RULES OR NO RULES? THAT IS THE QUESTION.
A look at Chomsky’s rule based approach versus Skousen’s analogical method.
What is a linguist? Well, this is very difficult question to answer. Maybe a more appropriate question would be, what does a linguist do? Linguists have been trying to define their work for centuries but still the answer is not consistent. The linguist’s answer could vary depending on the time period in which he was schooled, but also depending on which school of thought he wishes to associate himself with. For instance, an early linguist may have been trying to find a perfect language in which to preach the gospel of Jesus Christ to the various peoples of the world. Some linguists have been interested in creating languages while others are determined to explain language. Some have delved into analyzing language until it has been deduced into a meaningless philosophy. Others have tried to restore its virtue as a useful mode of communication and then pursued the challenge to describe it. Whatever direction a linguist dares to go, he makes a significant contribution to the field of which other budding linguists study, and join their throng. These divisions in the world of linguistics are the result of a passionate linguist’s logos and pathos that cannot be disconnected from him. Thus, we have a variety of linguistic study taking a variety of angles to look at the same thing: language.
1. TWO ANGLES. Out of the divisions of linguistic viewpoints, come many different methods to try and describe language. Why would a linguist want to describe language? Well, those reasons are as varied as the schools of thought. Describing language is, in itself, it’s own art form. Purposes can range from trying to find the true character of the language we speak to knowing how to explain it in a concise fashion to a learner of our language. Other reasons might include a linguist’s desire to explain what he hears and another’s desire to prescribe formulas to produce “correct” language. This paper will look at a couple of very different approaches to explaining language of which will be summarized in the next couple of paragraphs.
1.1. CHOMSKY. Noam Chomsky, the famous modern linguist, has sought to describe language according to rules. His research shows that one can take a given language and deduce formulas to explain the phenomena that make communication possible in that language. He asserts that people also have some biologically intrinsic rules in our makeup, called a universal grammar. This universal grammar can help us determine the concept of grammar and how to use it appropriately for communication. On his rule based approach to grammar he says:
A person who has learned a language has constructed a system of rules and principles- a grammar- determining a sound-meaning relationship of some sort over an infinite domain. The linguist’s grammar is a theory of attained competence, under conventional and entirely appropriate idealizations. The general theory of grammar- call it ‘universal grammar’ (UG)- is a system of principles that determines: (1) what counts as a grammar and (2) how grammars function to generate structural descriptions of sentences. (Chomsky 1976:303)
Here we can see that if a person attains competency in a language, then he has learned how to appropriately apply the rules that constitute his language. In other words, he has learned to scrutinize the application of these rules to make coherent communication possible.
1.2 SKOUSEN. Dr. Royal Skousen, linguist at
… the analogical approach makes no distinction at all between competence and performance. The same analogical procedure produces the rule-governed behavior as well as the tendencies towards change (including historical drift, dialect development, adult errors, and children’s language). (Skousen 1995:221)
Skousen’s approach with the analogical method is to use data that comes from real life examples. His model attempts to explain how imperfect language can be formed in spite of knowing the rules.
2. ANALYZING THE DATA.
2.1 CHOMSKY’S DATA. How can every conceived act involving the production of language have a rule associated with it? Well, Chomsky explains that we have our base rules, or the universal grammar mentioned earlier, and that deviations from those rules have had a system of transformations applied to it. Language goes through a medium of transformations to fit the particular situation of which we would like to speak. He says that in a phrase, there are two different structures. First is the deep structure, which is our phrase prior to any kind of transformations and then post-transformations we get a surface structure (Chomsky 1976:305). This is our utterance after it has been filtered down through various transformations. Chomsky says, “The grammar determines sound-meaning relations. I will assume that the mechanism is a set of rules that associate transformational derivations with representations of sound and representations of meaning” (Chomsky 1976:305). So, these transformations are necessary in order to deduce meaning. In another explanation of transformations, Chomsky says this:
In all theories of transformations that have been presented even in a semi-explicit way, a transformational rule is defined in terms of a structural description (SD) and a structural change (SC), where the SD specifies the domain of the transformation and the SC states what effect the transformation has on an arbitrary member of this domain. (Chomsky 1976:308)
Let’s look at an example of an application of transformations to see exactly what Chomsky has in mind with this idea.
2.2 EXAMPLE. This example is taken directly from Chomsky 1976:309.
For example, we might formulate the passive transformation in English with the following SD:
(vbl, NP, Aux, V, NP, by, #, vbl)
In this formulation, the two terms vbl are (end-) variables, so that the first and last factors of a string X to which the transformations applies are arbitrary. The second and fifth factors must be NP’s (each is an NP), the third and Aux(iliary), the forth a V, the sixth by and the seventh # (by and # are terminal symbols; we may think of # as an “abstract” representation of NP). In particular, the passive transformation, so formulated, will apply to the [next] string, where – indicates the break between factors:
Yesterday – the boy – PAST – read – the book – by - #
This might help to see these transformations of which Chomsky believes makeup the rules of the English language. I wish to clarify his work a little at this point. To do so I would like to again site Chomsky’s example.
(vbl, NP, Aux, V, NP, by, #, vbl) (Chomsky 1976:309)
Here the variables can be anything of the speaker’s choice, which will then be followed by a noun phrase. If this sentence were in the active voice, this first noun phrase would be the active voice’s noun phrase located in the direct/indirect object slot. Here, for the passive voice, it has been moved to the front to be the subject. Following the first noun phrase, we have the auxiliary verb followed by the main verb. This is the standard rule for constructing the passive voice. Our verb sequence is then followed by another noun phrase. This noun phrase would have a similar function in the active voice, that of direct or indirect object. Next, we add our characteristic passive voice feature, by + NP. This noun phrase would function as the subject of an active voice sentence. So, we can see that Chomsky takes some sort of base sentence having the characteristics of present, positive, and active and adds the applicable transformations. In the case “yesterday – the boy – PAST – read – the book – by - #” he added transformations of past and passive to a base sentence of The mother reads the book to the boy. This would give us an end result of Yesterday, the boy was read the book by his mother. Hopefully, this illustrates the use of transformations. Remember, these transformations are already in our universal grammars, so it is just a matter of application.
When Chomsky wrote this article, he realized the limitations of the rule-based system of describing language. He said:
Even if conditions are language- or rule-particular, there are limits to the possible diversity of grammar. Thus, such conditions can be regarded as parameters that have to be fixed (for the language, or for particular rules, in the worst case), in language learning. (Chomsky 1976:315)
Chomsky explains this deficiency in terms of competence and performance. He says that competence is the knowledge of these rules and that poor performance equals poor application of the rules or transformations. This is where Skousen’s research picks up.
2.3 SKOUSEN’S DATA. As mentioned earlier, Skousen’s analogical process to defining language considers performance to be equal to competence. His research stems from trying to explain the inconsistencies in language production. William G. Eggington comments on Skousen’s work by saying, “…actual language data constantly indicates that cognitive ‘rules’ cannot realistically account for the significant, non-random variation in language acquisition processes” (Skousen 1995:i). Skousen comes from a different angle to explain data. In his research, he explains that Chomsky, and his rule-based followers, divide data into neatly defined contexts and then apply a rule from the appropriate context (Skousen 1995:214). His alternative is this:
Instead of trying to determine the behavior for the contextual space, the analogical alternative stores exemplars in a multivariate contextual space and then uses these examples to predict behavior for specific given contexts. (Skousen 1995:216)
As we can see, Skousen’s approach is entirely different. He takes what is actually there and finds a logical way to explain it. Just as Chomsky, he gives many examples to illustrate his point. Let’s look at one here.
2.4 EXAMPLE. Summarizing Skousen’s work, he says that there are basic types of behavior to which the rule approach and the analogical approach react differently. One of these types of behavior is categorical, where data can be categorized (Skousen 1995:213). To understand the difference in approach, we must consider, as did Dr. Skousen, both methodologies of the two schools. First, if the rule school wanted to use the categorical method to describe the a/an distinction in English, they would do something like this, taken from (Skousen 1995:214):
Categorical an a
What is the rule approach trying to tell us with this simple diagram? They are constructing two contexts, an environment where a or an would come before a vowel, _V, and an environment where a or an would come before a consonant, _C. Their conclusions are obvious, where we can see an comes before a vowel and a comes before a consonant.
Skousen shows in his research that the a/an rule is not followed perfectly. He predicts that there will be leakage in this principle. If a given context is _V there is a small percentage that will use a instead of an. However, in the context is _C there will be no inclination to mistake and use an. The analogical approach has a similar model diagram to the rule approach but can take into account the imperfections that are associated with this rule. This model taken from Skousen 1995:216.
an an an an a a a a a
an an an an a a a a a
“categorical” an an an a a a a a a a
an an an an a a a a a
an an an an a a a a a
Here we can see that in most instances usage of an and a are consistent, however, careful inspection of the third line of data shows a small tendency for leakage. Skousen would say that this data is what is actually out there. He asserts that the rule-based approach cannot deal with what is actually out there. Again, on the performance versus competence issue, he says this:
…the rules themselves are inherently incapable of predicting the actual non-symmetric behavior. In fact, the rules are often declared to represent some kind of abstract (or non-empirical) ‘competence’. The rules per se do not have any built-in empirical interpretation, so the performative aspects must be added. In actuality, the whole competence-performance distinction may well be the result of first trying to use rules to describe language, and then adding performative factors to patch up the failure of the rules to account for dynamic aspects of the language (such as drift, fuzzy boundaries, leakage, and so forth.) (Skousen 1995:221)
This is where Skousen claims that for the analogical method, performance and competence cannot be separated.
As a teacher of English as a second language, I can agree that language has rules and I teach them daily in grammar lessons. However, are we to say that language can only be understood according to those rules? These inherently important rules that are not even followed perfectly by native speakers! But this is not the entire issue. When we come to the point where we want to describe language, which angle is more appropriate. I would argue that the analogical method is more accurate than the rule method. As Skousen puts it, “The analogical method is a procedural one. By this, I mean that it can only react (or make a prediction) in terms of a given context…. On the other hand, rule approaches are declarative” (Skousen 1995:227) The remainder of this paper will look at the argument in favor of a procedural method for describing language as opposed to a declarative method, of which I tend to agree.
3. THE ARGUMENT. Skousen’s analogical method is not only more accurate but also more succinct. Of course, his method requires the collection of data, but can also use predictions. Chomsky’s rule based approach must produce many different rules to account for the occurrences of inconsistencies. In his research, Chomsky lists the various different kinds of transformations. He says:
Among these rules there are syntactic transformations, lexical insertion rules, phonological rules, morphological rules of various sorts, rules that place word boundaries and that determine intonational structures, rules that specify the class of items that are available for lexical insertion, with their intrinsic properties. (Chomsky 1976:305)
This seems to be too much data to work with, and for each of these rules that he conceives to explain language, there are exceptions and situations for which these rules will not apply. The problem here lies with an issue mentioned earlier and that is whether or not we can use rules first to describe language. Chomsky is guilty of this. Does he prescribe language or does he describe it? Do linguists prescribe language or describe it? Do grammarians prescribe or describe language? It is a known fact that grammarians prescribe and linguists describe. It’s a matter of what is out there. What is being said and how can we analyze that? What can we learn from language by those who are using it? So, what is Chomsky doing? Well, he prescribes rules in order to describe language. Skousen, on the other hand, describes language as he sees it and this really does seem to be the way we process language also.
4. CONCLUSION. This argument of exactly how necessary rules are to describe language is not easily resolved. Obviously there are rules that affect language, but to what extent should we rely on them to describe what we hear in a real life situation. Can competence be separated from performance? Probably, but should that affect the way we describe language. Is language more influenced by rules than governed by rules? I would say so. Rules are important to add consistency to language, but not necessarily always followed.
Chomsky, Noam. 1976. Conditions on rules of grammar. Linguistic Analysis 2:303- 349.
Skousen, Royal. 1995. Analogy: A non-rule alternative to neural networks. Rivista di linguistica 7:213-231.