مشخصه های مقوله گزینی در ساخت پوچ واژه: رویکردی کمینه گرا (مقاله علمی وزارت علوم)
درجه علمی: نشریه علمی (وزارت علوم)
آرشیو
چکیده
پوچ واژه ها یا اصطلاحاً فاعل های بدلی، عناصری نحوی اند که سهمی در شکل گیری معنای گزاره ندارند و به همین دلیل، نزد بسیاری از زبان پژوهان، از مهم ترین شواهدی به شمار می روند که تصریح می کنند تولید گزاره برپایه یک فاعل و محمول از الزامات نحوی، و نه معنایی، جمله است. هرچند در زبان های ضمیرانداز که تصریف غنی دارند، فاعل های ارجاعی می توانند آزادانه آشکار یا تهی باشند، پوچ واژه ها در چنین زبان هایی غالباً ناملفوظ اند. در نحو زبان فارسی، وجود قاعده درج پوچ واژه موضوع کم وبیش مناقشه برانگیزی در میان زبان شناسان بوده است. برخی از تحلیلگران استدلال کرده اند که این ضمیر غیرارجاعی در نحو زبان فارسی یافت می شود، اما برخی دیگر مدعی شده اند که فارسی فاقد هر دو نوع پوچ واژه آشکار و پنهان است. در پژوهش حاضر در چارچوب برنامه کمینه گرا و براساس مشخصه های مقوله گزینی محمول ها، استدلال می کنیم که شاخص گروه زمان در فارسی باید با یک اسم یا ضمیر پر شود و لاجرم، در آن دسته از محمول های نامفعولی که یک گروه اسمی حامل نقش تتا وجود ندارد، درج پوچ واژه ضمیری الزامی است. اما مهم تر از این نکته، می کوشیم تا به جای ارائه فهرستی تفصیلی از ساخت های حاوی فاعل های بدلی، تعمیمی برپایه آموزه های نظری و شواهد تجربی به دست دهیم تا اعمال قاعده درج پوچ واژه را در ساخت های نحوی فارسی پیش بینی کند. طبق این تعمیم، جایگاه فاعل ساختاری محمول های نامفعولی با پوچ واژه ضمیری پر می شود، اگر و فقط اگر یک گروه اسمی با حالت بازبینی نشده در اشتقاق جمله در دسترس نباشد. در چنین ساختاری اگر پوچ واژه فاعلی ناملفوظ درج نشود، اشتقاق نحوی به دلیل تخطی از اصل فرافکن گسترده فرومی ریزد. این تحلیل که برپایه داده های آن، پوچ واژه ضمیری آشکار در فارسی یافت نشده است، توضیح می دهد که چرا در حالی که اسم های ارجاعی و پوچ واژه ضمیری در توزیع تکمیلی اند، در برخی از اشتقاق های نامفعولی و از جمله در ساخت های ارتقایی، یک گروه اسمی آشکار ظاهراً می تواند در جایگاهی بنشیند که متعلق به پوچ واژه ضمیری پنهان است.Categorial-selectional features in expletive constructions: A minimalist approach
Expletives, or the so-called dummy subjects, are syntactic objects that contribute nothing to the semantics. For many linguists, these elements constitute the strongest argument of the requirement for propositions to be built from a subject and a predicate, that the subject necessitated by the syntactic motivation, not by semantic one. While referential subjects may optionally be overt or null in pro-drop languages, it is well-known that they generally lack overt expletives. Whether expletive insertion rule applies to some derivations has been a rather controversial issue in the Persian syntax. Some linguists argue that non-referential pronoun exists in Persian, however, some others claim that there are no overt or covert expletives in this language. Based on the categorial-selectional features of predicates and within the framework of minimalism, I argue that in Persian the Spec of TP is obligatorily filled by a noun or pronoun, so expletive constructions are found with some unaccusative predicates, where there is no theta marked Noun Phrase. More importantly, and at a more theoretical and empirical level, I attempt to provide a kind of generalization about the application of expletive insertion rule within certain constructions. According to this generalization, the subject position of an unaccusative predicate may be filled by an expletive pronoun if and only if a NP with an unchecked case feature is not accessible. In this case, if there is no null expletive subject, then the EPP will filter the sentence out. While the referential nouns and pleonastic pronouns are in complementary distribution, such an account which requires that a predicate be able to select different kinds of categorial features explains why in some unaccusative derivations, including raising constructions, an overt NP can apparently be inserted in the position occupied by a covert expletive pronoun.
Keywords: expletive construction, categorial-selectional feature, unaccusative predicate, structural subject, EPP
Introduction
An expletive refers to the element that may fill the surface subject position, but does not receive θ-role from the predicate. These pleonastic subjects constitute the strongest argument that there is a need to invoke the notion of predication independent of θ-role assignment, and more specifically of the requirement that propositions must be built from a subject and a predicate with the syntactic rather than semantic motivation, since it is assumed that expletives contribute nothing to the semantics of such a sentence. Previous works on pro-drop languages identified a set of parametric values associated with the availability of null subjects, including the presence of null expletive subjects. For example, Rizzi (1982) has argued that there are correlations between thematic null subjects and null expletives which means that pro-drop languages usually have no over expletive.
Persian is a pro-drop language with canonical SOV word order. It is well-known that richly agreeing Null-subject languages lack overt expletives, though Finnish is an exception to this general rule. It has been recently controversial whether there is an expletive construction in Persian. While some scholars argue that Persian has no overt expletives, and there is no evidence pointing to the existence of covert expletives, others argue that overt and covert expletives exist in this language (cf. Darzi, 1996; Karimi, 2005; Mansoori, 2015).
The aim of this article is to support the view that expletives are generated in this language. In the same vein, but more within the framework of MP, we will discuss the fact that referential subjects may be optionally overt or covert in Persian, while expletives are obligatorily null. In this regard, a covert pronoun expletive will be claimed to have merged in the surface subject position of all zero-place predicates and some unaccusative structures where the internal argument is a PP or CP, satisfying the EPP feature - a requirement forcing some DP to appear in the specifier of the clausal head.
This analysis is empirically supported by a number of facts arguing against the idea that pronoun “in” (this) can freely appear as an expletive subject in Persian. Moreover, we try to answer an important problem: while referential nouns and pleonastic pronouns are in complementary distribution, why in some unaccusative derivations, including raising constructions, an overt NP can apparently be inserted in the position occupied by a covert expletive pronoun. Finally, we provide a generalization which predicts the insertion of null expletive in syntactic structures.
Materials and Methods
Argument structure – the pattern of underlying relations between a predicate and its dependents – is at the base of syntactic theory, and the theory of the interface with semantics. In studying argument structure, therefore, the relation between predicates and their arguments, especially the category of theses dependent elements is studied. The standard way to express statements about the category of the arguments to which a particular θ-role is assigned by predicates, is by means of categorial selectional features (abbreviated to c-selectional features) also called subcategorization features. Within the argument structure of a predicate, a c-selectional feature is a categorial feature on a lexical item, which does not determine the distribution of the lexical item itself; rather it determines the category of the elements which will be able to merge with that lexical item.
C-selectional features or subcategorization features must be checked at the interfaces for appropriate interpretability. It should be emphasized that all formal features which relate sound and meaning come in binary divisions; they can be interpretable or uninterpretable. The property of interpretability, generally used as the driving force behind the establishment of syntactic dependency in the minimalist system, is supposed to play a central role in the syntactic computation to drive the transition into the interpretive LF component. The idea is that uninterpretability forces feature matching and as a result, any uninterpretable c-selectional feature which has been matched, will be deleted. To put it more concretely, unlike interpretable features which have an effect on semantic interpretation and can participate in more than one checking operation, the uninterpretable ones must be eliminated before they reach LF; otherwise, Full Interpretation will be violated.
Another important theoretical issue related to the present research is argument structure alternation. According to this concept, some predicates have more than one set of c-selectional features and they may choose different kind of categories as their internal arguments. For example, the verb “pazir” (accept) can select a NP, CP or even an AgrP (small clause) as its complement. This phenomena will explain one of our main questions and show why in some unaccusative derivations, an overt NP can apparently be inserted in the position occupied by a covert expletive pronoun.
Discussion of Results and Conclusions
In line with what holds for feature checking, we examine the properties of expletives within the Minimalist framework. As we mentioned before, an expletive is an element that does not receive theta role from the predicate. So how can we account for the fact that the some unaccusative predicates accept such an element in their subject position? The answer is that expletives are place holders for the subject and pronoun expletive is fully specified with categorial, Case, and phi-features. Therefore, these dummy subjects are inserted in some unaccusative structure for syntactic purposes.
As we mentioned before, Persian is a Null-subject, verb final language that exhibits an SOV order in the unmarked order, except clausal arguments that occur postverbally. In this research, we discuss that in this null subject language, the Spec of TP is obligatorily filled either overtly or covertly by a noun or pronoun, so expletive constructions are found with some unaccusative predicates, where there is no theta marked Noun Phrase. In this regard, a covert pronoun expletive will be claimed to have merged in the surface subject position of all zero-place predicates (1a) and some unaccusative structures where the internal argument is a PP or CP (1b & c), satisfying the EPP feature - a requirement forcing some DP to appear in the specifier of the clausal head:
1.
(a).
tufɑn
šod.
storm
became-3Sg
“The storm began.”
(b).
az
ɑnhɑ
hemɑyat
šod.
from
them
support
became-3Sg
“They were supported.”
(c).
be
mɑ
xabar
dɑde
šod
ke
ɑnhɑ
resid-and.
to
us
information
given
became-3Sg.
that
they
arrived-3Pl
“We were informed that they had arrived.”
In all above sentences, the subject position of matrix clauses is filled by null pronoun expletive. In (1a), “tufɑn” (storm) is the non-verbal element (NV) of a zero-place predicate and “šod” (became) is its light verb (LV). Since there is no internal or external argument, an expletive pronoun must be merged in the surface subject position to satisfy structural requirements of the derivation, including EPP and case feature. In (1b) and (1c), the internal argument is a preposition phrase and a complement phrase, respectively. On the one hand these constituents are not subject to the Case Filter (i.e. they don’t need case checking), and on the other hand they cannot check the EPP feature of the clausal head due to their phrasal category. So, a subject expletive which agrees with the LV in phi-features has to be inserted in the Spec, TP.
This analysis immediately brings up an important question: if the subject position of (1c) is non-thematic and receives no θ-role, how can it be filled by an overt NP like “in nokte” (this point)?
2.
in
nokte
be
mɑ
xabar
dɑde
šod
ke
ɑnhɑ
resid-and.
this
point
to
us
information
given
became-3Sg.
that
they
arrived-3Pl
“We were informed that they had arrived.”
Based on the argument structure alternation, we will argue that contrary to (1c), the internal argument in (2) is a noun phrase, modified by an extraposed relative clause. Putting the facts together, we have to conclude that the predicate of this sentence has a different c-selectional feature. It should be noted that theories of argument structure ignore long-distance displacements such as the extraposition of a relative clause.
This conclusion is also true for the so-called raising predicates, where a verb like “be-nazar residan” (to seem) selects either a complement phrase (3) or a small clause (4) as its internal argument:
3.
badihi
be
nazar
mi-resad
ke
do
kešvar
be
tavɑfoq
dast
xɑhand
yɑft.
obvious
to
view
Prog-arrive.3Sg
that
two
country
to
agreement
hand
will.3Pl
find-Infinitive
“It seems obvious that the two countries will reach an agreement.”
4.
in
qaziye
badihi
be
nazar
mi-resad.
ke
do
kešvar
be
tavɑfoq
dast
xɑhand
yɑft.
this
case
obvious
to
view
Prog-arrive.3Sg
that
two
country
to
agreement
hand
will.3Pl
find-Infinitive
“It seems obvious that the two countries will reach an agreement.”
In sentence (3) where “badihi” (obvious) is an optional speaker-oriented adverb, the internal argument of the main verb is a CP. Therefore, the subject position of the matrix clause is filled by a covert expletive. In sentence (4), the c-selectional feature of the main verb has been satisfied by a small clause, i.e. “in qaziye badihi” (this case obvious). So, “badihi” is the predicate of the small clause and cannot be deleted. The NP “in qaziye” (this case) which is modified by an extraposed relative clause raises to the highest specifier of the main clause in order to check its case feature.
The conclusion we draw from these data is that the subject position of an unaccusative predicate is occupied by a null expletive, iff a NP with an unchecked case feature is not accessible.
تبلیغات
