The architecture of the english lexicon
The treatment of prefixes
Yüklə 2,22 Mb.
|
- Bu səhifədəki naviqasiya:
- 5.4.1 Prefixes in suffixed words
- Stem-Coh
- NonFin( s ²)
- Non-Fin(F’)
- Non-Fin( s ²)
- Edgemost
- NonFin(Ft’)
5.4 The treatment of prefixes From the preceding sections it should be clear that prefixes such as /re /, /de-/, /pro-/ can surface in three different ways. For example, the suffix /re-/ can appear as [re³-], [rŽ-] or [ræ-]. This is interpreted here (following the arguments presented above) as due to the underlyingly monomoraic structure of the prefix. The lexical form of such prefixes is of the type /re-/, with a short vowel. When such a prefix is part of a foot, in the head position, as in (rŽsi)dent, the short vowel is stressed and surfaces with its full quality. When such prefixes are alone in the foot, they are lengthened as is permitted by Stem-Coh, and surface as long vowels, e.g., the noun (re³)ject. Finally, as was noted above in ¤ 5.2, when such prefixes appear alone in a footable non-main-stressed position, they may optionally appear with long or reduced vowels, but never with short full vowel quality, as in (reø)(prös)al Ü (ræ)(prös)al. The behavior of such prefixes depends entirely upon how it is footed and the status of its foot. 5.4.1 Prefixes in suffixed words The treatment of prefixes in words suffixed in /-ent/ has been detailed above (¤ 5.1.2), but the treatment of words in /-al/ needs to be slightly revised. Under the current constraint hierarchy (shown with relevant constraints only), the optimal candidate for the structure /re-pris-al/ is actually (r«)(prös)al , a candidate that was not offered in previous tableaus (cf. 5.22): (5.50)
Adding an intrusive mora to the prefix in the optimal candidate does not violate Stem-Coh, and produces the best candidate in terms of Lapse-s violations. However, there is an alternation between the prefix surfacing with a full long vowel, and with an unstressed schwa, e.g., (ræ)-(prös)al, which appears to be optional, perhaps related to rate of speech. Such optional alternations in vowel quality are seen in other unstressed feet, like the / ent/ suffix (e.g., (prŽsi)(dænt) Ü (prŽsi)(dent)) and in forms like (c—m)(pæn)(s)te Ü (c—m)(pen)(s)te. Heavy monosyllabic prefixes in these positions can also exhibit an alternation between full short quality and reduced quality, e.g., (con)(grŽs)sional Ü (cæn)(grŽs)sional. Long vowels in such positions appear either in full or reduced forms, but never with full short vowel quality, e.g., (c˜n)(n¯t)(ta³)tion Ü (c˜n)(næ)(ta³)tion. The behavior of monosyllabic prefixes in this manner is consistent across a variety of word types, the underlined vowels all appearing optionally long or reduced: (5.51) (pro)(cŽs)sion (de)(l“be)(ra³)tion (de)(te³ri)o(ra³)te (pre)(d—mi)(nant) (pro)(h’bit)ive In positions where the prefix is not adjacent to the stressed stem, it will always show full stress, e.g., (preø)(con)(ce³i)ve, where the vowel in the prefix /con-/ is reduced but the vowel in the initial syllable (pre³-) is footed and stressed, as it is not adjacent to a stressed syllable. This again suggests the "stressless" foot, seen above (¤ 5.3) in (at)(l‡n)ta, which optionally fails to display vowel quality in its single syllable. These feet would be predicted to appear at the word-edge, or between two other foot heads, as a result of NonFin(s²). That is, the reduction in quality in a syllable is related to the adjacency to other stresses. For example, while the suffixes in prŽsident, rŽsident may optionally appear with the full vowel [ƒ], those in prop—nent, resp—ndent, where the suffix follows the main stressed syllable, are always reduced. The prefixes which reduce always precede stressed syllables, and the reduced syllables in the other cases noted above are also adjacent to stressed syllables. These reduced feet should not be regarded as defooted but only as unstressed, and the optional reduction is in line with that of other unstressed syllables.91 A further constraint is required to restrict the proliferation of these unstressed feet; otherwise, forms such as *a(lek)(s‡n)(drine) would surface as optimal candidates: (5.52)` Lapse-s¡: NI(MWd: s¡, s², s¡) This states that all unstressed syllables are adjacent to stressed syllables, and will be violated by *a(lek)(s‡n)(drine) but not (at)(l‡n)ta or the other examples given above. Stressless feet should only appear when enforced by the various Non-Finality constraints. (5.53)
In words where multiple unstressed syllables do appear adjacently, e.g., —rigin, higher ranking constraints such as Non-Fin(F’) enforce these optimal forms despite the violation of the lower ranked Lapse-s¡. 5.4.2 Multisyllabic prefixes and constituency There are a number of multisyllabic prefixes, and although they are not as common as the monosyllabic ones mentioned above, many behave in a similar way. For example, prefixes like /omni-/, /ambi-/, appear to subcategorize for a stem on their right; they align their right edges to the left of the stem, and can become part of the stressed foot under the appropriate conditions: (5.54) (ˆmbi)({dŽx)tr}ous (ˆm)(b’{va)(l}ence) (˜m)(n’{po)(t}ent) When the root is able to form a foot on its own, as in ˆmbidŽxtrous, the prefix stands in its own secondary foot. But when the root is light, it requires the light final of the prefix to comprise the necessary bimoraic foot, as in amb’valence and omn’potent. Due to the presence of further syllabic material supplied by the prefix, the lengthening seen in the shorter form po³tent is unnecessary in omn’potent. (5.55)
More common among multisyllabic prefixes is the behavior seen by /inter /, /super / and /hyper /. These prefixes appear to behave as if they were morphological words in themselves, and indeed sœper and hy²per can appear as independent words. For example, in nouns like ’nterview, ’ntercom, sœperman, sœperstar the main stress falls on the antepenult, rather than the penult, which should be heavy, since it is a closed syllable. By regarding the prefix as a morphological word, this behavior can be explained using the current constraint hierarchy: (5.56)
The constraint Non-Fin(s²) prevents the main stress from falling on the final syllable in the given morphological word constituent; another syllable in the same morphological word must follow. Since [(in)(ter)] forms a morphological word constituent, its main stress cannot fall on its second syllable, which is morphologically word-final, despite its weight in that position. The final syllable of the prosodic word, which itself forms a second morphological word constituent, fails to take main stress because of the stronger restriction imposed by NonFin(Ft’), whose domain is the prosodic word. The final syllable in the prosodic word may not take main stress. Of the possible candidates, ’nterview causes the least violations of the various non-finality constraints. In most words, the second syllable of /inter-/ fails to be stressed, even when followed by a light, unfooted syllable, e.g., “nterdepŽndent, “ntercollŽgiate. The restriction on stress in that position remains, even when the stress in question is not the main stress in the word; it will always be the main stress within the morphological word constituent /inter-/, and that is the domain addressed by the constraint Non-Fin(s²). There are a minority of words, however, in which prefixes of this kind are in fact not behaving like morphological words, e.g., intŽrcalary, intŽrpolate, intŽrpret, supŽrfluous, supŽrlative. For words in /inter-/, it is easy to parse an alternate structure prefixed in /in-/, i.e., /in-tŽrcal-ary/, /in-tŽrpol-ate/, /in-tŽrpret/. In other cases, if the sequences in question were interpreted as prefixes, the stems would be defective, e.g., */super-l-ative/. In other cases, the prefix is clearly behaving as a stem, and is further suffixed, e.g., supŽrior, intŽrior, intŽrnal, supŽrnal. Thus, the sequences in question do not necessarily subcategorize as morphological words, although in most cases they do; the process is lexical. Similarly, in a few cases prefixes like /ambi-/ appear to subcategorize as morphological words, e.g., ˆmbisyll‡bic. The question of how this subcategorization should be handled in an OT grammar will be addressed in ¤ 7.1.1. Words with multiple morphological word constituents provide further information for fleshing out the constraint hierarchy. In previous examples, words contained only one morphological word constituent, and so there was no issue regarding which morphological word would contain the main stress. In forms like ’nterview, the rightmost morphological word, although it contains the "root" rather than the prefix, is restricted from taking main stress by NonFin(Ft’), which applied over the domain of the prosodic word. But in other words where the rightmost morphological word consists of more syllables, there is no mechanism by which this choice can be made: (5.57)
Clearly, in such cases the rightmost morphological word accepts the main stress. This implies a second Edgemost constraint, this one applicable to the domain of the prosodic word, rather than the morphological word:92 (5.58) P-Edgemost: NI(PrWd: Ft’, PrWd, s) This constraint is ranked below the other relevant constraints; in previous cases it would have had no effect, as the Edgemost constraint, effective in the domain of the morphological word, would have always superseded it in words with a single morphological word constituent. (5.59)
There is one further effect seen due to the presence of two morphological word constituents in such words. Bisyllabic stems, when surfacing alone in the prosodic word, are forced to eliminate their final syllable from the main foot whenever possible, in order to avoid violating NonFin(Ft’). However, when such stems make up the second morphological word component in a word, it is possible for the first morphological word, the prefix, to take the main stress, leaving the second stem free to be footed entirely. As a result, such words are stressed on the prefix; example include sœpermˆrket, hy²permˆrket, sœpers˜nic: (5.60)
While (m‡r)ket is the most optimal simplex form, with a prefix that forms a morphological word it can appear as a secondarily stressed -(m‡rket). When stems are suffixed, as in “nterpŽrsonal, this effect, which is due to NonFin(Ft’), is removed since the stem is no longer word-final, and such suffixed forms show main stress on the final, root morphological word constituent, enforced by P-Edgemost. This series of constraints will also affect stress in compound words, and compounds where the final member forms a stem that would violate NonFin(Ft’) are stressed on the first member, e.g., v’deotape, pol’ceman, Žlderbrry. Suffixed compounds take stress on the second member, e.g., nymphom‡niac. Monosyllabic prefixes can also form morphological words, although again subcategorization for this appears to be lexical. Examples of this include d«-cont‡minate and r«-crea³te (cf. rŽcreaøte; this is a minimal pair, subcategorization of the prefix being the only difference). Such suffixes tend to have a specific meaning, rather than being virtually meaningless (i.e., /re / ‘repeat’, /de / ‘undo’, /pro / ‘in favor’). This can be understood as a lexical subcategorization, and these meaningful prefixes might even be viewed as separate lexical entries, subcategorizing for the morphological word, simply homophonous with their meaningless counterparts. Yüklə 2,22 Mb. Dostları ilə paylaş:
|