Menu
Home
Log in / Register
 
Home arrow Language & Literature arrow Word and paradigm morphology
Source

Block order and the status of portmanteaux

In a standard realizational model, the organization of rules imposes a clear division of labour. The grouping of rules into blocks permits a description of the sequential structure of a form. A disjunctive ordering condition contributes to an effective procedure for determining which of a set of rules is applicable at a given point in an analysis. While, as noted above, there is no compelling evidence that disjunctive

Table 6.12 Partial future paradigm of Swahili taka ‘to want’ (Ashton 1944:70ff.)

IV

Positive

III

V

Negative IV III

1sg

ni-

ta-

taka

si-

ta-

taka

2sg

u-

ta-

taka

ha-

u-

ta-

taka

3sg

a-

ta-

taka

ha-

a-

ta-

taka

1pl

tu-

ta-

taka

ha-

tu-

ta-

taka

2pl

m-

ta-

taka

ha-

m-

ta-

taka

3pl

wa-

ta-

taka

ha-

wa-

ta-

taka

ordering applies across blocks, other types of disruptive patterns are more robustly attested. Cumulative exponence is the source of one type of block-spanning relation. The cumulative realization of features presents no difficulties at the level of individual rules. The problems arise if some of the features spelled out cumulatively by oneruleare spelledout separately, indifferentblocks, by otherrules. Theanalysis of the Finnish paradigm in Figure 6.2 averts this situation by exploiting the fact that rule blocks need not be morphosyntactically coherent. The rule that realizes nominative and plural features by -t can thus be placed in the same block as the rule that realizes plural features by -i, rather than in the block that realizes the case features. The more specific nominative plural rule blocks the plural rule and defines the form talot ‘houses’, which remains unmodified in the ‘case’ rule block.

However, this solution only works because there is no nominative rule in the second block. In other cases, multiple features from a cumulative exponence rule are realized in different blocks. As Stump (1993c: i44ff.) shows, the portmanteau marker si- in Swahili presents a case of this kind. The agreement features realized in position ‘IV’ of the positive conjugation in Table 6.12 include a lsg marker ni-. Negative features are realized in position ‘V’ bythe marker ha-. Yet in the negative conjugation, 1sg negative features are not realized by the sequence *hani-, but instead by a single marker si-. As Stump (1993c) notes, the rule that introduces si- must block both the rule introducing ha- in block ‘V’ and the rule introducing ni- in block ‘IV.

In formulating an analysis of this pattern, Stump (1993b: 144) proposes initially that “si- simultaneously occupies both slots”. However, he goes on to implement this proposal in terms of a slightly different intuition, by treating si- “as the sole member of a portmanteau position class pre-empting slots V and IV”. This structure is exhibited schematically in Figure 6.23.[1]

In this structure, the rules introducing ha-, ni- and ta occur, respectively, within the blocks ‘V’, ‘IV’ and ‘III, whereas the rule introducing si- occupies a portmanteau block that spans blocks ‘V’ and ‘IV’. This organization permits the rule introducing

Portmanteau ‘isg negative’ rule block in Swahili

Figure 6.23 Portmanteau ‘isg negative’ rule block in Swahili

Total rule ordering in Finnish

Figure 6.24 Total rule ordering in Finnish

si- to block the rules introducing ha- and ni- in the realization of any feature bundle containing isg negative features.

The organization of rules in Figure 6.23 establishes the desired blocking relations. Yet mediating these relations through ‘portmanteau blocks’ amounts in effect to relaxing the usual assumption that rule blocks must be totally ordered. The idea of a block-spanning portmanteau reinforces the impression that blocks represent an independent level of structure which is associated with sets of rules, much as the ‘timing units’ of a phonological description are associated with segments. However, blocks need not be treated as objects, but can instead be understood as generalizations over ordering relations among rules. On this conception, a block consists of a set of rules that are (extrinsically) unordered with respect to each other and which all stand in the same ordering relations to rules outside the set. Hence the blocking relations exhibited by markers like si- do not necessarily call for a novel type of block, but can instead motivate a reconsideration of the ordering relations that blocks represent.

The rules in a realizational model are implicity ordered by a strict partial (irreflexive, asymmetric and transitive) precedence order Th’. The extrinsic order defined over rules must be partial, to permit rules that apply at the same point in an analysis to remain unordered. These extrinsically unordered rules are regulated by a disjunctive ordering condition. The further assumption incorporated in the realizational models of Anderson (1992) and Stump (2001) is that ordering relations among rules partition a rule set into totally ordered subsets. In a rule set that conforms to this description, such as the Finnish rules in Figure 6.24, the rule order is mirrored by the block order.

What portmanteau patterns like the one in Table 6.12 show is that realization rules cannot always be partitioned into totally ordered blocks. Instead, in the partial ordering of realization rules, there may be rules that are unordered with respect to multiple sets of rules, even though the members of those sets are themselves ordered. In the Swahili pattern, exhibited in Figure 6.25, the rule introducing si-

Portmanteau rule ordering in Swahili

Figure 6.25 Portmanteau rule ordering in Swahili

is extrinsically unordered (hence disjunctively ordered) with respect to the rule introducing ha- and the simple agreement markers.

As Figure 6.25 indicates, a block order can be projected from these partially ordered rules. The order between blocks will just reflect the order between the rules in each block that are not also in the other. Block ‘V’ will precede block ‘IV’ because once the shared rule introducing si- is disregarded, the remaining rule in ‘V’ precedes all of the remaining rules in ‘IV’. Since both blocks precede the rules in block ‘ITh the same is true of any rules that span ‘V’ and ‘IV’. These blocks retain their usefulness for indexing rules; here the sole revision involves cross-indexing portmanteaux rules with multiple, overlapping blocks.

As this brief comparison shows, portmanteau rule blocks are an almost exact realizational counterpart of portmanteaux morphs. When the assumption that features and forms were in a biunique correspondence turned out to be false, the Post-Bloomfieldians did not initially revisit their starting assumptions but instead introduced a host of ‘special’ morphs. When the assumption that rules could be partitioned into totally ordered blocks encountered analogous problems, Stump (1993c, 2001) proposed portmanteau blocks, amongst other types of special blocks. The eventual rejection of special morphs was not due to the discovery of a pattern that could not be assigned a brute-force special-morph analysis. Rather, the Post- Bloomfieldians came to recognize that these extensions merely compensated for the inadequacy of the original conception of‘primary’ morphs. Special rule blocks can be seen to have essentially the same compensatory function in realizational models.

  • [1] Anticipating the ‘realizational pair’ format in Section 6.4.1, exponence rules are expressed as pairs(F, A), specifying the features F they realize and their output A.
 
Source
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >
 
Subjects
Accounting
Business & Finance
Communication
Computer Science
Economics
Education
Engineering
Environment
Geography
Health
History
Language & Literature
Law
Management
Marketing
Mathematics
Political science
Philosophy
Psychology
Religion
Sociology
Travel