The architecture of the english lexicon
Extended forms in /-at-iv/
Yüklə 2,22 Mb.
|
- Bu səhifədəki naviqasiya:
- NonFin( s ²)
- 6.5 Other suffixes with word subcategorization
- Moraic Prominence
- Ft’ to MWd below NonFin( s ²)
- WdEdgemost
- Edgemost
6.4 Extended forms in /-at-iv/ The forms in / ative/ present a complex situation similar to that seen with / atory/. While words with monosyllabic stems all show initial stress, longer stems display three types of stress pattern, seen in both American and British English. Again, many words display more than one type, and the same words can show differing but overlapping stress patterns between the two dialects. The first pattern illustrated below shows a long / t/ syllable, while in the other two types that syllable is unstressed: (6.29) [1,0,2,0]: lŽgislaøtive [1,0,0,0]: n—minative [0,1,0,0]: prov—cative
As was the case with / atory/, the various forms in / ative/ will be accounted for via alternate morphological structures. Many words show variant pronunciations, e.g., c—mmutaøtive Ü commœtative, lŽgisla³tive Ü lŽgislative, qu‡ntita³tive Ü qu‡ntitative, c—ntempla³tive Ü contŽmplative, c—nnotaøtive Ü conno³tative, and the distinction between contrasting morphological structures in such words must be lexical. Most words with the first pattern, for example, show variants with the second. The first pattern is what would be expected if / at ive/ subcategorized like its individual parts, i.e., [{quantit}]{at}iv: (6.30)
However, the majority of forms do not exhibit this structure, and of those that do, most also present alternate forms with the second stress pattern given above in (6.29), i.e., qu‡ntitative, with stress three syllables from the right edge. Such words have been cited as evidence of ternary feet in English (Selkirk 1982b). They require an alternate structure, which is slightly different than that given above, and is not at all transparent with regard to the internal elements of the suffix, i.e., [{quantit}{at}]iv: (6.31)
The optimal candidate [(qu‡nti)(t)]tive is expected to show two possible surface forms. Since the second foot (t) is in an unstressable position (due to NonFin(s²)), it must be an unstressed foot and can surface either with or without vowel quality, as in other unstressed feet seen in prefixes and suffixes (¤ 5.4). In other words, forms showing the first two alternating types of stress can both be derived from such a structure; both surface forms are equally optimal. It is harder to explain the structure itself; nothing in the subcategorization of the elements /at/ and /iv/ seem to require a morphological structure [_{at}]iv, yet this structure yields the correct results under the current constraint hierarchy. Another subcategorization constraint specifically referring to / ative/ would be necessary to impose this structure on the candidates. The final stress pattern, that seen in prov—cative, is restricted almost entirely to stems made from prefixed monosyllabic roots. These form a stem with the / at/ element of the suffix, suggesting the / at2/ proposed above for similar stems in / atory/. The same distributional patterns apply to these roots. Such a subcategorization means that the root syllables will never lengthen, and thus / ative/ gave the appearance of being a "shortening" suffix (¤ 2.2.1). The structure of such forms would then be like [com{par at2}iv] (or [com{par at2}]iv):
No forms with lengthened stem vowels could surface, due to the high-ranking constraint Stem-Coh (¤ 5.2). Of course, underlyingly long vowels remain long here as elsewhere, e.g., conno³tative, and underlyingly long roots can show both patterns (e.g., c—nn¯taøtive, c—ntemplaøtive Ü contŽmplative) while true monomoraic roots cannot (e.g., comp‡rative Ü *c—mparaøtive Ü *c—mparative). All monosyllabic roots show stress on the root syllable, e.g., lu³crative, f—rmative, v—cative. All this suggests that the distribution of / at2/ might be not be purely lexical, as suggested above, but rather enforced by constraints interacting with the root type. It could be possible to propose that a number of ranked subcategorization constraints govern the suffix / at/:
Stem subcategorization: Align( at, R; Stem, R) = {_ at} These constraints conflict; as long as /-at/ remains outside the morphological word, it is not likely to form part of any successful stem constituent, as feet which cross the morphological word boundary will be dispreferred by a variety of the constraints in the current hierarchy, and perhaps others. But if /-at/ is forced into the morphological word by other constraints, it can then display its stem subcategorization. The interaction between root length and stem-formation would need to be explained: only monosyllabic roots join in this way with /-at/, while longer roots appear to remain in their own stems. Relatively low-ranking subcategorization constraints which interact with some prosodic constraints might be able to account for this; in any case, the relationships involved are complex and this is left to future investigation. A more conservative account of these facts would rely on an allophonic / at2/, which has a clear if lexically enforced distribution among monosyllabic roots. 6.5 Other suffixes with word subcategorization A number of other English suffixes also show the subcategorization for the morphological word seen in the cases reviewed above. This section will round out the discussion of such suffixes, highlighting the differences between them and true "level two" suffixes, which subcategorize for the prosodic word. One affix showing subcategorization for the morphological word is the nominal suffix / ment/. Stems taking / ment/ surface identically in words unextended by the suffix. The initial /m/ of the suffix fails to close preceding syllables, as it would if the suffix subcategorized for the stem: (6.34) b‡nishment mŽasurement m‡nagement g—vernment w—nderment sŽttlement In other words, the presence of successive consonants which cannot form an onset (e.g., /rm/, /ïm/, /lm/), due to the initial /m/ of the suffix, would be expected to lead to the projection of a mora (by Moraic Prominence) and a heavy penult. However, the penult in these forms does not receive the primary stress. Furthermore, monomoraic stems which show compensatory lengthening due to / æ/ suffixation in their verbal forms continue to show this lengthening despite that fact that the /m/ in / ment/ should close the stem syllable (as above), making these syllables bimoraic, without any need for further lengthening: (6.35) conce³alment ref´nement aba³sement Note that under the derivational approach proposed by Myers (1987) (see ¤ 2.2.3), which assumed underlying long vowels for such forms, these long vowels should have been shortened by "closed syllable shortening", which was the central rule of the shortening theory. Here, subcategorization for the morphological word constituent will instead account for these surface forms: (6.36) [{bani°}]ment [{govern}]ment [{setel}ment] [con{cel}æ]ment [re{fin}æ]ment [æ{bas}æ]ment Under the current constraint hierarchy, in such cases the penult can never be stressed, due to NonFin(s²). Thus *govŽrnment will never be optimal. For words in / æ/, the root syllable is not in this constituent-final position, because the /æ/ syllable is. Schwa fails to be parsed into prosodic structure in morphological word-final position, as was laid out in ¤ 6.3. Stems which never appear unsuffixed nevertheless adhere to the principles of word-formation enforced by the constraint hierarchy, e.g., d—cument , showing the structure [{docu}]ment. Some interesting effects occur when / ment/ interacts with other suffixes. A few words in / ize/, such as ch‡stisement, form nouns in / ment/. The suffix / ize/ also subcategorizes for the morphological word, and there appear to be alternate structures which can both appear optimally, [{chast}]izment and [{chast}{iz}æ]ment, resulting in the alternants ch‡stisement and chast´sement. The lexically optional presence of /æ/ in the structure appears to account for the difference. In a few other words, an element /a/ appears on the right of the stem, producing an unusual stress pattern, e.g., tŽmperament, mŽdicament, with structures like [{temper}a]ment, which would produce the attested forms. When extended with further suffixes, such as / al/, / ment/ can form a stem in itself, e.g., [(ex)(pri)(mŽnt)al]. The suffix / al/, in addition to requiring a stem, also must be inside a morphological word constituent, and the constraint requiring this evidently outranks the subcategorization of / ment/ for the morphological word. A similar effect is seen verbs extended in / ary/, which also compete with / ment/ for the morphological word, e.g., c˜mplimŽntary, with the structure [com{pli}{ment}æ]ary. Alternate forms like c—mplimentˆry suggest that / ment/ can also win this competition in some grammars. Verbs in / æ/ can be formed from / ment/ words, although any constraint which might subcategorize / æ/ for the morphological word is evidently not ranked so high as that for / al/, since the morphological word boundary apparently remains to the left of / ment/:108 (6.37)
However, the presence of /æ/ requires that / ment/ be stemmed, and the resultant vowel that surfaces is usually full (c—mpliment, verb) rather than reduced (c—mplimænt, noun). In the data in general vowels in stems appear less likely to reduce than those in affixes, outside the stem.109 For noun/verb pairs like sŽgment Ü segmŽnt, fr‡gment Ü fragmŽnt, it is not clear whether these words are being interpreted as forms in / ment/, or as stems {fragment}, {segment}. In the latter case, / æ/ suffixation in the verbal cases would yield the attested finally stressed verbal forms. In British dialects, the ranking of Ft’ to MWd below NonFin(s²) would correctly yield [(seg)](mŽnt)æ beside [(c—mpli)](ment)æ, suggesting that the monosyllabic stem forms can be interpreted as suffixed in / ment/.
The corresponding American forms, which should show initial stress (¤ 6.1), usually do, and forms with final stress, such as augmŽnt, tormŽnt, can be understood as not suffixed in / ment/, but as unitary stems (i.e., /augment/). Another suffix which subcategorizes for the morphological word is / able/. This suffix is similar in form and meaning to the / ity/ class suffix / ible/, and the segmental structure of the latter element in these suffixes is apparently /bil/, as becomes clear in forms that are further suffixed:
ch‡ritable ame³nable inŽvitable As with the other affixes of this kind noted in this chapter, / able/ can combine with word constituents ending in / æ/, e.g., [re{dem}æ]able, as well as unextended stems, e.g., [{arit}]able. Forms which show apparent shortening appear to be regarded by the grammar as suffixed in the homophonous / ibil/, e.g., fl‡mmable, impl‡cable, form’dable (which can alternate with f—rmidable), “nappl’cable (also in‡pplicable). Confusion in the grammar is further exacerbated by the many words which could well be suffixed in either / able/ or / ible/, e.g., commŽndable, delŽctable. A further type occurs, as with the suffixes / ary/, / ory/, / ment/ above, when the suffix / æ/ is lexically not included in the suffixed morphological word constituent: (6.41) [c—m{par}]able [irrŽ{voc}]able [irrŽ{fut}]able Alternants such as “rrevo³cable, “rrefu³table show that the grammar may also construct "input" forms like [irre{voc}æ]able. Reasons for this kind of alternation in the lexical forms will be further discussed in ¤ 7.4.2. As with the other suffixes of this kind, the combination of two suffixes subcategorizing for the morphological word results in alternate surface forms, e.g., [rŽ{cogn}]{iz}æable Ü [r{cogn}{´z}æ]able. It might be asked why words in / able/ do not surface with secondary stress on the suffix, i.e., *ch‡ritˆble, similar to nŽcessˆry. The best explanation of this is that the suffix is segmentally / æbil/, with initial schwa, and thus cannot be stressed. Similar situations exist among the suffixes / ist/, / ism/, some cases of / y/, and a number of other, less frequent suffixes. The distinction between suffixes subcategorizing for the morphological word, and those which subcategorize for the prosodic word (the level two suffixes), appears in stems which would interact with NonFin(Ft’), a high-ranking constraint which takes the prosodic word as its domain. An example is the alternation seen in tŽlephone Ü telŽphonist. The form tŽlephone, consisting of two stems, the stemmed prefix {tele} and the stem {fon}, is unable to be stressed on the rightmost stem due to NonFin(Ft’): (6.42)
Any form stressed on the final two syllables will violate the non-finality constraints. However, the final stem shows a lengthened vowel due to the requirement for stems to be footed, and feet to be bimoraic. Adding a suffix to the right of the morphological word, but still within the prosodic word, shifts all syllables in the word one syllable further from the right edge of the prosodic word, the area affected by NonFin(Ft’). Thus, the prosodic structure given to the identical morphological word will be different: (6.43)
The constraint WdEdgemost, which is the former Edgemost constraint from ¤ 5.3, replaced above in ¤ 6.1 in a refinement of the working constraint hierarchy, governs cases such as this, where two stems are present in the morphological word. Although a head foot on the right edge of both {tele} and {fon} would satisfy Edgemost (which only specifies that the head foot is aligned to the edge of some stem), WdEdgemost tries to align the head foot to the right edge of the morphological word. Although {fon} itself cannot take main stress due to NonFin(s²), the foot (lŽfon) satisfies the various word-edge constraints even though it divides a stem. Apart from cases like this however, suffixes of this kind which subcategorize for the morphological word will appear to have little effect on the prosodic structure assigned to the morphological word constituents in question, maintaining the consistent "look" of suffixed stems in these cases. Yüklə 2,22 Mb. Dostları ilə paylaş:
|