MA Liguistic Theory Topic One Introduction I call my part of this - - PDF document

MA Liguistic Theory Topic One Introduction I call my part of this lecture 'methodology' since I have not been able to cook up a better title. The topics I wish to discuss will be about the shape and making of grammar, ie. how to make grammars, what are the ingredients of a grammar and what approaches there are to language and, finally, how we argue in grammar to defend a particular analysis. A) What is the object of linguistic analysis In t.he first part we will distinguish between (1) linguistic phenomena, (2) linguistic data and (3) linguistic facts 1) linguistic phenomena a) results of speaking/writing b) speech situations c) speech/writing activity and its perception/understanding

• utterance ('megnyilatkozás'): the basic unit of linguistic phenomena, the result of

speaking/writing activity

• utterances are observable: the result/the situation/the understanding empirical method

2) linguistic data

• are only a set of the linguistic phenomena
• what counts as data depends on the purposes of the research, and this restricts the phenom-

ena since it is impossible to examine all linguistic utterances from all possible aspects at the same time. The aspects of linguistic research that forces the grammarian to choose the rele- vant aspect from the linguistic phenomena are the following: i) phonology researches the sounds of a particular language or dialect; it may also study speech sounds in general of many languages to find regularities between them ii) morphology studies the structure of words that are built up of speech sound segments or phonemes iii) syntax concentrates on how words are organised into a linear sequence to create mean- ingful elements iv) semantics attempts to find meaning relations between different structures of various complexity v) pragmatics examines how language is used in actual situations, how use determines/ modifies grammar, in other words, the interaction between grammar in the abstract sense and the situations in which language is used These are large fields and each can and must be further restricted. For instance, if we want to describe the Hungarian word order, we will only examine the phenomena that are relevant for this research, ie. we will not be concerned with the phonological makeup of the utterances we are examining. What has been said presupposes that we have hypotheses about what there is in language, we have a theory about language, we have concepts that we use when we change the phenomena into data and leave the rest in peace. In other words, before we set out to study any aspect of linguistic phenomena, we have to have firm ideas about language. 1

3) linguistic facts A linguistic fact is a general statement about a linguistic phenomenon/group of phenomena,

• ie. it is an empirical (= based on observation) generalisation. A linguistic fact is a rule of
• grammar. (Example 1)For example, it is a rule of English grammar that in clauses where there

is a verb which shows present or past tense there should be an explicit subject even if it is

• meaningless. Put differently, finite clauses have subjects irrespective of their meaningfulness.

For example, The dog/He bit the new postman. Both clauses have an explicit, meaningful subject: the dog and the personal pronoun he. However, in the clause It was raining heavily the explicit subject, the personal pronoun it does not refer to anything/anybody, and from a semantic point of view it is a superfluous element. But the clause *Was raining heavily is

• bviously ungrammatical, even if it contains all the necessary semantic elements. (The same

applies to what is widely known as existential sentences: There were three spiders on the wall vs *Were three spiders on the wall.) To arrive at the above analysis we need to have concepts

• f subject, meaning, personal pronoun, clause structure etc., that is, the linguistic phenomena

do not explicitly show these grammatical concepts, they only offer some acoustic phenomena. (Example 2) Adjectives referring to size can be paired off: tall-short, big-small... -an obser- vation which is a fact of language. Further, it is also observable that one member of each such pair is 'unmarked' (the positive) while the other is 'marked' (the negative). Consider the fol- lowing sentence: John's house is as big as Mark's. This clause does not inform us about the absolute size of either house: both can be big and small. However, the clause John's house is as small as Mark's implies that both houses are small. Markedness, a notion that can be ap- plied to various linguistic phenomena, induces a presupposition. That is, the negative adject- ives presuppose some absolute degree of the characteristic they describe. Also related to the markedness of negative adjectives is their use with measure expressions. It is quite natural to say that John's house is two metres high but not that John's house is two metres small. To summarise the examples: the use of the marked forms always requires some specific circum- stance, either linguistic or situational. As is also shown by the above discussion above, lingu- istics facts are more abstract than linguistic data and require a more sophisticated theoretical

• background. If we wish to study the semantics of the English modal auxiliaries, we will need

logical concepts, such as 'epistemic' and 'deontic'. Consider the following clause. Heavy win- ters can cause serious problems on the main roads. This sentence implies that we have some 'knowledge' (Classical Greek ðéóôçìç (epistémé)) about heavy winters and their consequen-

• ces. So, a clause with the modal auxiliary 'can' expresses what logicians traditionally call 'e-

pistemic statement', a statement whose truth depends on our knowledge of the world. Con- sider another clause with the same modal auxiliary. We can use the bathroom on the fourth

• floor. This clause illustrates a different meaning type of the modal auxiliary 'can', which is

referred to as 'deontic'. The second 'can'-sentence means that we are allowed to use the bath- room because the regulations make it possible for us. The expression 'deontic' is related to

• bligation, what must and can be done. (It comes from the Classical Greek verb äÝù ( deó) 'I

must/have to'.) Linguistic facts, that is, grammatical rules or generalised statements concerning various linguistic phenomena can be falsified, ie. can be found incorrect or inadequate or not true. Three cases may happen: either (i) we made the wrong generalisation or we (ii) were not careful enough and did not examine all the data that are relevant. Also, it might be the case that (iii) we used grammatically incorrect data. As to the first problem. (i) We would like to e- xamine the English tense system and we find that there is only present and past tenses in Eng-

• lish. Therefore, we conclude that there is no expression of futurity in English -an obviously

2

false generalisation, which confuses tense with time reference. Tense is a grammatical, while time reference is a semantic category. It is true, there are only two tenses in English -present and past-, referring primarily to present time and past time, respectively, but there are ways to refer to future time though not with the help of a morphologically coherent future tense mark- ing affix but with seven different structures, some of which may also have present time refer- ence as well. (ii) We may make a statement that in English all clauses must have explicit sub-

• jects. However, this statement/rule is not true because we did not examine non-finite clauses.

In finite clauses the presence of a subject is obligatory but in some non-finite clauses the pres- ence of the subject is impossible while in others possible, again, in a third type subject is ob-

• ligatory. So the above generalisation can be easily falsified on the ground that we did not e-

xamine all the relevant data, not all possible sentence types. (iii) The third case: the gram- matical rule is false because we did not use grammatically correct data. Hungarian students tend to use the English word 'own' following the distributional pattern of the Hungarian e- quivalent of this word: *He has an own house ('Van egy saját háza') We could say that many Hungarians use ungrammatical data: they apply a rule of their mother tongue to English, so they produce an ungrammatical English sentence, instead of the correct one: He has a house

• f his own. (I know that this last example is not the best: a Hungarian student of English is

not an English grammarian.) To summarise this section: we made the threefold distinction between linguistic phenom- ena (whatever is uttered or written), linguistic data (a part of the utterances/phenomena which are relevant for the research we are conducting) and linguistic facts (which are generalised statements about linguistic data, also known as grammatical rules. B) Collection of linguistic data In this section I will discuss where grammarians find the linguistic data that they will study in

• rder to make general statements (ie. establish grammatical rules). Linguistic phenomena are

an integral part of our lives, however, not all utterance types are used by grammarians. (i) Traditional grammar used printed literary works as source for their investigation. Sour- ces of linguistic data are called 'corpus', the plural: 'corpora' (as in the original Latin). For in- stance, most Hungarian descriptive grammars take their data from well-known Hungarian lit- erary works. The idea is that a printed text is more reliable for analysis since it is the result of some deliberation, so final, and also there are full sentences in literary works (mostly), unlike every day language with its sentence and phrase fragments, therefore, research is easier and the results are reliable. Unfinished, fragmentary utterances cause problems for analysis. Also, there has been another view of literary works that promoted them to the object of linguistic

• study. Prescriptivism as well as purism influenced the choice of the best pieces of Hungarian

literature in the hope that those works give the best and purest examples of the Hungarian language, therefore, the language of literature is worth studying because it also sets an exam- ple for the socially prestigious (also called 'correct') use. (ii) More modern schools of grammarians (eg. structuralists) tend to rely on introspection for data, that is, they rely on their own linguistic intuitions and create the necessary structures, which will feature as data in such grammars. In other words, their data do not represent utter- ances but sentences, that is, structures abstracted from situational utterances. The grammar- ian's intuition is a good source for the relevant data, and also the grammarian decides which form is grammatical and which is not on the basis of his/her knowledge of the language. An-

• ther point is that a written grammar also shows the potentialities of use. In contemporary

corpus-based researches (and this type of data collection could be point (iii)), however, a sit- 3

uation can easily appear when there is no relevant data in the corpus which can illustrate a particular language problem/use, which the grammarian would like to analyse in his/her

• grammar. Such missing data can be provided by the linguist's intuition. For example, if a

grammarian studies quantifiers (quantity expressions, such as 'all, many, each, none' etc.) it is possible that no corpus will provide data, such as All the boys danced with all the girls or Two boys love a girl or Two boys love one girl or Everyone knew some things etc. Interim summary. In this section we started (but have not finished yet) to examine the sources that grammarians can take their data from. The traditional source was literary works, later grammarians used unattested/created data according to their needs, and with the advent

• f modern information technology large linguistic corpora came into being and these also

provide sources for grammarians. Such IT based research is called corpus-linguistics. Literary sources can be criticised for not representing everyday language use, the problem with gram- marians' intuitions is that in the heat of argumentation linguists might create structures which do not either actual or potential use or simply incorrect. Corpora, thirdly, may not contain enough data for the grammarian to exhaustively study a particular phenomenon. (iii) Historical linguistics can only rely on written corpora, and it has no other choice: no native speakers are alive from, let us say, the Middle Ages, no recorded utterances, films etc. are available from the pre-twentieth century world. Linguists cannot check their grammatical hypotheses with native speakers. Traditional grammarians who use literary corpora as sources for study of a particular language, in fact, follow the steps of historical linguists. This practice

• f traditional grammarians was related to what is called the 'classical fallacy' in the linguistic

literature: on the one hand, grammarians thought that written language was primary in relat- ion to spoken language because to produce a piece of writing require a lot of deliberation and care while spoken language is careless, therefore, it reflects the grammar of a particular lang- uage to a lesser extent than the written language. On the other hand, grammarians believed that writers', especially classical writers', language was better, purer, more perfect than the everyday language so literary language is worth studying. Further, an approach to language which is widely known as prescriptivism also grew out of these fallacies: the view that the role of a grammar is to teach people to correctly speak, the job of the grammarian is to show the correct forms of a particular language, to weed out forms and usages which are incompat- ible with the language of the classical writers. The radical change came in the twentieth cen- tury, with the advent of modern structuralism. This approach claimed that spoken language is primary, therefore, ordinary people's every day language should be studied, and literary cor- pora and prescriptivism have no place in a scientific description of language. Also, since spoken language was considered as primary, reference to the history of language in explana- tions lost its relevance. The structuralist approach to language had a consequence on what counts as corpus: linguistic phenomena were collected from speakers rather than from books. Currently, there are large data bases available for language study. Some words about prescriptivism are in order. Prescripitivism, ie. the practice when gram- marians tell people which forms of their language are correct and which ones are incorrect, basically also relies on descriptive practice. In general, a descriptive grammar shows us what there is in a language; however, a grammatical description can only analyse noun phrases, or the semantics of adjectives, or the relationship between word order and meaning etc.. In other words, what we call 'descriptive grammar' does not necessarily give a wholesale picture of a language, it rather means a particular, impersonal, objective approach to certain areas of a particular language (irrespective of the corpus). It is a characteristic feature of all language that one and the same phenomenon can be expressed in more than one way. Some of these 4

competing varieties take on negative social value, and a prescriptivist advises you not to use these socially stigmatised varieties. Consider the following Hungarian sentence pairs: A la- kásba tarcsák a farkaskutyát vs. A lakásban tartják a farkaskutyát and Ezek a könyvek meg- találhatók az intézeti könyvtárban vs. Ezek a könyvek megtalálhatóak az intézeti könyvtárban. In the first pair, examine the verb forms tarcsák vs. tartják. What is the difference? From a communicative point of view both forms are all right, both sentences convey the same mean-

• ing. These are two forms which are widely used by native speakers, they are both part of the

Hungarian language. So both are correct. However, the tarcsák-form is socially stigmatised, it is not incorrect, it is only associated with speakers who we do not want to be identified with. And this social aspect of certain language forms is what is a battleground for a prescriptivist. Consider the other pair. Here we find a choice, too: megtalálhatóak vs. megtalálhatók. This difference, however, is not freight with any social content or stigma; perhaps people would not even perceive that there are two ways to say the same thing. So in language there are grammatical/phonological varieties of expressing one and the same meaning, for some un- known reason, some of these varieties will receive some social value, others will not. What the prescriptivist does is, in fact, nothing to do with grammar, a prescriptivist gives advice how to behave in a socially acceptable way. Also, prescriptivism is useful when one teaches/ learns a foreign language. It is fairly obvious that when teaching Hungarian to international students, we do not want to teach them to say A lakásba tarcsák a farkaskutyát or as learners

• f English we do not expect our teachers to teach us some socially stigmatised form of the
• language. For instance a form such as I bought the book what was expensive. To summarise

the paragraph on prescriptivism. The competing grammatical/phonological forms in a lang- uage perform the same communicative function: neither is more correct that the other. How- ever, most varieties are associated with some speakers belonging to some social layer and the 'correctness' or 'incorrectness' of the varieties depend on the social status and prestige of the people who use it. People who use the -suk/-sük and -szuk/-szük-forms are mostly uneducated,

• ie. they have little social prestige, therefore, if someone uses this variety will also have little

social prestige. That is basically behind prescriptivism. ...................................................................................................................................................... T2 MA Linguistic Theory Explanation in grammar Introduction An important element of a grammatical discourse is explanation. Once certain facts of gram- mar (ie. rules) are established grammarians are interested in the 'whys' and 'wherefores' in very much the same way as is natural in all aspects of life. For instance, a child may ask why it has to stop when the traffic lights are red. The explanation in this case refers to situations in traffic and personal safety. Explanations, however, become more abstract when someone wants to know why the sun rises every day and in many situations explanation consists in the explication and detailed analysis of the particular phenomenon or in finding other types of phenomenon which are thought to be similar to the one we wish to explain. In the following I will be trying to give some examples of the possible types of explanation. Type 1 A linguistic fact is explained with reference to (an)other linguistic fact(s). It is generally suggested that active and passive clauses are synonymous in the sense that 5

they both identify the same situation. (The choice between them reflects communicative pur- pose rather than grammatical rules.) John broke the window vs. The window was broken (by John). In grammars the passive construction is explained with reference to the corresponding (= synonymous) active clause, which is considered basic while the passive the derived struc-

• ture. The explanation concentrates on the change in the verb form and the possible suppress-

ion of the subject noun phrase. Whether students of grammar are satisfied by such an ex- planation is another question. Type2 A linguistic fact is explained by translating it into another language. For instance, such and such a grammatical fact is so in English because it is of the same structure as its Hungarian equivalent. Type 3 A linguistic fact is explained by stating that it is a consequence of a general linguistic fact. A general linguistic fact is a statement which is supposed to explicate some characteristic of a particular language. Such general linguistic statement are the result of the study of specific features of many languages. For example, it is a general linguistic statement that there are sentences/clauses which describe reality (as it is) while the other type of clause does not de- scribe reality: they refer to wishes, future events, conditions, orders etc.. It is generally claim- ed that reality describing sentences can undergo logical processes, that is, (i) they can be questioned, (ii) they can be negated, (iii) they can appear in implications, (iv) they can be co-

• rdinated. I grammatical terms, we can say that clauses describing reality can be subjects to

various syntactic processes. This general linguistic statement is in the background of the di- stinction between three types of adverbial categories in English (and Hungarian as well): ad- juncts, disjuncts and conjuncts. Disjuncts and conjuncts make clauses/sentences non-reality describing therefore they do not undergo the syntactic processes that reality describing ones

• do. In other words, whether a sentence describe reality or not has syntactic, ie., grammatical
• consequences. So, adjuncts, unlike disjuncts and conjuncts, are closely integrated into clause

structure: they can appear in the (a) focus of cleft sentence (It was because of his injury that Hilda helped Tony), in a structure called (b) alternative negation (Hilda didn’t help Tony be- cause of his injury but to please his mother), also in (c) alternative interrogation (Did Hilda help Tony because of his injury or to please his mother?), they can serve as (d) focus of other adverbs (Hilda helped Tony only because of his injury), they can fall (e) within the scope of ellipsis (Fred carefully cleaned his teeth but Jonathan didn’t carefully clean his teeth), they can appear (f) within the scope of pro-forms (Fred carefully cleaned his teeth and so did Jonathan), and, finally, they can be (g) elicited by question forms (Why did Hilda help Tony?) On the other hand, disjuncts (also known as sentence adverbs) are not integrated into the clause structure in the above sense; from a semantic point of view, they express the speaker’s attitude to, and evaluation of, the sentence (Frankly, Mr Foster neglects his children/Briefly, there is nothing more I can do about it/If you ask me, he was drunk). The third group of adverbials, conjuncts, are not integrated, either, into the clause structure in the above sense; they connect two (or more) linguistic units and express what logical connection the speaker thinks holds between them (It was snowing and, in spite of this Mona went cycling/I tell you in addition that she has written ...). To summarise. General linguistic statements help us to i- dentify three distinct categories of adverbial function in English, and the truth of this abstract statement is borne out by the syntactic tests shown above. 6

Type 4 Typological explanations. Typological explanations are also a sort of general linguistic statement. Typological studies attempt to find what are called implicative universals, that is, regularities in languages where

• ne phenomenon presupposes the presence of another phenomenon. For example, if there are

voiced obstruents (b, d, g) in a language, there are also voiceless obstruents (p, t, k); in other words, the presence of one class of grammatical or phonological phenomenon presupposes the presence of some other. Or, if there is inflexion in a language, there is also compounding. Or, if a language has grammatical gender, the grammatical category number also exists there. Or, if the basic constituent order of a language can be characterised by the following order: subject-complement-verb, then that language has postpositions (rather than prepositions). Such typological implicative universals are the result of observations of many languages and they only work in one direction: eg. if there are voiceless stops (p, t, k) there may not be voiced stops (b, d, g) in the language; etc.. The idea of explanation here is that the reason why a particular grammatical category/phenomenon exists in a particular language is because it is an exponent of a universal rule. (Of course, we cannot explain why we have that universal rule, ie. why languages have the rules they do). Type 5 Historical explanation A type of explanation which used to fashionable in the nineteenth, early twentieth centuries is to relate the present day structures to their historical antecedents, ie. to earlier forms of the same language. For instance, grammarians felt that they had to explain certain irregularities, such as irregular plurals in English, such as goose-geese, foot-feet etc., also known as 'muta- tion plurals' and irregular past and past participle forms, such as see-saw-seen. The explana- tion of the first type of phenomenon, irregular plural, goes back as far as the period of English commonly referred to Old English -a misnomer since there is very little in common between Present Day English and the dialects spoken in the British Isles between 500 and 1100 AD; (some people call it Anglo-Saxon). In Old English (OE) one type of plural inflection contain- ed the /i/ phoneme, which cause regressive, partial assimilation, characteristic of the Ger- manic languages, called Umlaut in German philology. This assimilation caused the fronting

• f a back/velar /o/ phoneme: /fo:t/ + /iz/ > /fö:tiz/ > /fe:ti/ > /fe:t/ > /fi:t/ <feet>. The historical

process consisted in the following steps: the addition of the plural suffix caused the velar vowel to change into a front vowel (/fö:tiz/), then the plural suffix disappeared, which even- tually resulted in the contrast between the singular and plural forms: /fo:t/ vs. /fe:t/ in OE, and the long mid-vowels became high vowels (fu:t, fi:t/ before Shakespeare's time during the per- vasive sound change which is known as Great Vowel Shift. What ever grammars call irregul- ar past tense and past participle forms go back to an even earlier period: most of these forms are survivors from the ancient language which was supposed to be the language from which most European languages come: Proto-Indo-European (PI E). PIE is a hypothetical language, no documents exist from that era, but the similarity between many European (Germanic, Ro- mance, Slavic) and Asian (Indo-Iranian) languages suggest that they must have had a com- mon ancestor, and this ancient 'mother language' is PIE. Historical explanation however in- tresting are not preferred today any more. Type 6 Theory specific explanation Consider the following sentences: Mary persuaded John to buy a new car. Everybody who understands English will confidently explain that Mary did the persuading while John will/ would (probably) buy a new car. In more a technical language we would say that the NP 'Ma- 7

ry' functions as subject of the main clause, while the NP 'John' performs the function of direct

• bject, and the way the sentence is put together indicates that 'John' is also the subject of the

non-finite subordinate to-infinitive clause. The speakers' intuition that 'John' is the subject of the subordinate clause is expressed in descriptive/structuralist grammars saying that the logic- al/understood subject of the subordinate clause is the same as the object of the main clause. In more current, mainstream syntax it is claimed that the subordinate clause also has a subject: PRO, an abstract (= unpronounceable, ineffable) pronominal element, which picks up its ref- erence/receives its interpretation from the full noun phrase ('John'), which appears in the main

• clause. What is theory-oriented in this account? The following features characterise the theory
• f grammar that offers the second explanation. (i) The supposition that there are abstract ( =

non-phonological) elements in a language, called empty categories. (ii) These (phonological- ly) empty elements have the same grammatical properties as the phonologically full items of the language, more specifically, the empty elements also belong to some grammatical catego-

• ry. In other words, abstract elements form an integral part of this grammatical theory and a-

nalysis that relies on it. (iii) Specifically, PRO is a pronoun and similarly to pronouns it re- ceives its interpretation (picks up its reference) from a non-empty NP; in this grammar empty pronouns are characterised in the same manner as non-empty, phonologically full pronouns. Consider the following sentence where there are no empty pronouns. Mary tried to persuade John but he didn't listen. Here the personal pronoun 'he' receives its interpretation from the NP 'John', which is the object of the first clause, in a phenomenon which is identified as ana- phoric (backward pointing) reference and this syntactic and semantic relation is indicated by

i i

subscripts: John , he . In this view PRO is just another exponent of pronominal elements:

i i

Mary persuaded John [PRO to buy a new car]. It must be obvious that the current syntactic account attempts to explain how the two clauses relate semantically to each other. 8