Notes on Nominalization

-As is probably already clear, there are certain similarities between...sentences, on the one hand, and their nominalizations on the other. These similarities are found in terms of morphological form, semantic interpretation and the selectional features of verbs and their nominals.

Once these similarities were noted, the question then became, how can these similarities be accounted for within a grammar of English? It became necessary for the study of nominalization to precisely define what this relationship was - because obviously there was a relationship.

-overtime, two solutions emerged, what are called the transformationalist and lexicalist approaches.

-The transformationalist approach theorizes that sentences and their nominals are all derived from one sentence-like underlying structure, essentially a deep structure. Thus the morphological, semantic and syntactic similarities are due to the fact that sentences and their nominalizations are all transformed out of virtually the same base structure.

-Lexicalist approach, on the other hand, states that the relationship between sentence and derived nominal is specified in the lexicon rather than through transformations. Now the essential difference between the two approaches concerns the status of derived nominals. The transformationalist approach argues that they are "transforms" - that is to say, they are derived from the same base structure as their corresponding sentences. The lexicalist approach, however, states that derived nominals are nouns in the base component.

-in "Remarks on Nominalization", Chomsky outlines three arguments against the transformationalist approach, and for the lexicalist approach.

-Productivity
-consider the sentence "John is certain to win the prize"
-gerundive nominal: "John's being certain to win the prize."
-derived nominal: "John's certainty to win the prize"
-same kind of difficulty arises with words like "easy" and "amuse."
-for every sentence, there is a corresponding gerundive nominal and vice versa; this is not so with derived nominals
-according to transformationalist approach, derived nominals should be formed as easily as gerundive nominals but this is not so
-lexicalist approach proposes that such words as "amuse" and "certain" appear in the lexicon with "strict subcategorization features" indicating what kinds of complements may be taken by each; thus ungrammatical forms can not be derived

-Internal Structure
-Consider a sentence like "John has failed the exam," - present perfect aspect
-the gerundive nominal for this is "John's having failed the exam" - the nominalization
still carries aspect, still retains the internal structure of the verb from which it is transformed
-no equivalent derived nominal; derived nominals can't carry aspect
-derived nominals can pluralize, take determiners, appear in any noun phrase structure; in fact they resemble nouns in every way
-the transformationalist hypothesis would expect that a derived nominal reflects the internal structure of its corresponding sentence
-lexicalist approach specifies that derived nominals are nouns in the base component and thus have the structure of noun phrases

-Idiosyncrasy
-laughter, marriage, construction, actions, activities, revolution, belief, doubt, conversion, permutation, trial, residence, qualifications, specifications, etc.
-such derived nominals exhibit a range of meaning and extremely varied semantic relations to their corresponding base forms. To accommodate this in the transformationalist approach, we would need to assign a range of meanings to the base form as well.

Chomsky opted for the lexicalist approach over the transformationalist approach, despite noting that a "compromise solution" may be possible, concluding that the lexicalist approach more clearly accounted for the internal structure, semantic idiosyncrasy and syntactic irregularity of derived nominals.

SOURCE

Chomsky, Noam. "Remarks on Nominalization." 1970.