Quirky subjects

In some languages having a morphological case-marking system, predicates which have sensory, or cognitive, or experiential features necessitate the appearance of subjects which are not nominative case marked. These subjects are called quirky subjects, or non-nominative subjects or oblique subjects.Footnote ⁷ Some of the Vafsi sensory/experiential/cognitive verbs are: /gow-an/ (to want), /xoš ^ owi-an/ (to feel good, to like), /xow ^ owi-an/ (to feel sleepy), /garm/sard ^ bi-an/ (to feel cold/warmth), which in all tenses have quirky subjects. These subjects, which have the experiencer function, will be overtly dative case marked, similar to Vafsi IOs which, based on the verb meaning, have one of the goal, recipient, or beneficiary functions.Footnote ⁸

However, Vafsi does not have relators specific to the dative case, so uses the oblique enclitics (see Table 3) as portmanteau morphemes used generically for all oblique cases. But, based on the genetic linguistics and areal typology, it can be deduced that the oblique case of the experiencer is actually a dative case.Footnote ⁹

According to Bickel, dative case constructions for experiencers is a feature specific to the South Asian and Indo-European languages. The following examples illustrate the Vafsi quirky subjects in context.Footnote ¹⁰

Example 1: quirky subject (experiencer)—/gow-an/ (to want)

In example 1 (a), the cognitive verb /gow-an/ (to want) has the first singular subject. But instead of using the subject pronoun /az/ (I) in the nominative case, the oblique pronoun (dative) /tamen/ (me) appears along with the first person singular pronominal morpheme /-m/ fronted to the direct object (DO); /siva/ (apple) is the DO which, according to the Vafsi differential object marking (DOM), will be direct case marked as it is inanimate and indefinite. Therefore, the sentence verb will show agreement with the DO by using the zero morpheme.

In example 1 (b), the cognitive verb /gow-an/ (to want) is in the present tense and its subject is a first person plural pronoun. Instead of using the nominative pronoun /āwān/ (we), the oblique (dative) pronoun /tāwān/ (us) is used along with the first person plural pronominal morpheme /-oān/ fronted to the DO, /xar/ (donkey). The DO is animate and by the appearance of the possessive pronoun /esda/ (your), as a pre-head modifier, will become definite, so according to the Vafsi DOM, the DO should be oblique case marked as /xar-i/; but because of the phonological phenomenon of hiatus (two vowels occurring in adjacent syllables, with no intervening consonant), the oblique morpheme /-i/, which in the sonority scale of ā>o>u>e>i is less sonorous than /o/, will be deleted.Footnote ¹¹ Hence, /xar-i-oān/ turns to /xar-oān/; consequently, in this sentence, both subject and object will be oblique case marked and as the Vafsi verb agrees with an argument which is direct case marked, the default agreement (third person singular) will appear by using a zero morpheme.Footnote ¹²

In examples 1 (c) and (d), the cognitive verb /gow-an/ (to want) is in the past tense, so these two examples show that the experiencer case pattern and the agreement system in clauses including cognitive verbs are not tense-based. /gow-an/is a polysemous verb. Therefore, if meaning “to like,” it will not be considered a cognitive verb in Vafsi; its grammatical subject will have the function of an agent rather than an experiencer.

Example 2: subject (agent) -/gow-an/ (to like) (C)

Example 2 (a) shows that /gow-an/ (to like) has the subject /az/ (I) in the direct case. /gol/ (flower) is an indefinite DO, so, according to the Vafsi DOM, it will be direct case marked, the same as the subject.

/gow-an/ (to like) is a transitive verb. So, in the past tense, it should follow the Vafsi ergative system. Therefore, the subject /tamen/ (me) will be ergative case marked and the indefinite DO /gol/ (flower) will be direct case marked (according to the Vafsi DOM).

In Hindi/Urdu, Pashto, and Balochi, as in Vafsi, quirky subjects will appear.Footnote ¹³ The following example from Butt shows the appearance of quirky subject in Hindi.Footnote ¹⁴

Example from Hindi: quirky subject in Hindi

Generation of the dative case marking (DCM) constraints for the experience. According to the universal case licensing principles introduced by Chomsky, every argument requires the appearance of a licensed case.Footnote ¹⁵ Therefore, in OT analyses, any input which violates these principles will be outcast before the application of other constraints.Footnote ¹⁶

But the universal case licensing principles only observe the case marking and do not specify the case type. So the specification of the case type will be determined by using the markedness constraints. Hence the constraints will be classified into two groups—faithfulness constraints and the markedness constraints—which will be extracted in the following sections.

Faithfulness constraints

Within the principles and parameters (P&P) model, the existence of the case marking (license giving) feature in every verb is part of the subcategorical features of that verb to which the related case will be assigned in the lexicon level in the D-structure. If a verb has the case licensing feature, this feature should be controlled for each input. So we need to introduce a constraint which will observe this feature at the initial stage. In optimality terms, it means that this constraint should be highest in rank and observe the consistency of the input and the output. The priority of the inherent cases in the Chomskyan generative grammar is justified by their assignment in D-structure, as the structural cases will be assigned in S-structure which is a higher stage relative to D-structure. Therefore, the inherent case assignment has priority over the assignment of the structural case.Footnote ¹⁷

Optimality equivalence of the priority of the inherent case assignment in D-structure is the faithfulness constraint in Model 1.

Model 1:

This faithfulness constraint is context-based—i.e. this faithfulness should be followed only if the specific context like the transitive verb or the perfective aspect occurs.Footnote ¹⁹ In Vafsi, this constraint will act in the context of transitive/intransitive verbs: in verbs which include the case licensing feature in their subcategorical features.

Markedness constraints

The faithfulness constraint introduced in Model 1 is only applicable in the control of the existence of the inherent case feature in output. Therefore, in order to specify the type of the case of the output, markedness constraints are used. According to Dixon and Woolford, the nominative case is the least marked and is a universal case which will appear in all world languages.Footnote ²⁰ According to Grimshaw and Woolford, inherent cases are more probably morphologically case marked relative to the structural cases; and hence are more marked.Footnote ²¹ According to the data, the markedness constraints hierarchy can be extracted as follows:Footnote ²²

Model 2:

The ranking of the faithfulness and markedness constraints mentioned in Models 1 and 2 respectively is shown in Table 4. According to the constraints hierarchy in Model 2, *ERG is ranked higher than *DAT and as the faith-lex is context-based, if the past transitive clause occurs, the ergative structure will appear and faith-lex (ergative) will be relevant. Otherwise, this constraint will be irrelevant.

Table 4. OT tableau of the case of Vafsi subjects

Note: The shaded cells have no effect on the outcome because the competition has been decided by the higher-ranking constraints.

Table 4 has six inputs from which inputs 1‒4 react to the constraints similarly in pairs, so they are classified and analyzed in these same reacting pairs. Inputs 1 and 2 both have cognitive verbs with the subcategorical features of +dative and -ergative. So the first constraint—i.e. faith-lex (ergative)—is irrelevant. Candidate “a,” because of ergative case marking, will be eliminated. Candidates “c” and “d,” because of failure in satisfaction of faith-lex (ergative context), will also be losers; therefore, candidate “b,” despite the violation of *DAT, will be the winner and the optimal candidate.

Example 3:

(a) subject-intransitive cognitive verb (-ergative, +dative)—(present)

Example 3 (a) is based on the input 1 structure. In this example, the compound verb /sard ^ bi-an/ (to feel cold) is a cognitive verb which requires the subject (experiencer) /tamen/ (me) in the dative case. But the compound verb /sard-ome/ (I am cold), by using the first person singular morpheme /-ome/ from set 1 (see Table 3) will show agreement with the sentence subject despite being in the dative case.

(b) subject-intransitive cognitive verb (-ergative, +dative)—(past)

Example 3 (b), like example 3 (a), has the same cognitive verb and the same subject in the dative case. But the sentence is in the past tense. These two examples show that the Vafsi cognitive/experiential/sensory verb will license the dative case for subjects in both past and present tenses.

Example 4: subject-transitive cognitive verb (-ergative, +dative)—(present)

In example 4, the cognitive verb /gow-an/ (to want) has a first person singular subject. But instead of using the nominative subject pronoun /az/ (I), the object pronoun /tamen/ (me) in dative case is used in coordination with the first person singular pronominal morpheme /-m/ fronted to the DO /siva/ (apple). The DO, according to the Vafsi DOM, because of being inanimate and indefinite, is direct case marked. This example is evidence of the prediction of the OT tableau in choosing the dative subject for the structure related to input 2.

Inputs 3 and 4 both have non-cognitive and non-ergative verbs. So the faith-lex constraint for both the ergative and dative contexts will be irrelevant. Therefore, the verb does not require any oblique case marking on the subject, so it will appear unmarked. Consequently, all three candidates “a,” “b,” and “d” will be eliminated because of fatal violation of the related constraints and “c,” the only candidate which is nominative case marked, will be the optimal candidate and the winner.

Example 5:

(a) subject-intransitive verb (-ergative, -dative)—(present)

In the above example, the intransitive verb /a-ch-óm/ (I go) is in the present tense so the nominative-accusative system will rule.Footnote ²³ Therefore, the nominative pronoun /az/ (I) will appear unmarked.

(b) subject-intransitive verb (-ergative, -dative)—(past)

In example 5 (b), the intransitive verb /ba-ss-im/ (I went) is in the past tense so the nominative-accusative system will rule. Therefore, the nominative pronoun /az/ (I) will appear unmarked.

Example 6: subject-transitive verb (-ergative, -dative)—(present)

In example 6, the transitive compound verb /bar ^ at-ār-ènde/ (they bring out) is in the present progressive tense, so the nominative-accusative system will rule. The subject is the deleted pronoun /āne/ (they) which is nominative (direct) case marked and its co-referential dependent pronoun is the third person plural pronoun /-ènde/ of set 1 (see Table 3), which is suffixed to the sentence verb as an agreement marker (AM).

Inputs 5 and 6 both have the ergative structures but input 6 has also a cognitive verb, so it will require the satisfaction of the constraint *DAT. Candidates “b,” “c,” and “d” are all eliminated because of violation of the highest ranked constraint, which is faith-lex (ergative context), and candidate “a,” despite violating the markedness constraint *ERG in input 5 and the double violation of the constraint *ERG and faith-lex (dative context) in input 6, will come out as the optimal candidate. The optimal output for both inputs 5 and 6 is candidate “a,” but the difference between these two inputs is that the existence of the dative context in input 6 will be more marked relative to that of candidate 5.

Example 7: subject-transitive verb (+ergative, -dative)—(past)

In example 7, the non-cognitive transitive verb /b-ard-a/ (brought) is in the past tense and licenses the ergative case for the subject, so the dependent third person plural subject pronoun /-san/ is of set 2 (see Table 3), which as the optimal candidate “a” predicts, will be ergative case marked.

Example 8: subject-transitive cognitive verb (+ergative, +dative)—(past)

In example 8, the cognitive transitive verb /gow-an/ (to want) is in the past tense so it licenses both ergative and dative cases for the subject. As Vafsi does not have special morphemes for ergative and dative cases, it cannot be judged morphologically whether the case of the subject pronoun /tawan/ (us), which is oblique case marked, is ergative or dative. For distinguishing this case, the OT tableau will help us to find out that the oblique case of the subject of the transitive verbs in Vafsi is actually the ergative case. The interesting point here is the power of prediction in OT tableaux, so we can deduce a universal rule for Vafsi according to the optimality theory. This rule would be as in Model 3:

This rule is shown in Table 4 by the optimal candidate of input 6 and is illustrated by example 8.

The OT analysis of the dative case of IO

The Vafsi IO, like that of Magar (an Indo-European language spoken mainly in Nepal, Southern Bhutan, Darjeeling, and India by the Magar people), which conveys one of the three functions of goal, or recipient or beneficiary, is always dative case marked. But Magar DO, always as a patient, follows a DOM pattern based on the animacy or definiteness features.Footnote ²⁶ Therefore, in case marking of the Magar DO, DDCM (differential dative case marking) will appear, which is sensitive to the animacy and definiteness features—that is, in this language, the human DO (definite or indefinite) will be oblique case marked and will receive the dative case morpheme /-ke/ and the non-human DO (animate or inanimate) will be dative case marked only if it is definite, otherwise it will receive the direct case marking.Footnote ²⁷ In Magar as in Vafsi, oblique case marking of the experiencer appears.Footnote ²⁸ As mentioned earlier, the faith-lex constraint is context-based, so it will require the emergence of the dative case for the IO in Vafsi double object construction (DOC).

As the focus of this section is on the OT analysis of the dative case and is not related to the analysis of the ergative case, the *ERG constraint in Model 2 will be redundant, resulting in its elimination in the constraint hierarchy in Model 4.

Consequently, the constraint rankings and the OT tableau for the dative case of Vafsi IO will be as in Table 5. Table 5 is relevant only to the inputs including the Vafsi DOC. The Vafsi DOC includes an IO which is always dative case marked.

Table 5. OT Tableau of the Dative Case of Vafsi IO

Note: The shaded cells have no effect on the outcome because the competition has been decided by the higher-ranking constraints.

Candidates “b” and “c” will be eliminated because of violating the highest ranked constraint faith-lex. As a result, candidate “a,” despite the non-fatal violation of the markedness constraint *DAT, will be selected as the optimal candidate and the winner.

Example 9: subject-DI-IO-ditransitive verb (DOC)

In example 9, the ditransitive verb /ad-do-ø/ (gives) is in the present tense and its DO is

/yadegāri/ (memorial), which according to the Vafsi DOM, is direct case marked because of being inanimate and indefinite and the IO /berā-y/ (brother) will be dative case marked by suffixation of the singular masculine oblique morpheme /-y/ (see Table 1).

The OT analysis of the dative case of object of adposition (OA)

As Vafsi does not have a specific morpheme representing the dative case, in case the oblique portmanteau morpheme (set 2) (see Table 3) appears in the animate objects of the preposition /da/ (to), or the preposition /az/ (to), or the postposition /rā/ (for), or the enclitic /-o/ (to), the OT analysis of the dative case of the OA will be the same as for the Vafsi DACM (differential adpositional case marking), which is analyzed in another paper by the authors titled: “DACM in Vafsi within OT framework.”Footnote ³⁰