb (to be) becomes the copula of the verb in the gerund form. Again, the new (aspectual) marking is pleonastic right from the beginning, because it adapts to the context (anacoluthic construction, concentration on the verbal action). It can develop further from this basic pattern, so that also more specific rules for its usage emerge, but it always keeps its pleonastic-redundant character, because it is not only used to distinguish the progress of an action from a sequence of actions ('distinctive' function!), but also in those cases in which the context in itself is enough to avoid ambiguities. Thus, in many cases, the progressive form is only a concomitant form of a verbal action in progress. When a foreigner whose English is not very good doesn't use it, he will still be understood in most cases. In English, to do has become a similarly intensifying periphrastic construction (he does go home). Syntactic condition is again the construction consisting of a finite verb + verb in the infinitive; if the causative (to do) is emptied, the auxiliary verb will be considered as a concomitant morpheme of the verb in the infinitive (he does go), so to speak as an analytical marking of person and tense of this verb in the infinitive. This little specific periphrastic construction became later, with increasingly schematic usage, the periphrastic question and the periphrastic negation (do you go? I do not go). But, this last step of grammaticalization towards obligatory usage leads also towards the complete devaluation of the form.
We hope we were able to illustrate with the preceding examples how, especially in the Indo-European languages, new grammatical categories and markings develop again and again: A variety of forms is required (e.g. participle or gerund, past participle, different stems for 'rectus' and 'obliquus' as far as pronouns are concerned: ego - me), with a number of usages and sub-usages (e.g. dativus commodi), as well as the existence of constructions which also form a morpho-syntactical unit (finite verb + infinitive; congruency of the case in laeta mente), and in addition a flexible syntax. Thus old constructions can be dissolved and new ones can develop, or existing constructions can be re-interpreted as a new unit with a new signification while keeping an element which has only a syntactic justification (je vais aller, I do go home).
Languages with less morphological markings and less complex and intertwined constructions, especially isolating languages, have therefore little which could be dissolved or newly interpreted, and languages with a rigid syntax don't permit rearrangement and new classifications either.
We can ask ourselves now what these considerations have to do with our topic ('Universality of grammar, grammatical universals'). The Indo-European forms and categories are not universal, of course. But there are universal conditions on which new grammatical categories and markings can develop, that is those which are used highly schematically and pleonastically, just the same as one can state universally which functions and relations can be expressed by morphemes (cf. chapter two).
Two further questions follow, which we will only treat briefly: a) how integrated and obligatory can grammatical morphemes become, and b) can markings also disappear again, are grammaticalizations reversible?
The first question is already so complex that we can only mention briefly some aspects here: the question of integration in the 'system' of grammar (in structuralist terminology) seems to be especially dependent on one circumstance: can we find a simple rule according to which the grammatical morpheme in question can be used? This is not a problem as far as abstract indications of function (nominative case and accusative, genitive) and significations (conjunctions, relative pronouns) are concerned, but this is not the case to the same extent as far as morphemes are concerned which also convey additional meanings (we only have to think of the usage of competing tense forms, for example future tenses, or aspectual markings with their nuances). More subtle differences and nuances don't permit any schematic usage per se, because the speaker has to make additional decisions. Now, we could argue that such more subtle differences will disappear again when the usage becomes more and more schematic. The additional decision of the speaker is then limited to a minimum of considerations, as such a schematization of the usage can also be observed everywhere in the field of grammar. But it also comes up against limiting factors which are outside the morpheme in question, which has to be delimited and established within the 'morpheme field'. If the paradigmatic surroundings are already occupied, because there are already other morphemes at our disposal for similar purposes, it becomes more and more difficult to find a simple and schematic usage rule for the new morpheme. A partitive article which assumes the same function as the (indefinite) plural, will probably also develop usage norms, but they will always remain fragile, unless the older plural is replaced by this new form (as in French, in contrast to Italian, for example). This is true in a similar way for the future form in English and in French (e.g. futur proche and futur simple), or the synonymous possibilities to ask a question in French (Avez-vous…, est-ce que vous avez…, vous avez?) and quite a few things more. A lot of sub-usages and affinities (of tense to person, to adverbs of time etc.) develop in turn from this problem of mutual delimitation, the fragility (also diachronically) of which only proves once more that these 'stylistic' differences are by no means necessary and, at the same time, that the new grammatical morphemes and markings arise by no means from a need, but that they are due to coincidences in the history of a language. In this sense, the new categories and morphemes are also 'luxury phenomena'!
The difference between fully integrated and thus also obligatory markings and rather free or stylistic ones can very well be observed if we compare the English progressive form with its equivalents in Italian and in Spanish (I am singing - sto cantando - estoy cantando). The difference is surely founded on the conditions of these languages: the English form does not meet with an already established form of similar content and can expand without any obstacle. In the Romance languages, however, there exists already an imperfect tense which expresses the state or the progress of an action. It is therefore extremely difficult for the aspectual form to get established; it remains a stylistic variant of this imperfect (cantava/ cantaba - stava cantando/ estaba cantando) and this is why it does not expand in the present tense either, where the distinction between state/ progress vs. beginning doesn't play any role. In English however, the analogy to the usages in the past tense might have played a decisive role.
These brief indications may be sufficient here, so that we can turn to the second question: can grammatical markings also be cut back again or are they obligatory and fixed in so far as they can only be replaced by other markings. In other words: why can that which has once emerged not be replaced again? What, in the end, is the reason for this typological irreversibility, which shows that isolating languages will also remain isolating in the future and that inflecting, that is highly redundant languages will also remain pleonastic-redundant in the future?
This question has only been asked in passing so far, and this is why it was always taken for granted that some markings are of a more fragile nature (e.g. the marking of the adverb by -ly in English), and others are of a more stable nature (marking of the person, and also the tense, as far as the verb is concerned). We don't want to claim that we are able to see all possibilities of the languages of the world or merely of the Indo-European languages, but we would like to put forward a hypothesis which is pragmatically founded. It goes as follows: only that can be removed again which can be deduced from the context without any marking, without there ever being any ambiguities and misunderstandings. Thus, the marking of singular and plural is irreversible, because in I see the house. the singular cannot be taken for a plural and in I see the houses. the plural cannot be taken for a singular, so that any 'wrong' usage leads instantly to misunderstandings; in contrast to the Japanese language, we can no longer interpret flexibly and neither does the context give any explicit information as far as the number is concerned.
Neither can a tense of the past be removed again: Der Kaffee war kalt. is simply different from Der Kaffee ist kalt. We don't have any neutral verb forms anymore as far as tense is concerned (for the predicative verb), thus, the listener does no longer interpret with regard to the context, but orientates himself by the grammatical morpheme. Connected to this is the fact that a copula, too, at least in the past tense, cannot be cut back. There is no marking of the past tense other than the verb in the Indo-European languages.
The progressive form in English, however, was by no means necessary and it also developed only very late, but a potential reversal of the development including a more general usage of the simple form would also lead to misunderstandings.
Thus, the underlying hypothesis goes as follows: redundancy before risk. The speaker prefers to stick to a distinction before risking misunderstandings; therefore, the distinction between a definite and an indefinite article is hardly reversible even though it was hardly necessary at the time of its genesis; this is also the case because we have given up alternative possibilities of former times because of this development (cf. Latin homo quidam 'a certain person, some - not yet mentioned - person'!). Some usages of the article may fluctuate, but the basic distinction between 'mentioned - not yet mentioned' will be kept.
Thus, there is obviously no direct proportion between redundancy and fragility or, in turn, between information and stability: even few possible ambiguities are enough to preserve the category in question and to avoid its removal.
Thus is also preserved the marking of the person as far as the verb is concerned by means of an ending or a subject pronoun, which has the function of word formation as well as a grammatical function (marking of the subject). The schematic marking is economic for the speaker in so far as he doesn't have to decide constantly whether the marking is necessary in the context. The economy consists paradoxically in a pleonastic, thus not economic usage!
Thus, even such markings are preserved which are amalgamated with the lexeme, so that a reanalysis is no longer possible, cf. the adjective formation in French honnêtement. The removal of the suffix leads to a feminine form of the adjective, which would not be understood at all anymore if there was a sentence with a masculine subject (***Son père travaille honnête). This is the fundamental difference to the adjective formation in English with the suffix -ly, which uses adjective forms which are neutral as far as gender is concerned. Only such markings and distinctions are thus reversible which express purely pleonastically that which is already clear from the context without them. Even if there is no resultative perfect one can recognise the resultative character of a verbal action by the fact that no point in time is mentioned and that there is neither a succession of actions (I won the price). Such temporal distinctions could then also be removed again, and this is also true for further temporal distinctions, for instance between different future forms, which are, above all, of a stylistic nature, so that no ambiguities occur if one form replaces the other. Analogously, this is similarly true for pleonastic conjunctions (und, daß). Such purely pleonastic markings are by no means to be removed in any case, but they can be preserved for a very long time, as is shown by many languages. Therefore, the term economy only offers very few explanations for the area of grammar (and also of pleonastics and redundancy!).
Conclusion:
In this piece of work, we have tried to explain in which functions and at which positions in the sentence or in the text grammatical markings, and also morphological markings can occur universally, and why, almost inevitably, they become pleonastic-redundant. These are, above all, functions in the sentence or in the complex sentence and (also temporal) relations, which the speaker always has to know if he wants to produce a meaningful sentence or text, so that the little additional effort for the marking isn't too much trouble either. The use of such markings, which are expression of a previous planning, becomes a schematic habit, so that they are used almost 'automatically', without considering the question whether they are absolutely necessary in the context or whether they are only pleonastic-redundant.
The concept of grammar which we have developed here, and which we put up for discussion, is primarily based on observations which derive from the comparison of isolating (and economic) with inflecting languages.
At the same time, it became clear how extremely complex our Indo-European grammar is, and how it is also connected to word formation. One further characteristic of this grammar is the syntactic freedom which, on the one hand, encourages the development of morphological markings right from the beginning (in contrast to isolating languages with inflexible syntax) and, on the other hand, also facilitates re-interpretations and shifts in structure and thus the development of new grammatical categories and markings, including the transfer of morphological markings onto syntax (cf. the disappearance of the cases in English or in the Romance languages).
Thus, this kind of grammar also remains in motion from a diachronic point of view, and where, on the one hand, markings disappear, on the other hand, new categories and markings appear. For different reasons, the development towards an economic and isolating language is excluded.
In the grammar of an inflecting language, the term development must therefore not be understood in terms of an optimization of the 'system'. Such developments don't pursue an aim either, but they consist in the very schematization of a usage (also to the detriment of another one, cf. replacements), so that we can ask ourselves at the very most what has caused this development.
Our concept of grammar sees itself first and foremost as a functionalism which is historically founded, but which does not look for a deeper meaning in all places, but which emphasises the high relativity of functions and differences especially in the grammar of our languages.
We hope to have given stimuli for further discussion.
(Translated by Ute Eckstein and Ulrike Schmidt).