Group 4: "Problematic suffixes" (6 of 43)

We may now turn to the last group of suffixes, which Fabb labels "problematic" (1988:536), because they do not display the kind of clear-cut restrictions Fabb posits for the suffixes discussed so far, namely that they either attach only to non-complex bases, to one other suffix, or with no restriction at all. However, as we have learned from our close inspection, the stacking restrictions of the suffixes dealt with so far are much more intricate than Fabb conceived. We will see that his "problematic" suffixes are no more (and no less) problematic than the ones discussed earlier.

The six suffixes of this group are denominai -al (which, according to Fabb, attaches to -ion, -ment, and -or), the nominalizing suffix -ion (which attaches to -ize, -ify, and -ate), the nominalizing suffix -ity (which attaches to -ive, -ic, -al, -an, -ous, and -able), and the three deadjectival suffixes -ism, -ist, -ize, which may all be preceded by -ive, -ic, -al, and -an. Fabb speculates about possible explanations, like blocking effects, etymological restrictions, and affix-driven restrictions, but does not arrive at a satisfactory solution to the - in his framework - somewhat strange behavior of the six suffixes.

Contrary to Fabb's claim, denominai -al attaches not only to -ion, -ment, and -or, but also to derivatives involving nominal bases in -ure )apertural, cultural), -ent/-ant )presidential, componential, consonantal), -ancel-ence (concordantial, conferential), -cide (insecticidal, suicidal), -ory (laboratorial), -ary )secretarial), -ive )relatival, substantival). Most of these nominal suffixes take the adjectival suffix -al regularly, but rival processes, especially -ous, may intervene.¹The data clearly speak against the first of the two possible solutions Fabb discusses, namely that "-al belongs with affixes like -ism in that it attaches to underived words and to three specified suffixes" (1988:536). The - equally unsatisfactory - alternative Fabb proposes is that "-al is in principle free to attach, but there are other (unknown) selectional restrictions involved which prevent it attaching to these suffixes" (1988:536). As will become clear shortly, the Latinate Constraint, in conjunction with individual base-driven selectional restrictions can nicely account for the attested combinations.

With reference to Anshen et al. (1986), Fabb discusses briefly the role of the feature [+ Latinate] as a possible solution of the combinatorial properties of -al. He comes to the conclusion that positing an etymological constraint of the kind that -al attaches to (all) Latinate suffixes cannot be correct, since there are no attested combinations of -al preceded by the Latinate nominal base suffixes -ity and -ism². In our view, this fact does not necessarily speak against the operation of the Latinate Constraint, since the Latinate Constraint does not preclude the operation of additional restrictions of a different kind. Thus the absence of *-ity-al and *-ism-al can be straightforwardly explained by base-driven selectional restrictions and semantic constraints of these nominal suffixes, which have to be stated anyway. As was mentioned above, -ism only allows adjectivization through -ist(ic), which automatically rules out -ism-al, and there seems to be a general semantic constraint that does not allow the formation of relational adjectives by suffixation to quality nouns that are derived from adjectives (cf. *-ity-al, *-ity-ary, *-ness-y), which probably has something to do with the fact that such forms would be synonymous with the root.

We have already argued that there is a base-driven restriction with respect to the Latinate verbal suffixes -ize, -ify, and -ate, namely that they all take -ation (i.e. the respective allomorph) as an abstract-noun-forming suffix. The "problematic" distribution of -ion is therefore only apparent. In the base-driven model presented here the behavior of -ion is the natural consequence of the properties of the Latinate verbal suffixes.

With respect to -ity, Fabb offers a more promising account. He observes that -ity attaches to Latinate bases in general, which rules out all combinations with Germanic suffixes. Furthermore he posits that -ity only attaches to non-glide consonants, which rules out the supposedly non-occurring combinations involving -ory and -ary³. However, at least with adjectives in -ary, the nominalizing suffix -ity is possible, if not necessary; consider complementarity, supplementarity (rare). Selectional restrictions seem to be responsible for the apparent lack of forms in -ority. All derivatives ending in -ority given by the OED are nominalizations of bases ending in the string -or (like minority), not in the string -ory. With adjectives in -ory, standard nominalization involves either replacement of -ory by -ion )satisfactory / satisfaction), or suffixation by -ness. In any case, adjectival derivatives in -ory, but not those in -ary, are outside the domain of -ity.

There are, however, more adjectival suffixes that can be followed by -ity, namely -ar, (as in polarity, peculiarity, scalarity), and -ile (infantility). These are both [+ Latinate] suffixes and, pending other constraints, the Latinate Constraint predicts their existence. One other mechanism, blocking, is responsible for the only remaining possible combination Fabb considers, -ant-ity, which is blocked by -ance. Under a base-driven approach, all of the suffixes that can be followed by -ity need to be marked for this property. Since -ory is not marked for any specific abstract noun suffix, it undergoes the most general process, -ness suffixation, whereas -ant is marked for -ce or -cy. Alternatively, and probably more elegantly, one may conceive the domain of -ity as comprising potentially all Latinate stems (including derived stems with a Latinate suffix, of course), unless otherwise specified.⁴

In sum, the distribution of -ity can be tentatively explained by a careful definition of its domain, the Latinate constraint, and blocking effects due to base-driven selectional restrictions of potential bases. Such base-driven selectional restrictions are needed in any case to rule out forms like *notorious-ity, or * adventurosity.

The deadjectival suffixes -ism and -ist seem to attach freely to all kinds of adjectives, provided that the combination is licensed semantically. Thus, in addition to the four suffixes Fabb finds, the adjectival suffixes -ile, -able, and -ar are also attested to precede -ism (as in infantilism, probabilism, particularism), and it seems that only semantic-pragmatic factors speak against forms involving other more picturesque adjectival suffixes like -esque preceding -ism/-ist. Consider the putative kafkaesquism, which could certainly denote a theoretical framework developed by a circle of literary critics who try to find kafkaesque traits in any piece of fictional writing (with a kafkaesquist being a member of this circle). It is the oddness of the denotation and not of the morphological form that makes kafkaesquism a presumably unacceptable derivative. Parallel arguments hold for putative derivatives involving other adjectival suffixes, consider ?girlishism, ?peacefulism, Ί wholesome ism. Factors like blocking may additionally be involved (as always), as can be seen with ?helpfulism which is probably blocked by altruism.

The last "problematic" suffix is -ize to which roughly the same arguments apply as to -ism/-ist, namely that a whole range of additional adjectival suffixes may precede -ize. Consider redundantize, privatize, probabilize, permeabilize, respectablize, terriblize, particularize, familiarize, seniorize, exteriorize. The fact that certain combinations (like e.g. -able-ize) are not often attested seems to be not due to a morphological constraint but rather due to phonological or pragmatic factors and token-blocking. For example visualize 'make visible' blocks visibilize (rare). The Latinate Constraint seems again operative to exclude bases with Germanic suffixes. Derivatives on the basis of adjectives in -ory and -ary are not attested at all, which is most probably due to the prosodic shape of such adjectives. Featuring a disyllabic unstressed suffix, they exhibit a three-syllable stress lapse (e.g. o̕rdinary, mándatory), which is strictly prohibited with -ize verbs.⁵ The intricacies of the phonological constraints are laid out in detail in and the reader is referred to that discussion for more detailed information.⁶

To summarize, the discussion of the last group of suffixes has shown that these suffixes are not more problematic than any others, since they are subject to the same kind of idiosyncratic, paradigmatic, and semantic pragmatic constraints as are all supposedly non-problematic ones.

¹ Sometimes the form involving -ous is lexicalized (cf. adventurous), sometimes doublets exist (medicamental (rare)/medicamentous/medicamentary), and quite often the -ous form has become obsolete in favor of the -al derivative (e.g. matrimonious).

² Furthermore, Fabb assumes that -ance may not precede -al, which, as we have just seen, is incorrect.

³ Fabb also mentions -y here, but adjectival -y is ruled out by the Latinate Constraint anyway.

⁴ The interaction of the domains of -ity and its rival systematic domains remains to be determined in a more exact fashion. The problem of -ity illustrates the above-mentioned general difficulty to account for the blocking effects between two word-formation rules with rival systematically restricted domains.

⁵ Deletion of the base-final vowel (as in forms such as summary - summarize) could reduce the stress lapse by one syllable. However, it seems that only vowel-final dactyls are truncated (cf. again súmmary - súmmarize). See The phonology of -ize derivatives. for a detailed discussion of stem allormphy effects observable with -ize derivatives.

⁶ The in-depth phonological analysis of verbal derivatives has led to a thorough revision of my earlier explanations for the apparent stacking restrictions involving -ize as given in Plag (1996:793).

مواضيع ذات صلة

Mirror in nominal morphology

Appendix 2: Hapax legomena from the Cobuild corpus

Appendix 1: 20th century neologisms from the OED

Morphological Productivity Structural Constraints in English Derivation Conclusion

Verbal derivation and the separation hypothesis

The separation hypothesis

Theoretical implications: Against the separation of form and meaning in morphology

Rival morphological processes 4: Where have all the rivals gone? Summary

Conversion vs. -ize, -ify and -ate

Rival morphological processes -ize vs. -ate

Rival morphological processes -ize vs. -ify

Rival morphological processes 4: Where have all the rivals gone?