Ma Xiu Jia

Descriptive Models of the English Language

The twentieth century has been the stage of some major changes in grammar-writing. In the beginning of the 1900s, De Saussure developed the structural model, combining old ideas with new concepts. The model was extremely popular, and was adapted and applied to linguism and to literature studies. In the latter part of the century, two linguists proposed their own models which quite fast got their proper following. On the one hand there is Chomsky, who developed a Generative Grammar. M.A.K. Halliday, on the other hand, composed a Functional Grammar. In the following essay, special attention will be given to the generative and the functional model, giving them a place in the long tradition of linguism and explaining some of their most important features.

1. The Formal Model (generative or transformational grammar)

Avram Naom Chomsky, born in 1928, caused a revolution in the field of linguistics. In "Syntactic Structures" (1957), he outlines his concepts on language and on how language should be studied. His most striking innovation is his choice to describe how language is organized in the human mind, rather than in written or spoken text. In Chomsky"s own terminology, we could say he preferred research of the I-language (internal language, mental representation and computation of language in the mind) or competence (knowledge that users have of a language) to research of E-language (external language, the study of the use and situation of use of language). This supposes a rupture with the Structuralists, whose model reigned when Chomsky began his academic career. Structuralists, as well as linguists in the centuries before them, have always put the emphasis on the actual usage of language (Chomsky titles this aspect performance).

Another important innovation is his belief that it is possible to compose a grammar which describes principles, conditions and rules applicable to all languages, a so-called Universal Grammar. He wants to distil the essence of all human languages. Chomsky deems this Universal Grammar present in every human being, on from the point of being born. This becomes clear when we view Chomsy"s theory on First Language Acquisition. Primary language input (language used to or around the child) is combined with an innate and language non-specific Universal Grammar to form the child"s internal knowledge of its native language (a mental grammar of principles, parameters and lexicon).

Though innovations, there are certain influences which can be discerned in Chomsky"s theories. First and foremost, there is Plato (c. 429 - c. 347 BC). Plato saw the actual world as a mirror image (and a bad one, moreover) of an ideal world of perfection. People in the actual world bear distant images and vague ideas of aspects of the ideal world. Plato himself claimed for example that his political ideas were based on the ones ruling that perfect world. Specifically in his theories on language acquisition, Chomsky talks about the innateness of Universal Grammar, and suggests the human language capacity might be a remainder of some previous life in an ideal world.

A second great influence is the philosophy of René Descartes (1596-1650) and other rationalists. They claim that knowledge of man and his understanding of the world is a consequence of mental activity (the "ratio"). Chomsky"s methods reflect this. As he is studying an aspect of the human psyche that is untill today out of scientific bounds, his theories envolve quite some philosophical hypotheses and emendations. Rationalists also accept, like Chomsky, the existence of some innate knowledge, some wisdom that humans bear with them from birth onwards (see above). His ideas on these topics are voiced in "Cartesian Linguistics" (1966) and "Language and Mind" (1968).

Using the distinction of competence versus performance, Chomsky echoes several of his linguistic predecessors. Wilhelm von Humboldt (1767-1835) respectively speaks about "innere Sprachform" and "language as energy". Although language is energy and ever changing, there are some unalterable fundaments. With Ferdinand de Saussure (1857-1913) an analogous partition became more widespread. The Swiss linguist distinguished between "langue" and "parole" (respectively the personal language capacity and the actual language use).

Some minor influences and similarities can be found as well. Chomsky"s desire to compose one grammar describing the common fundaments of all human languages could be compared to the medieval importance of Latin grammar, which was the model for any grammar made of any other language. In other words, scholars in the Middle Ages tried to describe the European languages with the same (Latin) grammar.

When Chomsky develops his theory of Phrase Structures, his usage of the words "mother", "daughter" and "sister" evokes the Romantic interpretation of languages as part of a greater family.

The terms "mother", "daughter" and "sister" as well show the importance Chomsky gave to hierarchy of language. Herein influenced by structuralist theory, Chomsky saw language as a highly structured concept: Structured in performance (especially written text is extremely structured on various levels: morpheme, word, phrase, clause-levels - as already structuralists described with near exhaustiveness) but also in the terrain of competence.

Chomsky distinguishes the Deep Structure and a Surface Structure of a phrase. The Deep Structure of a phrase is the version of the phrase prior to any transformations. Its pattern is in harmony with one of the prototypic re-write rules which language users possess. In Chomsky"s opinion it is so that language is transformed in the stage between the unconscious conceiving of the phrase and the actual production of it. The Surface Structure then is the form of the phrase as it is going to be produced.

Within his theory on transformations and different levels of structure, Chomsky takes the phrase to be his unit of language. He is keen on legitimising this by moving words and word-groups in the clause. Phrases can"t be separated by these shifts. Phrases are ruffly the same as what de Saussure defined as "syntagm".

In shaping his theory, Chomsky tries to affirm some universal principles and parameters. The phrase is the first principle. Every language and the knowledge of language users seems to be organised in phrases. Each phrase has a head. This is the most important part of the phrase, considering that it defines class and structure of the phrase. The position of the head in the phrase is language-specific. Since all languages use heads, but heads are not really a fixed concept, Chomsky calls the head a parameter.

The interior structure of phrases can be represented by so-called re-write rules. Those are conventionally displayed formulas conveying some unchanging features of the phrase. These symbolic rules generate structures (real phrases). Each phrase category has several possible construction types. A noun phrase for example can have the following prototypic forms:

NP -> (Det) N

NP -> (Det) N PP

NP -> (Det) N S"

NP -> NP N

Besides of these very general rewrite rules, there are other minor rules and frames which are used to explain category-specific or even word-specific exceptions. One of these is the projection principle. It tries to systemise the concept of transitivity. Some words (especially verbs) cannot be used correctly in the absence of a specific category of phrase. The verb "to carry" for example needs some NP, specifying that what is being carried. ("John carries a chair.") In this context, we speak of the subcategorization frame of a word, which is represented much like the rewrite rules (e.g. carry [ V ; NP].

Aother minor rule of restriction is the system of "Selectional Restrictions". It helps to exclude construction which are grammatically correct but semantically impossible. Each verb gets a small pattern specifying the possible nature of subject and object. (e.g. *"Carrots run very fast." In this example the verb "to run" has a selectional restriction for the subject [+ animate]. Since "carrot" is [- animate], the combination is impossible.) This example already makes clear in some degree the difference between syntax and lexicon. Whereas syntax is ruled by its own rules, termed "Phrase Structure" (PS), lexicon has a set of rules of its own which are called "Lexical Insertion Rule" (LIR). LIR inserts lexical items under terminal nodes, and it has two conditions attached to it:

1. The terminal node must match the categorial class of the lexical item.

2. The phrase containing the terminal nodes must attach the subcategorization conditions of the lexical item.

Sound and meaning are quite important to Chomsky. In 1968 he published a book on phonology, "The Sound Patterns of English". Phonological analysis is also one of the three fields he defines when speaking about his objectives as a scholar in his first work "Syntactical Structures". He gives phonology a place in his theory by taking it as one of the two faces of language, sound and meaning. In his theory, the phonological side gets the name Phonological Form (PF). On the other hand, Chomsky defines meaning as the Logical Form (LF) of language, implying that it is rather the meaning which the language user wants to give, than the meaning one might make up out of the sentence. Finally, he adds to these two branches of language the idea that sound and meaning are linked by syntax, and as a consequence, he concentrates on the study of it.


Deep Structure (PS Rules and LIR)


Surface Structure

Phonetic Form Logic Form

One of the groundstones of his grammar being his theory on transformations, Chomsky defines some different forms of transformation. First of all, there is the device of topicalization. It means that a phrase leaves its normal place and gets the leftmost position of the sentence. It is mainly used to highlight one aspect of the sentence. In Chomsky"s terms, the phrase leaves a "trace t" on the place where it would normally be. The "trace t" is part of Chomsky"s so-called trace convention. This is a co-indexation showing the relationship between a moved category and its trace. Trace convention is adequate to show the deep structure of a sentence in which a transformation like topicalization has taken place. (e.g. "[That train ] I will miss [t ].")

A second type is that of the Wh-movement. When there are wh-words at the beginning of a sentence (in English, e.g. "Which train did you take?"), there seems to be a contradiction to the rewrite rule which exacts a NP after the verb "to take" (take [ V ; NP]). Intuition of native speakers learns that these sentences are correct and very frequent. Chomsky solves this by formulating four rules on the Wh-movement:

1. Move Wh-xP to comp. provided comp. is [+Q]

This condition means that the phrase containing the Wh-word can only be moved to the front of a sentence if it is a interrogative sentence (+Q) in contrast to a declarative sentence or a statement.

2. Move Wh-xP to comp. provided it is not included in a Wh-island

This means that the Wh-word which is placed at the left has to be part of the embedded clause from which it is removed, and it cannot be replaced by another Wh-word. (e.g. "Which did you know what Gilbert had in his house?"

3. Move Wh-xP to comp. provided it is not included in a complex NP

A complex NP is a NP in which an embedded clause depends on the head of the phrase. If the Wh-word is part of this embedded clause, it cannot be taken out and moved to the front of the sentence, since it is not directly part of it. The rankshifting necessary to lift the Wh-word to the level of the sentence is prohibited.

4. Move Wh-xP to comp. provided it goes to the nearest Wh-phrase

A third type transformation is that of the NP-movement. The general application of this NP-movement is the so-called passive voice construction, or the "object-to-subject raising" construction. It consists of moving ("raising") an NP to an "empty subject position". The "empty" in "empty subject position" refers to the fact that the structural subject formed by this NP-movement is not the real semantic subject. (e.g. "Richard wrote a nifty novel." becomes "A nifty novel was written by Richard." in which "a nifty novel" is the structural subject, but not the semantical subject, which is "Richard"). A co-indexed trace marks the place behind the verb where the new structural subject was supposed to be. (e.g. "[A nifty novel ] was written [t ] by Richard.")

A second form of NP-movement or raising construction is the so-called "subject-to-subject raising". This construction is built up around an unpersonal modal expression ("it seems that ...", "it appears that ...") which has an embedded sentence. The raising construction takes the subject of the embedded clause and makes it the subject of the verb of the root clause. The verb of the embedded clause becomes a to-infinitive with the same subject as the root clause. (e.g. "It appears that snow in May is extraordinary" becomes "Snow in May appears to be extraordinary" in which the verb of the embedded clause, "is", turns into the to-infinitive "to be").

There are two conditions to the raising construction:

1. The sentence must have a non-TSC (non-tensed condition): "Snow in May appears to be extraordinary."

2. The sentence cannot be separated from the target position by a SSC (specified subject condition), or in other words, the root sentence and the embedded sentence must have the same subject.

The NP-movement, finally, has two more minor types. Both of them imply large Noun Phrases. The first one, extraposition, splits an NP in an NP and an XP (the complement of the head of a NP is not necessarily a NP itself) and puts the XP to the rightmost constituent of the VP. (e.g. "Reports on the ongoing elimination of the Amazone forest in Brazil were published." becomes "[Reports [t ]] were published [on the ongoing elimination of the Amazone forest in Brazil].") The second type is called the "heavy NP shift" and involves placing an NP belonging to the VP (e.g. the direct object) at the end of a sentence. (e.g. "I gave the book to the chairman of the very powerful Milk Federation yesterday." becomes "I gave the book yesterday to the chairman of the very powerful Milk Federation." in which the Indirect Object is moved to the end of the sentence.) At the basis of this move is the so-called end-weight principle. There seems to exist an unconscious tendency to move large phrases away from the center of the sentence, either to the left (although the phrase gets a special highlight then) or to the right.

Chomsky adds to his syntactic theory an aspect of semantics. In his "Thematic Roles", he distinguishes between agent (the doer of an action), the patient (the person or thing object of the action), goal (recipient or destination of the object of the action) and theme (the thing which is moved by the action). The distinction between patient and theme has to be clarified. One could say that patient is equal to the direct object in a sentence in active voice if there is no indirect object present. When there is an indirect object, the direct object is the theme. A term to summarize the sequence of thematic roles in a sentence is the "argument structure". An example would be: "Frank sent a box of cigars to his brother." of which the argument structure is: agent - theme - goal (the agent is "Frank", the theme is "a box of cigars" and the goal is "his brother").

To Chomsky, phrases have cases. In present English, remnants of case are to be found in for example the personal pronouns, which have a subjective, objective and a possessive case (respectively "I", "me" and "my"). Interrogative and relative pronouns bear some memory of case as well (who, whom, whose). These cases remind of the old nominative, accusative and genetive. The case of a phrase is determined, according to Chomsky, by the same "case filter", which can assign a nominative (subjective), an accusative (objective), a genetive (possessive) or an oblique (after preposition) case. The assignment of those cases are directed by a complex Government Rule.

a governs ß iff (if and only if)

1. a is a Xo category (terminal node)

2. a constituent-commands ß

a constituent-commands ß iff

1. the first branching node dominates a and also dominates ß

2. a does not dominate ß

Another theory, the Binding Theory, is concerned with anaphores, pronouns and r-expressions. Anaphores are mainly reflexive and reciprocal pronouns and they necessarily have an antecedent in the sentence (e.g. "She could hit herself."). Pronouns don"t normally have an antecedent, though they can optionally take one (e.g. "She could hit him."). Finally, referential expressions involve proper names and determinant phrases which cannot have an antecedent in the sentence (e.g. "She could hit Jorge."). These three categories of determinant phrases have a counterpart in the "null Determinant Phrase". These are either DP traces (something similar to anaphor; e.g. "The Pope seems to be old"), variables (traces of Wh-phrases and qualifier phrases, similar to pronoun as they don"t normally have an antecedent; e.g. "What did you see in San Pablo?") or PRO (similar to pronouns and anaphores, it is the subject of non-finite clauses; e.g. "Harris hoped to rest on sunday.").

Chomsky formulates his binding theory as followed:

a binds ß iff

1. a is co-indexed with ß

2. a c-commands ß

According to this rule, anaphors are always bound. In contrast to that, pronouns and r-expressions are free, since there is only exceptionally co-indexation and constituent-command (e.g. the stereotipical "Hildegarde saw the man who hit Dirk .")

Chomsky worked the principles and parameters of his Universal Grammar into a model of syntactical analysis which he calls "X-Bar Theory". It is based on the phrase, with its most important types of the Noun Phrase, Verb Phrase, Adjective Phrase, Adverb Phrase and Prepositional Phrase (these phrase-types necessarily have a head that belongs to the mentioned lexical category: e.g. Prepositional Phrase has a preposition as its head). The model is illustrated as followed:

XP (X"")

specifier X"

X (Head) Complement (S)

The only thing that might strike as being somewhat unnecessary is the place of the specifier, since it is very often missing.

Chomsky applies this model as well to full sentences, which he calls "functional phrases". As each element of the sentence has its own function, and as Chomsky considers the phrase the unit of his X-Bar Theory, the term "functional phrase" is very adequate. In the functional phrases, the most important concept is that of the Inflectional Phrase. It is illustrated as below:

IP (I"") (Inflectional Phrase)

I" (Inflection)

I VP (V"")

(tense, agreement)

present-past, singular-plural


V Complement

In this model, it is to be noticed that Chomsky opts for the inflection as the Head of the functional phrase, whereas the inflection is at times not even visible in an English sentence. (e.g. to do -> "How do you do?": we know the first, auxiliary "do" is the inflected verb, but there are no graphic nor phonic markers to distinguish it as such from the uninflected infinitive).

A last detail of the X-Bar theory that deserves our attention is the way in which is dealt with embedded clauses. Chomsky uses a Complementizer Phrase, consisting of a complementizer (the head) and an inflectional phrase.

To end this small summary on Chomsky, a remark on his "Minimalist Program" could be added. Chomsky was radical in his proposed change of grammar from the start of his career. In 1997, he published a book ("The Minimalist Program") in which he reduces his own principles to a minimum. He justifies this move by pointing towards two principles, which indeed seem to rule language: the principle of economy which implies that language users use as few elements as possible, and the principle of full interpretation which says that we try to be clear and unambiguous when using language. Chomsky takes these unconscious aspirations of language users and applies them to his grammar writing: to describe language as clear as possible with as few elements as possible.

2. The Functional Model

Halliday"s functional model generally has a different approach to language than Chomsky"s formal model. Chomsky considers it his task as a scholar to solve four rather philosophical problems (What constitutes language in the mind?, How is such knowledge acquired?, How is that knowlegde put to use?, What are the physical mechanisms that serve as the material laws for this system of knowledge and for the use of this knowledge?). Using these problems as a starting point, Chomsky seems to commit himself to the exploration of language as well as of the human psyche. Halliday concentrates his theories more on language, in particular one aspect thereof: meaning. What the functional model provides is a series of methods for the interpretation of texts, of the system and of the elements of the linguistic structure. The first preoccupation of this model is to find meaning, to interpret, not to analyse morphologically or syntactically. These analysis can be applied though, in trying to answer the question how meaning is expressed in a text, one of the main concerns of the Functional Model. The fact remains that where Chomsky boldly chose to map "competence" after a long tradition of studying "performance", Halliday goes back to almost solely studying "performance". The text is the center of attention, not human language capacity.

The units of language with which the Functional Model deals are texts. This presents an important contrast with the earlier linguistic traditions.

Halliday"s contemporary Chomsky took the phrase as his unit, after centuries in which linguism was mainly restricted to the study of words, both morphologically and phonologically. Dionysos of Thrax might well have been one of the most important iniciators of this centuries-long trend. His study on Greek morphology in which he distinguishes eight different word classes laid the pattern for centuries of linguism. Dionysos" theories were applied on Latin by the Byzantine grammarian Priscianus Caeseriensis (6th century). The classification which Priscianus made was very important. It was the ruling thinking pattern on grammar during the Middle Ages, and untill today, some of his terminology is still in use (nomen, verbum, ...).

Though Halliday introduced a new concept of grammar, the emphasis on meaning in dealing with language is not new. His heritage includes Aristotle (384-322 BC). In "De Interpretatione" Plato"s student speaks about the difficulty of uniform meaning. For Aristotle, language is the representation of experiences in the mind, and this leads him to claim that meaning changes interpersonally because different people have different experiences. Halliday as well talks about meaning as a shifting concept. In his view, much is determined by the "context of use". Inspired by anthropologist Malinovsky, Halliday sees a context of situation and a context of culture. The first is determined by the environment of a text, the latter by the used language system. (This latter clearly echoes the theories made public by the American linguist and anthropologist duo Sapir-Whorf. They claim that languages are predetermined by the social and cultural characteristics of its users. A famous example is the fact that Innuit -Eskimo language- has some seven separate words to talk about types of snow.)

A definite contrast between Aristotle and Halliday is the importance the Greek gives to writing, which for him is the representation of language. Halliday does not put much emphasis on written text, one might even say he prefers studying spoken text, being more spontaneous and natural. Moreover, Greek tradition before Aristotle was mainly centered on speech in the so-called "rhetorikè" (art of speaking), which was one of the most important subjects in Greek education of young aristocrats.

Besides Aristotle, two other linguists discuss topics which reoccur similarly in Halliday"s Functional Model. On one hand, the Roman scholar Marcus Terentius Varro (116-26 BC) stressed the social function of language (communication between people). On the other, Wilhelm von Humboldt (see above) envisioned language as energy, which sounds very much like Halliday"s conception of text as a process, something that is continuously under production.

Returning to Halliday"s theory, I will try to present the general idea and give some explanation on some of the most important aspects. I have already explained the importance of meaning, and the influence of context. This goes together with the idea that all units of language function within the text to so help transfer meaning. It is in this respect that Halliday uses the term constituency. With it, he denominates all the elements (in hierarchical order) making up a clause (Halliday"s early point of interest), a poem or a text. (e.g. in a sentence: letters - morphemes - words - phrases - subsentences - sentence).

The most important distinction Halliday made is expanding the concept of the subject and assigning functions to the clause. The subject is divided into three concepts. Firstly there is a psychological subject, being that what the message is concerned with. The subject then is named "theme". A second possibility is the subject as that of which something is being predicated. This subject, which is titled "subject", is the grammatical subject of a clause. A last subject is the actor. It is the so-called logical subject, involving the doer of the action predicated in the clause. An important consequence of the conception of a subject on different levels is that units of language become multifunctional: one unit, like "the child" in "The child said bo.", can be theme, actor and subject at the same time.

The clause can equally be considered threefold. It can be a message, an exchange or a representation, and each of these have their proper implications. The different clause and subject considerations run parallel. Judging a clause a message, we have to look for a psychological subject, a theme. Clause as exchange and clause as representation demand respectively a grammatical subject and a logical subject or actor.

Of course, in whatever way you look at a clause, in a large majority of cases there are more units of language than just those of the psychological, grammatical or logical subject. For each of the above mentioned clause-considerations, Halliday provided a full terminology to cover the other functions in the clause.

If the clause is regarded a message, there are two elements. The theme is always the starting point of the message, it is the word-group preceding the verb. The rheme is the part of the clause that doesn"t belong to the theme. Various classes of word-groups can take on the function of theme (nominal group, adverbial group, ...). A clause-type worth mentioning is the "Thematic Equative". In this case, theme and rheme are joined by the copula "to be" and the theme takes on the form of a nominal, pronominal or adverbial group with a relative clause attached to the noun, pronoun or wh-word (e.g. "What the boy yelled was goodbye." -as is made clear by this example, the antecedent can be silent: "(That) what the boy yelled was goodbye.").

Themes can be simple or multiple. Simple themes comprise only one word-group or phrase, whereas multiple themes stretch over several phrases, so forming an internal structure of its own. Multiple themes are mainly restricted to spoken text. Within them, there are a series of three different metafunctions that its "constituents" can have. The textual metafunction connects the clause to the previous clause (e.g. conjunctive adjuncts). Interpersonal function serves to communicate with the receiver of the text. Theme-components with an ideational function finally bring forward a topic of the clause, in the same way as did the simple theme.

A different feature of the theme is that it can take on two forms, being marked or unmarked. The unmarked theme is the one coinciding with the grammatical subject. (e.g. "The girl is running." A marked theme can have any grammatical function except the one of the grammatical subject. It is mainly used to highlight one aspect of the clause in particular ("Next year, I will be older than I am today.").

Another way to consider the clause is that of the exchange. In an exchange, there are two producers: a listener and a speaker. They each have a role, respectively giving and demanding. The exchange can be of information or of goods and services. These four elements are together Halliday"s four primary speech functions: offer, statement, command and question. Combining a speech role and a type of exchange delivers one of the clause types or speech functions mentioned above. (e.g. giving goods and services is an offer, demanding it a command; giving information is as statement, demanding it a question)

Each one of these speech functions provoke two possible answers: an offer can be completed with an acceptance or a rejection. A statement can encounter an acknowlegdement or a contradiction. A command can be met with an undertaking or a refusal, and a question with an answer or a disclaimer.

The clause as an exchange can have two functions. If the exchange concerns information, we can speak of proposition. The exchange of goods and services makes the semantic function of the clause a proposal.

The main elements in the clause as exchange are mood and residue. The mood consists of the subject and the finite of the clause. The finite expresses tense or modality and as consequence, it can have temporal operators (past, present, future) or modal operators (with varying intensity: low, median or high). The residue can consist of the predicator, the complement or the adjunct of the clause.

A last way in which we can consider a clause is the representation, referring to how language users represent the real world in their minds in the form of a process. This process has three main components: the process itself, the participants in the process and the circumstances associated with the process.

The terminology used for the participants of a process and its circumstances depends of the type of process we encounter. If the process is "material" (meaning that something is done in the course of the process), we have an actor and a goal. The actor is the one doing the action, the goal is that or he to which the process extends (e.g. "The man bought a sofa" with "the man" being the actor, "a sofa" the goal and "bought" the process).

The process can as well be "mental", depicting a process with a human(like) participant in whom a mental process completes itself. These mental processes can be predicated by verbs of perception (see, hear), verbs of affection (like, loathe) or verbs of cognition (know, ponder).

A last type of process is relational. It is either attributive or identifying, and the participants are the token and the value. The token is that which is being identified or being attributed with something, the value is that with which is being identified or attributed.

A final remark could be spent on minor types of processes: behavioural, verbal or existential. In the behavioural type, psychological events are portrayed (weep, cry), having a behaver and the behavioural process. The verbal process includes an act of saying, with sayer and target as participants. The existential process is a process with only one participant. It includes a grammatically "empty" subject "there" and a participant which is called the existent.

Linguism today has changed profoundly because of these two linguists. Language teaching has been influenced by both these theories. Nowadays, it happens that a university department decides on the type of grammar it wants to teach. Often is opted for a proper mixture of the structuralist, generative and functional model. It is not untill one takes a look at the distinctive theories in their own context that one realises in what degree only three men have influenced today"s thinking about language.