The architecture of the english lexicon
Extrametricality and word-final syllable effects
Yüklə 2,22 Mb.
|
- Bu səhifədəki naviqasiya:
- NonFin
- NonFin as NonFin(Ft’) : (5.40) NonFin(Ft’)
- Edgemost
- NonFin( s ²)
- 5.3.1 Monomoraic words revisited
5.3 Extrametricality and word-final syllable effects It is relevant at this stage to return to the unaffixed words and to reexamine them under the expanded constraint hierarchy arrived at above. While extrametricality effects had previously been handled by NonFin and Edgemost, it will be necessary to further explain why certain unusual foot shapes can occur at the right edge of the word. For example, the simplified tableau offered above in (4.4) suggested that words of the shape /LH/ should appear with final stress, and indeed a small majority of all bisyllables (53% as in table 3.20) do. However, most of these words are verbs and adjectives which are to be understood as suffixed in /-æ/,89 and the percentage of nouns showing this pattern is smaller (42%). Of these, again a majority are deverbals of the type revŽrse, accl‡im, and so are actually suffixed deverbal forms of the shape /L(H)-æ/. The prevalent outcome for unaffixed words of this kind, once the schwa-affixed forms are removed from consideration, is actually /(LH)/, as in lŽgend, br’gand (which would need to be accounted for in any case). Such a violation of NonFin is expected in bisyllables with light initials, such as (c—met), because it is not possible to build a stressable foot from such stems without including the final. However, the candidates /(LH)/ and /L(H)/ both equally violate NonFin, which states that the head foot in the word does not appear finally in the word. Yet, the form /(LH)/, with initial stress, is clearly preferred to /L(H)/, although it violates FtForm (a candidate /(L)H/ would be precluded by Min-2). The solution to this problem lies in refining the constraint hierarchy with regard to word-final effects. The evidence suggests that there is actually a strong restriction on main stress occurring on the word-final syllable, and that a word-final weak (unstressed) syllable (in some cases dominating a deleted schwa) must always appear, except of course in monosyllables. This is not the same as restricting the final foot from coinciding with the word edge, the current definition of NonFin, since feet can be mono- or bisyllabic. This distinction can be addressed by proposing a constraint NonFin(s²), which may be formalized as follows:
The preceding constraint, however, would be satisfied by any number of syllables which intervene between the head foot and the right edge of the word, so long as it is at least one. However, it is clear from the data that the head foot in English tends to be as close to the right side of the word as allowed for by the NonFin constraints. Above, the number of syllables which could intervene between the head foot and the right edge of the word was normally limited to one by Edgemost, which was presented above as an Alignment constraint (¤ 4.1.1). This can also be re-formalized as a No-Intervening constraint, which penalizes syllables that intervene between the head foot and the edge of the word: (5.41) Edgemost: NI-R(MWd: Ft’, MWd, s) Note that this new formulation of the Edgemost constraint refers to the Morphological Word rather than the Prosodic Word (see ¤ 5.2.1). Applying this expanded constraint hierarchy to /brigand/ proceeds as follows: (5.42)
This revised constraint hierarchy does not affect previous conclusions about other types, but does account for trochaic feet of the sub-optimal shape (LH) occurring at the right edge of the word. This non-canonical foot can also occur elsewhere in the word. Again, the source of this situation appears to be the presence of a heavy syllable in an unstressable position. For example, when such a syllable appears before the main-stressed syllable, and a preceding light syllable is available, an (LH) foot appears in the optimal candidate, e.g., ˆlex‡ndrine, v˜luntŽer, ˆnalgŽsia, sˆturn‡lia, tˆrantŽlla. This can be enforced by modifying the NonFin(s²) constraint presented above in (5.39) to specify that the syllable following the stressed syllable should be stressless:
The adjacency of the two stressable syllables (e.g., *alx‡ndrine) results in one giving up its potential stress. Since main stress cannot move off the penult without violating Edgemost, it is the antepenult that is restricted from becoming the head of a foot. The same effect can be seen for prefixes, e.g., (rcon)(c´l)e. For trisyllables in this situation (e.g., (at)(l‡n)ta ), the antepenult remains footed, as there are no other syllables to join this foot. Due to its proximity to the head foot, such a foot can appear unstressed, showing an optionally reduced vowel (see ¤ 5.2), but it is nevertheless a foot. Regarding such a monosyllabic foot as stressless prevents the optimal candidate (at)(l‡n)ta, with no secondary stress on the initial, from fatally violating the NonFin(s²) constraint. Such stressless feet are further discussed in ¤ 5.4.1. Another word-edge phenomenon governed by the extrametricality constraints is seen in the treatment of some four-syllable stems, such as those seen in amŽrica, par‡meter. The most optimal candidate with regards to the Lapse-s constraint would be the unattested *(‡me)(rica), which foots all syllables yet does not treat the word-final foot as the primary stressed foot, which would violate the principle of non-finality. However, the gap between the head foot and the word edge is two syllables, incurring two violations of Edgemost. The attested type a(mŽri)ca thus succeeds despite its Lapse-s violations: (5.45)
5.3.1 Monomoraic words revisited At this point the revised hierarchy may be applied to the bisyllabic and trisyllabic unaffixed words discussed in chapters three and four. The expected stress patterns based upon underlying syllable weight will be laid out, and the minority patterns resulting from gemination and / æ/ suffixation will also be presented. Note that the issue of prefixation will be deferred until the next section. The set of bisyllables which are expected to surface in true monomoraic stems is as follows: (5.46)
Note that gemination in true bisyllables will not be detectable, as the penult will be stressed regardless. Cases in which apparent final stress appears will always be examples of suffixation in / æ/; where a stressed apparently light final is involved, both suffixation and gemination are to be understood, e.g., baguŽtte. Such words are suffixed and so possess internal morphology; in the table below, the symbol ‘H’ is used to cover both ordinary heavy syllables and proposed geminates: (5.47)
These account for all finally stressed bisyllables, which must show a heavy penult, geminate or not. Due to the fact that the suffix in question is a single light syllable, the patterns which appear in schwa-suffixed bisyllables are quite similar to those seen in trisyllables. The optimal trisyllables are similar to those yielded by the earlier, primitive hierarchy explored in (4.3): (5.48a)
Trisyllables with intrinsically light penults will always take initial stress, regardless of the weight of the initial or final syllables. Likewise, those with heavy penults will always take penultimate stress: (5.48b)
Trisyllables with any other pattern (for example, with final stress, or a stressed light penult) are understood as being suffixed in /-æ/. The optimal forms for these suffixed cases accord with the minority patterns seen for surface trisyllables. Again, the symbol ‘H’ refers to both typical heavy syllables and the proposed geminates: (5.49a)
Suffixed forms with heavy stem-final syllables will display final stress on the surface; underlyingly, these heavy stressed syllables are penults, as in the unsuffixed forms above in (5.48b). Suffixed forms with light stem-final syllables will show surface penultimate stress: (5.49b)
When the penultimate syllable in the stem is light, as in (5.49b) this results in a form which shows a stressed light surface penult, in contrast to the corresponding unstressed forms. When the penult is heavy, the surface stress pattern is identical to that seen in unsuffixed forms with the same stem type: (5.49c)
As the crucial constraints for determining prosodic structure involve interactions between the final three syllables, monomoraic words of greater length should be expected to show the same head feet, with secondary stresses feet iteratively projecting from the left side of the head foot. Heavy syllables adjacent to stressed syllables should surface either as weak members of feet (LH), or as unstressed monosyllabic feet (H), as was outlined above in ¤ 5.3. The investigation into the relative frequencies of these stress patterns, presented in ¤ 3.2, was relevant for determining both the proposed constraint hierarchy and the lexically determined morphological structures which are provided to it as "input". Presumably, the acquisition of the grammar works in the same way. OT constraint hierarchies usually yield up a single optimal candidate for a given structural type. Thus, if a language presents to the learner what appear to be different realizations for an apparently unitary type, the grammar must choose one as the unmarked case, the case in which the general constraint hierarchy applies and the "input" structure is transparent. It is sensible for the grammar to choose the most frequent pattern as its basic one, and this is what has been done here as well. Kager’s first pattern (that showing final syllable extrametricality) has been chosen as the basis for the development of the constraint hierarchy. Using that constraint hierarchy, which addresses the majority of forms, as the strict foundation for the grammar, the remaining forms are able to be understood as containing further structures (geminates, morphological structure) which allow them to also be accounted for using the same constraint hierarchy. The assumption behind the trends seen in ¤ 3.2 is that the high percentages are indications of 100% mixed with structurally marked cases, while the low percentages are indications of 0% likewise mixed. The degree of mixture depends greatly on the unusualness of the structural marking, which ranged from low frequency structures, such as gemination plus suffixation of / æ/ (Kager’s "idiosyncratic" third type), to relatively high frequency things such as the suffix / æ/ itself (present in Kager’s second type, which he ascribed to verbs and adjectives generally and characterized by loss of final syllable extrametricality). However, if some other pattern had been most frequent, or if such a pattern becomes the most frequent stress pattern of English, it would be expected that a reanalysis would take place of both the constraint hierarchy and the lexical forms which are fed into it. It is likely that this has happened numerous times during the development of English, and as can be seen from various alternant forms (¤ 6.3, ¤ 6.4, ¤ 7.4.1), regularization and adjustment of lexical forms according to the current hierarchy are still taking place in the present-day dialects of English. The preceding tableaux are applicable to stems consisting of a single morpheme. Some bisyllabic and trisyllabic words are however prefixed, and this internal morphological structure can have an effect on how words surface, especially with regard to mora augmentation, which is prohibited in all the above forms by Stem-Coh, since they are stems with two moras or greater. Some of the effects which can be seen in prefixed words are discussed in the next sections. Yüklə 2,22 Mb. Dostları ilə paylaş:
|