Some Issues in the Theory of Transformations

  • Robert Fiengo
  • Howard Lasnik
Part of the Studies in Natural Language and Linguistic Theory book series (SNLT, volume 20)


The specification of formal criteria 1 delimiting the class of possible transformations is of fundamental importance in linguistic theory. Ceteris paribus, if the notion transformation is narrowly defined, the class of possible grammars is thereby reduced. By restricting the class of grammars in empirically justifiable ways, we approach an explanation for the ability of human beings to acquire language. Postal (1975) argues that the theoretical framework we were assuming2 in Lasnik and Fiengo (1974) imposes overly restrictive formal criteria. Since such an argument implies that the class of possible grammars must be increased, the burden of proof clearly rests on its proponent. We maintain that Postal has failed in this task.


Direct Object Subject Condition Prepositional Phrase Indirect Object Grammatical Relation 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    We use the term “formal criteria” for conditions specifying the class of possible grammars, as against “conditions on applicability”, which determine the class of derivations, given a particular grammar The A-over-A condition, for example, is a condition of the latter type.Google Scholar
  2. 2.
    Basically, that which was developed in Chomsky (1955).Google Scholar
  3. 3.
    Cf. Chomsky (1956).Google Scholar
  4. 4.
    Two phrase structure derivations are equivalent if and only if they differ only in the order in which the rules are applied. It should be noted that a minor extension in this account of P-marker is required. As Chomsky (1961) has noted: Since transformational rules must re-apply to transforms, it follows that the result of applying a transformation must again be a P-marker, the derived P-marker of the terminal string resulting from the transformation. A grammatical transformation, then, is a mapping of P-markers into P-markers.Google Scholar
  5. 5.
    Cf. Chomsky (1961).Google Scholar
  6. 6.
    This transformation can be stated more simply. Cf. Fiengo (1974).Google Scholar
  7. 7.
    Cf. Peters and Ritchie (1973).Google Scholar
  8. 8.
    Which appears as (1), in Postal (1975).Google Scholar
  9. 9.
    Cf. Jespersen (1969, 155). We see this work as an example of traditional grammar.Google Scholar
  10. 10.
    Cf. Chomsky (1965, 71). A further confusion concerning grammatical relations on Postal’s part involves a failure to distinguish a transformation from its name. That is, Postal provides no evidence that Subject-Verb Agreement could not be stated within the theory of grammar that we accept, yet he indicates in his footnote 8 that we would be forced to claim that such a rule as Subject-Verb Agreement does not exist. We can only conclude that he feels that the mention of “subject” in the name of the rule renders the rule unstateable in our theory.Google Scholar
  11. 11.
    Cf. Postal ( 1975, fn. 11 ). In fact, Postal allows a wide variety of these. Rules may be stated in terms of subject, direct object, and indirect object at the levels of initial structure, end of (sub)cycle, and shallow structure.Google Scholar
  12. 13.
    Cf. Postal (1975, fn. 31). Actually, this observation bears on descriptive power only under the assumption that Postal’s theory prohibits all extrinsic ordering statements.Google Scholar
  13. 14.
    Cf. Postal (1975, fn. 31).Google Scholar
  14. 15.
    Curiously, for stage (c) of this derivation, Postal gives The songs were all written by Frank. We understand the passive construction to be derived by two rules, one postposing an NP, the other preposing an NP, cf. Fiengo (1974). Postal, we assume, takes Passive to be a rule making an object into a subject. We see no way, therefore, that Passive can produce his (c) from (b). It should be stressed that nothing in Postal’s rule of Q-Floating, or in his theory, determines the ultimate position of the quantifier. It is interesting to note that if one were to accept Postal’s assumptions, even in the face of the overwhelming arguments against them, one would be forced to conclude that the causative analysis of kill and other such verbs is incorrect. From an underlying structure such as Bill caused all the men to die the ungrammatical *Bill killed the men all would be derived, since the men is the cyclic subject of to die.Google Scholar
  15. 16.
    We abstract away here from the specification of exactly which quantifiers undergo the process.Google Scholar
  16. 17.
    Cf. Chomsky (1970).Google Scholar
  17. 8.
    Actually, for a number of examples, Postal advances proposals only to argue against them. He states that “if the SSC were a valid universal principle, or even a principleGoogle Scholar
  18. holding for English, it would in fact suffice to block… at least the cases (28b), (29b), (30b), (31b), (32b), (33b), (34b), (35b), (36b), (39b), (41b).“ He then presents several arguments against the SSC, to which we return below.Google Scholar
  19. 19.
    Cf. Chomsky (1973).Google Scholar
  20. 20.
    There are various possible formulations of the notion “subjacency”, cf. Chomsky (1973). We suggest this as the most adequate.Google Scholar
  21. 21.
    In the structure s[NP VP], the NP will, of course, always be assigned as the subject. In the case of VP complements, thematic relations will, in general, determine the subject.Google Scholar
  22. 22.
    It might appear that condition (25) would block the application of Object Deletion in (i): (i) John was easy for Bill to please. The optional presence of for Bill, and the fact that it can be moved, however, together suggest that for Bill is a benefactive and is not assigned as the subject of the VP. If this is the right analysis, then the generalization can be preserved that only NPs that are the sisters of VPs can be assigned as subjects.Google Scholar
  23. 23.
    For example, it seems clear that Sequence of Tenses is not essentially a rule of sentence grammar, although properties of sentences may affect the application of that rule. Another process of this type is discussed in Lasnik (1976). Assignment of pronominal reference is shown not to be a rule of grammar even though properties of sentence structure limit the possible interpretations.Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 1990

Authors and Affiliations

  • Robert Fiengo
  • Howard Lasnik

There are no affiliations available

Personalised recommendations