Spektator. Jaargang 4
(1974-1975)– [tijdschrift] Spektator. Tijdschrift voor Neerlandistiek– Auteursrechtelijk beschermd
[pagina 231]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A reanalysis of D-Deletion in Dutch
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 232]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
paper I will follow Smith's analysis as closely as possible, indicating where it fails to adequately account for the facts, and in doing so 1 will build up a reanalysis. Let us examine first of all the alternations which motivate Smith's rules of D-DELETION and J-INSERTION. First consider the forms derived by means of both rules.
Given no other data, a description of this phenomenon can be expressed quite straightforwardly in a rule which replaces d by j in between a tense vowel and word boundary (≠)Ga naar eind1. followed by a reduced vowel, and deletes d in between diphthong with a lax vowel part and again word boundary followed by reduced vowel. Since diphthongs with a tense vowel part do not appear within the same morpheme before final d, and since lax vowels will definitely have to be excluded (cf. non-alternating forms such as vodden ‘rags’, hadden ‘had, pl.’, etc.) the most economical way of stating the rule appears to be as in (2). I will refer below to this rule as D-WEAKENING, also for the cases where it deletes d. Angled brackets abbreviate the two subrules required: j replaces d after tense vowels, and d is deleted after diphthongs.Ga naar eind2. Interpreting <j> as j in the longer subrule and as Ø in the shorter appears to be a natural consequence of the role of angled brackets as a notational device in the theory of phonology. The employment of these brackets makes superfluous Smith's rule of GLIDE ASSIMILATION, which applied, for instance, in the derivation of gouwen (/xawd= ən/→ /xawj=ən/-/xaww=ən/) and a rule of DEGEMINATION of glides not stated by him, which would have to send, for instance, /xaww=ən/ → /xaw=ən/, and also /krʌjj= n/→ kr j=nn/, and so on. The negative condition on lax vowels (for similar proposals for the phonologies of English and Menominin see Hoard, 1971, 1972) takes away the need for positing ‘unitary’ vs. ‘non-unitary’ long vowels, a markedness convention stating that the unmarked length for tens vowels is long, and a Dutch-specific rule of HIGH TENSE VOWEL SHORTENING, which requires 9 features. Smith invokes this machinery to arrive at an economical statement of the left-hand environmental condition of his rules. Furthermore, there is no need to require that the vowel or diphthong be stressed, since there are no crucïal examples to show that the rule does not operate after unstressed vowels or diphthongs. There are, however, other relevant data, and, as Smith argues, these show that the above rule of D-WEAKENING should be halved into rules of D-DELETION and of J-INSERTION, which both happen to be independently motivated in Dutch phonology. This is shown, he claims, by forms such as those in (3), where in an environment very similar to the one where a də/jə alternation holds in (i), a də/Ø alternation is found; and by forms such as those in (4), where Ø alternates with j, again in a very similar environment. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 233]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Thus it seems indeed reasonable to propose that D-WEAKENING should be separated into a rule deleting d and a rule inserting j. D-DELETION wilL then delete d fromr both the forms in (1) and (3), while J-INSERTION would insert j after tense vowels in both the forms of (1) and (4). Before we surrender, however, let us examine more closely paradigms (3) and (4), in that order. In relation to the forms of (3), Smith himselfadmits in the first place that ‘...the [də/jə alternation] is quite productive, whiL[e the [də/Ø alternation] is decidedly unproductive.’ (p. 423). Next to this distinction in productivity, which in itself seems sufficient reason to speak of two different processes, we may observe that one cannot say that the deletion of d with deletion of schwa, as exemplified by the forms in (3), depends on style, context, colloquialness, etc., in the same sence D-WEAKENING does. A handful of examples will suffice. Of the alternations vader/vaar ‘father’, moeder/moer ‘mother’, and broeder/broer ‘brother’, vader, moeder and broer are normal Modern Standard Dutch. Broeder is very formal in the sence of ‘brother’, but it is the only variant allowed in the senses of ‘friar’ and ‘male nurse’ and in compounds such as broederdienst ‘brother's service’. broedermoord‘fratricide’, and so on. Moer and vaar occur in id ioms (loop naar je malle moer‘get lost’; een aardje naar z'n vaartje hebben ‘take after one's father’); they are the only forms allowed in compounds such as moervos ‘vixen’, parelmioer ‘mother of pearl’ and bestevaar ‘granddad’, although moeder and vader are obligatory in grootmoeder, -vader ‘grandmother, -father’ and stiefmoeder, -moeder ‘stepfather, - mother’. Moer is the only form possible with the meaning of'female screw, nut'. As Van Haeringen (1963) has noted, mede- is generally used with nouns if not derived from verbs. Both mede- and mee- are used with verbs, but in this case mede- has the meaning of ‘in cooperation with’, while mee-approximates the meaning of ‘also, with’. A clear difference in meaning exists in pairs such as the following: vergaderen ‘assemble’/vergaren ‘gather’; buidel ‘pouch’/buil ‘lump’; ijdel ‘vain’/ ijl ‘thin’; beademen ‘breath on’/beamen ‘affirm’; gevederd ‘feathered’ geveerd ‘pinnated’, and so on. Given these idiosyncracies, and given the fact that no such idiosyncracies are the results of D-WEAKENING, it again seems much less than obvious that the same rule of D-DELETION is involved in the derivations of the forms of both (1) and (3). Thirdly, an ordering anomaly results if only one rule of D-DELETION is allowed. As is clear from examples such as kruiden ‘herbs’, such a rule would have to follow the rule of plural affixation. Alternations such as lade/la, however, require the ru le to precede plural affixation, since the respective plurals are laden and la's (see also below). 1f two separate rules exist, one can be ordered before, and one after plural affixation. Curiously enough Smit recognizes all these arguments in one manner or other, but instead of arriving at the conclusion that two rules will have to be posited, he | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 234]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
counters each one of them with what are clearly ad hoc solutions. He proposes to account for the productivity distinction by adding to the grammar two redundancy rules, one of which almost completely repeats the structural description of J-INSERTION. Secondly, the peculiarities of d-deletion cases must be listed in the lexicon. This is obvious, but does not release one of the duty to analyze what is really going on as accurately as possible. Thirdly, the ordering anomaly must be solved either by a derivational constraint, or by adding an extra environment to the rule of plural affixation. In view of this we cannot but conclude that everything seems to indicate that a rule deleting d must be present in the grammar of Modern Standard Dutch independently of the rule of D-WEAKENING (2). Since this is so, we also must conclude that it remains to be shown that D-WEAKENING must be separated into a d-deletion rule and some other part(s). It is not immediately clear that the appropriate manner to express the alternation in (3) is a rule deleting d in the environment [-cons, -tense] -, followed by a schwa-deletion rule. In the first place I know of no independent motivation for Smith's schwa-deletion rule, plausible though the rule may seemGa naar eind3. Thus it seems more appropriate to posit a rule deleting -də (DE-DELETION below). Secondly, this rule wille have to be generalized to the point where E is a centra] vowel, given alternations such as weduwe ‘widow’/ weeuw ‘(young) widow’, and for some people also schaduw/schauw ‘shadow’. Thirdly, note that the facts of the language as presented here do not force a choice between an DE-DELETION rule on the one hand, and a DE-NSERTION rule on the other. Actually, given the fact that the rule operates on a relatively restricted class of forms, featurecounting will select DE-INSERTION over DE-DELETION, since it is obviously cheaper to insert də into underlying forms without it, than to delete də from underlying forms with it. I will return to this below. But whatever the appropriate device to deal with these alternations will turn out to be, it has hopefully been shown that it will be irrelevant to the facts of D-WEAKENING. Let us turn next to the forms in paradigm (4). The paradigm has been divided into two subclasses, which will be dealt with one after the other. In the first place, the j in the plural forms of (4a) can be explained by means of the rule which inserts a homorganic glide in between vowels. Smith is of course aware of this rule (p. 435, n. 9), but does not relate it to paradigm (4). Thus the forms of(4a) can adequately be accounted for even in the absence of Smith's rule of J-INSERTION, ceteris paribus. We may note that his verbal statement of the rule is incorrect in two ways. He writes: [it inserts] glides between high vowels and following unstressed vowels as in Ria (girl's name) [rījā], Mantua (Mantua) [mantüwā]. (p. 435, n. 9)Ga naar eind4. But the rule not only inserts glides after high vowels, but after non-low ones, cf. Lea (girl's name) [lējā], Samoa (Samoa) [samowa], and so on. Secondly, the application of the rule does not hinge on an following unstressed vowel, as is shown by examples such as idioon ‘idiom’, hiaat ‘hiatus’, poëet ‘poet’, and so on. This gives the following formulation of the rule, which although no revision will follow in this papier, can only be considered a first approximation. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 235]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
This leaves us with the only possibly crucial forms of the type (4b), vlooien and koeien, forms which have a non-low back stem-vowel. To be fair we may add to this couple the mass noun stro ‘straw’ which has the derived adjective strooien, but to the best ofmy knowledge these three items exhaust the list. Thus the evidence for a rule of J-INSERTION in Dutch phonology is reduced to three items. Even for these last three forms, however, Smith's analysis collapses. His analysis has as a consequence that koe and vlo will have to select the nasal plural affix, which is exceptional for monosyllabic nouns ending in a back vowel, cf. gnoe's ‘wildebeests’, soe's ‘farthings’, clou's ‘points (of a story)’, po's ‘poes’, do's ‘does’, ro's ‘roes’, sjah's ‘shahs’, roe's ‘rods’, la's ‘drawers’, ka's ‘quays’, and so onGa naar eind5.. If the forms which enter plural affixation are taken to be / koei-pl/ and /vlooi-pl/, on the other hand, these will select the nasa] affix regularly, cf. fooien ‘tips’, kooien ‘cages’, boeien ‘buoys’. As underlying forms /koei/ and /vlooi/ (and /strooi/) have independent motivation, since they meet the otherwise valid morpheme structure condition that monosyllabic noun stems of the native portion of the lexicon cannot end in a vowel, that is, end in either a consonant or a glide. In a preliminary way this may be expressed as in the negative condition (6).
Note in this respect that gnoe, soe, clou, po, ro, do, and sjah are all non-native; as far as 1 am aware they exhaust the list of forms of this type. If constraint (6) is valid, it also forces the choice between a DE-DELETION rule and a DE-INSERTION rule for the alternations in lade/la, kade/ka, roede/roe, slede/slee, and so on, in favour of deletion: the bisyllabic forms will now be underlying (consequently, in order to preserve consistency the same DE-DELETION rule will be posited for the alternation in moeder/moer, veder/veer, and so on, although these alternations are neutral with respect to constraint (6)). Forms like ree ‘roe’, spie ‘pin’, and drie ‘three’, and a few others, belong to the small class of exceptions to constraint (6). Thus the underlying forms posited here conform to an independently needed constraint (which they violate in Smith's analysis) and they explain both the glide in the forms koeien, vlooien, and strooien, and the plural affix in koeien and vlooien (which remains unexplained in Smith's analysis). To arrive at koe, vlo and stro a rule of final glide deletion will have to be posited, but this rule may be needed in the grammar anyway to account for the alternations found in zee ‘sea’! Zeeuw ‘Zealander’, Hebreeër ‘Hebrew, pers.’/ Hebreeuws ‘Hebrew, adj.’, Chaldeeër ‘Chaldean’/ Chaldeeuws ‘Chaldaic’. Earlier I argued that the forms with d- and schwa-deletion as in (3) cannot be claimed to show that D-WEAKENING must be separated into rules ofd-deletion and j-insertion. The same has now turned out to hold for the forms in (4). All forms which Smith takes to motivate his rule of J-INSERTION can be explained also in the absence of this rule from the grammar of Modern Standard Dutch, without loss (End actually with some gain) of generalization. 1 thus conclude that the grammar of Modern Standard Dutch has one rule of D-WEAKENING for the forms of (1), the present rule (2). The above forms the principal part of my reanalysis of ‘the phenomenon of d-deletion in Dutch’. 1 have argued for a rule 1 called D-WEAKENING for the alter-Ga naar eind6. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 236]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
nations as found in goede/goeie, rijden/rijen, and so on. I have rejected Smith's (1973) evidence for breaking up this rule into rules of D-DELETION and J-INSERTION, mainly on the basis that his evidence can be explained independently in J-INSERTION cases, and that d- with schwa-deletion is a marginal and idiosyncratic phenomenon, while D-WEAKENING, given its exact conditions, appears to be fairly regular. Moreover, an ordering paradox is solved if d can be deleted by two different rules. I am convinced that the new analysis has flaws in various placesGa naar eind7., not in the least since it is based on a very shallow analysis of the language. In general, however, the account seems to me to be a not unreasonable first approximation. In the second part of this paper I will discuss varjous troublesome phenomena, for which Smith contemplates the devices of ‘positive exception feature’ and derivational constraint, both peeking and transderivational. For the cases of derivational constraints and one class of positive exceptions I will offer a reanalysis within the standard theory of generative phonology. One class of positive exceptions will remain as an unexplained residue. First consider the fact that the present analysis, contrary to the one proposed by Smith, has a rule of D-WEAKENING and a rule of DE-DELETION. As noted above this has as a consequence that an ordering anomaly is eliminated, for which Smith contemplates a ‘peeking’ derivational constraint. The problem was that lade has the plural laden, but la has la's. But "at the time when the morpheme - plural- is rewritten as /s/ or /ən/, la still has the form lade’ (Smith, p. 432) Thus we somehow have to know beforehand (i.e. we must ‘peek’) that lade at a stage in the derivation after plural affixation becomes la. This problem is simply non-existent, as noted above, if we ensure that DE-DELETION applies before plural affixation. Thus we also correctly derive roe's, la's, sneeën, sleeën, an so on. Secondly, i fail to see why the device of transderivational constraint should be invoked to ‘explain’ differential behaviour as regards the choice of plural affix for nouns of the lade-type, which take -en, versus the type palissade ‘palisade’, methode ‘method’, periode ‘period’, which take either -en or -s. Quoting Smith, this constraint would have to read: ...if another derivation of a word results in D-DELETION, then in any derivation that does not involve D-DELETION, the plural morpheme (if present) must be rewritten as/ n/. (p. 433) Apart from the fact that this constraint predicts incorrect plurals for forms with morpheme final consonants, such as broeder, moeder, buidel,, and so on, which all take -s, we may note that all nouns of the bolide type belong to the Romance (say, non-native) portion of the lexicon. This feature will explain the behaviour of the two subdivisions as regards plural formation in an adequate manner. It is also reasonable to suppose that this feature exempts non-native words from the rule of D-WEAKENING, since the rule never seems to apply to a member of this class. If the analysis proposed here is accepted, Smith's transderivational constraint, just as his ‘peeking’ derivational constraint, is without motivation in this area of Dutch phonology. Thirdly, consider forms in which Modern Standard Dutch allows D-WEAKENING, while no internal boundary is present as required in the structural description of the rule. All examples I know of are the place-names Leiden, IJ, Lei-) Muiden, and St. Truiden, the common nouns poeder ‘powder’, schouder ‘shoulder’, and for some people also bodem ‘bottom’, and the adverb heden ‘today’. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 237]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Of these Smith only provides poeder. To mark these forms idiosyncratically as somehow undergoing D-WEAKENING while not meeting its structural description (i.e., as ‘positive exceptions’ to D-WEAKENING, essentially similar to the cases for which Harms, 1968, p. [19, proposed his ‘structural description features’) is to deny their common structure, and their close resemblance to the forms of (1). Yet they are exceptional in that most forms of their type do not undergo D-WEAKENING. I have as yet no clear analysis to offer for them. Finally, let us consider somewhat more detailedly the cases where ‘D-DELETION has analogically spread to forms where the phonological environment for D-DELETION is not present’ (p. 428). By way of introduction, consider the following paradigm.
This formulation, however, does not bring out the structural similarity of this rule and the rule of D-WEAKEN ING (2). But since nothing will prevent us from ordering (2) and (8) adjacently, the two rules can be collapsed into a schema (see below). Given this readily available alternative, and given the fact that the phenomenon exemplified by the forms in (7b) is fairly productive, I am at a loss why Smith should want his D-DELETION rule and J-INSERTION rule to apply exceptionally in the derivations of these forms via ‘positive exception features’. As a conseque:ce of his analysis Smith has to express the facts of(7b) by means of two, as far as his analysis is concerned completely unrelated redundancy rules. One, Smith's rule RR3, states that the forms with a stem-diphthong, such as rijd and houd, may exceptionally undergo D-DELETION. The other, which he does not formulate, has to state that forms such as laad, raad, and bloed can exceptionally undergo both D-DELETION and J-INSERTION. An immediate consequence of the introduction of rule (8), collapsed with (2), is the elimination of the need of this pair of rules in the grammar. D-WEAKENING will have to undergo one more revision. Note that the forrns of (7) all have ei.her a stem-diphthong or a back stem-vowel. Comparable forms with front stern-vowels differ from these in that the (b)-type of forms lack the final glide, cf.:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 238]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Smith comments on these forms: Thus we assume that J-INSERTION will only apply exceptionally to singular forms whose corresponding plurals normally undergo it when the resulting sequence is one that matches those strings possible in underlying forms, eg. /aj/. An alternative explanation would be that J-INSERTION does in fact analogically apply and that the resulting impossible sequence wil be corrected by later rules. In the first place we would have a derivational constraint, in the second a conspiracy. (p. 432) Thus the choice seems to be between a derivational constraint, or a new rule which ‘conspires’ with others. Again, however, there is a solution which stays well within the standard theory. The problem can be solved if the insertion of j via D-WEAKENING is limited to back vowels, that is, if D-WEAKENING is formulated as in (10).
In forms such as deeën, reeën, sneeën, gleeën, breeër, and so on, j will now be assumed to appear via HOMORGAN]C GLIDE INSERTION, rule (5). This gives some of the following derivations.
I finally conclude that the facts of Dutch D-WEAKENING, although they Jeave one as yet unexplained class of ‘positive exception’, are thus shown not to support the notion of derivational constraint, either peeking or transderivational. The cases for which Smith (1973) proposed these constraints can be explained within a standard theoretical framework, without loss in explanatory force. Until further notice, we can not but hope that the situation described in this paper is typical in Dutch phonology. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
10. Appendix: rules(6) ~ + c [v/+N/+native] + DE-DELETION
PLURAL AFFIXATION
GLIDE DELETION | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 239]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
[pagina 239]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
I am grateful to M.K. Brame, H. Gilijamse, and H. SchuItink for valuable criticism on earlier versions of this paper. Needless to say that its contents have only benefited from their suggestions. | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
BibliographyChomsky, Noam. and Morris Halle 1968. The Sound Pattern of English. Harper & Row, New York Haeringen, C.B. van 1963. ‘Weder en weer, neder en neer, mede en mee’. MKNAW. Afd. Lett., N.R., dl. 26:8, pp. 259-80. Harms, Robert T. 1968. Introduction to Phonological Theory. Prentice-HaIl, Englewood Cliffs, New Jersey. Hoard, James E. 1971. ‘The New Phonological Paradigm: The SoundPattern of English’. Glossa 5:2, pp. 222-68. Hoard, James E. 1972. ‘Naturalness Conditions in Phonology, with Particular Reference to English Vowels’. in Michale K. Brame (ed.), Contributions to Generative Phonology. Texas University Press, Austin, pp. 123-54. Smith, N.S.H. 1973. ‘The Phenomenon of D-DELETION in Standard Dutch’. Spektator 2:7, pp. 421-37. |
|