Studies in Semitic Syntax

CONTENT

INTRODUCTION

I. TRANSFORMATIONAL GRAMMAR. A BRIEF OVERVIEW
1. Looking At Other Linguists´ Theories

II. A STUDY OF SEMITIC SYNTAX
1. ANALYZING FASSI-FEHRI`S THEORY
1.1. How Are Words Formed In Arabic?
1.1.1. Summary
1.2. Word Order In Arabic
1.2.1. S-V-O versus V-S-O
1.2.2. Case Assignment
1.2.3. The EPP
1.3. Summary
2. SHLONSKY`S APPROACH
2.1. The Order Of Inflectional Elements In
Hebrew And Arabic
2.1.1. Imperfectivity In Hebrew And Arabic
2.2. Word Order In Hebrew
2.2.1. Negation And Word Order In Arabic
-The ‘ laa ’ Negation
-The ‘ laysa ’ Negation
-The ‘ maa ’ Negation
2.3. Summary

III. CONCLUSION

IV. BIBLIOGRAPHY 83

INTRODUCTION

This work’s aim is to first show how linguistic theories, in this case the syntactic theory of Transformational Grammar, which have been developed on the basis of Indo-Germanic languages, especially on English and French, apply to other language families, in this case to the Semitic languages Arabic and Hebrew. Moreover it is Arabic which I wanted to examine more thoroughly, since this language has not been analyzed very much linguistically.

This work is based on the principles of Transformational Grammar. The main question that covers this work is along which parameters Arabic varies compared to English for example. However, I will not be able to answer this question in detail since the examination of this subject would exceed my work, but I will analyze certain fields of this subject.

Finally, this work can also lead to a better understanding of the Arabic language and thinking if we consider Sapir and Whorf´s theory which associates language with thinking.

I.TRANSFORMATIONAL GRAMMAR. A BRIEF OVERVIEW.

What is Transformational Grammar? Which part of a natural language does it constitute? And what are its rules like?

These questions, among other things, will make up this introductory part in order to state the means by which the main part of this work will be analyzed.

The name ‘Transformational Grammar’ contains the word ‘transform’ which leads to the conclusion that Transformational Grammar is the description of a grammar in which transformations occur. But what kind of transformations are they? And what will be transformed?

To start with I will first present the rules that constitute this theory.

One of the most important rules is the X-bar (X’) theory which, along with the Projection Principle which says that lexical information is syntactically represented, states that all syntactic structure is built on the basis of a binary and layered representation, the X’ format.

illustration not visible in this excerpt

The syntactic structure of an English declarative sentence would then look like the following:

illustration not visible in this excerpt

Within this structure we find predicates and arguments which are designed to be such by theta (q) theory. This theory states the semantic relationship between predicates and their arguments following the Theta Criterion which says that each argument is assigned one theta role and each theta role is assigned to one argument. Following this we can now say that the verb in the above English sentence is the predicate assigning a theta role to the subject and another one to the object.

The Extended Projection Principle furthermore maintains that every sentence must have a subject. Thus there can be no sentence containing just the verb and the object as we can see in the following English example:

illustration not visible in this excerpt

Having now formulated these rules we now come to the question whether there is a general ordering rule for the subject, the verb and the object that is valid for all natural languages, and whether the S-V-O order in English sentences, which we have stated above, is an order that applies to all types of English sentences or not.

Transformational Grammar distinguishes between four levels of syntactic representation. First of all there is the Deep Structure (DS) representing the basic argument relations in a sentence followed by the Surface Structure (SS) which reflects the actual ordering of the elements having undergone move alpha, a process of various movement transformations consisting of Head Movement, A-Movement (or NP-Movement) and A’-Movement (or Wh-Movement)^[1]. At Logical Form (LF) the acquirement of sentential scope triggers movement which has no overt reflex. Finally the Phonetic Form (PF) is the phonologically overt realization, the spell out of a sentence.

Now we can reformulate the question from above - what is the Deep Structure of an English sentence and what is its Surface Structure?

Various linguists agree in classifying English as being an S-V-O language. This order still hides movement, namely vacuous movement, which of course cannot be seen in the linear ordering of the constituents, but which takes place when the level of representation changes. Therefore the Deep Structure representation of the English sentence (1a) is not the same as its Surface Structure representation.

illustration not visible in this excerpt

We can see that at Surface Structure the subject has moved leaving a co-indexed trace in its base position inside VP. In the linear ordering, however, nothing has changed.

But what kind of movement is this?

As mentioned above the kind of movements we are concerned with in this work are Head Movement, A-Movement and A’-Movement. Before explaining each single movement transformation it is important to add the Structure Preserving Principle to our list of syntactic rules, which states that all structures established at Deep Structure must be preserved at Surface Structure.

Head Movement is a movement transformation involving movement of heads (X°) to head positions leaving traces in the base-positions. This kind of movement is triggered by the richness of agreement morphology of a verb. English and French for example differ with respect to this possibility. Starting with the Split-INFL hypothesis which states that the structural node IP (Inflectional Phrase) is divided into Agreement Phrase (AgrP) and into Tense Phrase (TP), linguists have found out that in English only auxiliaries can undergo movement from V° to T° and then to Agr°. According to the Head Movement Constraint (HMC) saying that Head Movement has to take place in a strict cyclic fashion. In French, for example, both auxiliaries and lexical verbs can and obligatorily must move due to their morphological richness as we can see in the tree diagrams of the Surface Structure sentences below.

illustration not visible in this excerpt

English lexical verbs stay in their base position V° with the inflectional endings Agr and T lowered onto them, while in French both auxiliaries and lexical verbs raise to Agr° passing T°, in negative sentences also passing Neg°, to attach to their rich inflectional endings.

These differences in the behavior of English and French verbs can also be explained in view of the Minimalist Program which states that verbs are base-generated together with their inflectional endings. Agr° and T°, on the other hand, dominate bundles of abstract features which have to check the morphology associated with the verb. This feature-checking is done by adjoining the inflected verb to the relevant functional head. The abstract features will then be eliminated. This feature-checking has to take place before spell-out (PF) if the morphology on the verb is strong. If the morphological realization on the verb is weak, however, the verb does not have to move to check inflectional features.

A-movement (or NP-movement) and A’-movement (or Wh-movement) are similar to each other in that they obey similar conditions by moving phrases to A-positions in the first case and to A’-positions in the second, in a strict cyclic fashion also leaving traces in the base-positions. It is the reason for movement and the behavior of landing site and trace that are different.

A-movement is an obligatory movement resulting from the need of a noun phrase (NP) to be case-marked. In its base-position an NP succeeds in being assigned a theta role, but no case, which violates the Case Filter which says that every overt NP must be assigned abstract case. Abstract case renders an NP visible and therefore allows it to be theta-marked according to the Visibility Requirement. The NP has to move out of its base-position into a position in which it can receive case. This position is an A-position or a theta-bar position, a position to which case, but no theta role can be assigned. The trace is theta-marked and by virtue of chain-building the NP receives the theta role from its trace. A-movement takes place in raising structures in which raising verbs or adjectives occur or in passive constructions. An example of passivization will make this clear.

illustration not visible in this excerpt

While A-movement moves an NP to an A-position A’-movement moves a Wh-element to an A’-position, to [Spec,CP], a specifier position. Wh-movement takes place whenever a question is formed. The base position of the moved element, the Wh-trace, is both case- and theta-marked.

illustration not visible in this excerpt

Coming back to the Extended Projection Principle we take a look at the following examples.

5a) To have eaten the cake is good.

5b) PRO to have eaten the cake is good.

6a) Parlo.^[2]

6b) pro i parloi.

speak-1sg

‘I speak.’

In both cases there seems to be no subject visible that might carry theta role or case. PRO and pro, however, are the non-overt subjects that are generated in order to render the sentence grammatical.

PRO is assigned a subject theta role by the verb without being case-marked respecting the PRO Theorem which claims that PRO must be ungoverned.

pro on the other hand is a non-overt pronoun which is co-indexed with and also theta-marked by the verb which carries the matching agreement morphology. pro, however, is not a language universal element. We can see that it does not occur in English, but does so in Italian in example (6).

7a) * Speak.

7b) I speak.

English, in contrast to Italian, is not a pro-drop language, it is not a language in which a pronominal subject is allowed to be left unexpressed. The subject pronoun cannot be dropped. But why does this work in Italian?

It is again the richness of inflectional morphology associated with the verb that determines whether a language is pro-drop or not. Modern Hebrew for example allows pro-drop only in cases in which inflectional morphology is rich. Therefore pro is not allowed in morphologically poor present tense and is restricted to first and second persons in main clauses with future and past tense.

8a) pro i ‘Axaltii ‘et ha-tapu’ax.^[3]

ate-1sg ACC the–apple

I ate the apple.

8b)* pro i Oxeli ‘et ha-tapu’ax.

eat-sg ACC the-apple

8c) ‘Ani oxel ‘et ha-tapu’ax.

I eat-sg ACC the-apple.

Therefore not all types of inflection are rich enough to allow pro-drop.

1. Looking At Other Linguists´ Analyses

Having now taken a look at the summary of the general theory of Transformational Grammar we can now turn to the theories of some other linguists namely to that of Ian Roberts and Hilda Koopman.^[4]

In his work “Comparative Syntax” Roberts is pleading for the VP-Internal Subject Hypothesis claiming that subjects are base-generated in [Spec,VP] and then raised to [Spec,(AgrSP)IP]^[5] having been assigned a theta role under minimal m-command by V. He further shows that the I° node has the properties of a raising predicate since the subject cannot be case-marked by I° in [Spec,VP]. It will have to raise to [Spec,IP] where it can receive case. This, however, only has to take place in S-V-O languages like French, but not in V-S-O languages like Welsh or Arabic in which both versions are allowed.

In V-S-O languages the subject can be case-marked inside VP due to the fact that I° has strong V-features, meaning that it has more morphological realizations than I° in S-V-O languages, and moreover weak N-features, nominal features. The verb moves out of VP, but the subject does not as we can see in the Welsh example.

illustration not visible in this excerpt

Furthermore Roberts claims that “when a subject follows the verb in a VSO language, it cannot agree in all morphological features with the verb^[7] ” which he calls the Anti-Agreement Effect. He illustrates this with an example from Classical Arabic where the verb shows full agreement (person, number and gender) when it follows the subject, whereas it only shows gender agreement when it precedes the subject.

^[8]

illustration not visible in this excerpt

This fact I want to support with another example in which I also found out that it is not only gender that is being left on the verb in V-S-O order, but also person agreement.

illustration not visible in this excerpt

Roberts claims that “the parameter which distinguishes Welsh [and Arabic] from French would be one which determines whether or not the subject is able to be assigned Case in SpecVP”^[9].

The question of why and how languages vary with respect to this parameter I will analyze together with the help of other linguists later in the main part of this work.

Hilda Koopman focused her work on searching for the canonical position of subjects. She also found out that languages such as English and French generate their subjects inside VP which they then move out of VP in order to case-mark them while in languages such as Italian, Welsh and Arabic the subject which is base-generated in [Spec,VP] does not have to raise to

[Spec,IP]. A V-S-O language cannot be base-generated as V-S-O, she states, but this structure involves movement of the verb to I°. Koopman also observes that both S-V-O and V-S-O orders are allowed in Arabic simple clauses with the V-S-O order only exhibiting a default number agreement while in the S-V-O order the verb fully agrees with the subject. Therefore in Arabic tensed clauses raising is not obligatory. The subject in the V-S-O order which stays inside VP must then receive NOM case from I° which governs the subject. Koopman argues that NOM is a both structural case assigned under government as well as case by agreement. She therefore distinguishes between case assignment under government and case assignment under agreement. In French and English, both S-V-O languages, I° can only assign NOM case to the subject by agreement which explains why the subject has to move out of the VP, while in Arabic, an optional V-S-O language, I° can govern into the VP and thus assign case to its specifier. Arabic then seems to be a language which tolerates both types of case assignment, either by agreement, representing the Spec-head agreement between subject and verb, or by government into the VP which has the effect that the subject fails to raise to [Spec,IP] which on the other hand correlates with the lack of number agreement on I°. Koopman continues that since I° in Arabic can assign case either by agreement or by government the [Spec,IP] position is an A’-position in the first case and an A-position in the second.

II. A STUDY OF SEMITIC SYNTAX

1. ANALYZING FASSI-FEHRI`S THEORY

1.1. How are words formed in Arabic?

In his work Issues In The Structure Of Arabic Clauses And Words Abdelkader Fassi-Fehri is pleading for a syntactic approach to Arabic word formation. He rejects word formation as taking place on a pure lexical level in favor of the hypothesis that Arabic words can be bracketed internally mirroring syntactic clause structure rules. He claims that in Arabic there is no such thing as a stem or a root which is labeled for the category of a noun, a verb or an adjective. No inflectional or derivational morphemes can be added to form a conjugated verb as is the case with English, but there are consonantal roots which inherit category labels only when they enter into syntax. Affixes, on the other hand, are categorially specified mirroring the syntactic order that occurs within phrases and clauses within words. He argues that a verbal ‘stem’ like kataba ‘write+Past+3m.sg.’ is already a conjugated verb inflected for Tense and Active Voice. In English, however, this word syntactically forms a sentence.^[10]

illustration not visible in this excerpt

In example (2) there is the consonantal ‘stem’ [?llm] onto which the inflectional affixes are attached forming a morphological unit obeying syntactic rules. These affixes, however, do not occur in an accidental order, but syntactic structure rules to apply to them as is shown by the next example:

3) * hu - na - a?llim - sa - uu - yu

In syntax each element must occur under an X° projection the order of which would constitute the general order of constituents in clauses. How would a tree diagram of this configurational structure look like?

illustration not visible in this excerpt

In the proposed structure there is no solution found for the person-gender, number and mode affixes. If we now change the linear order of AgrP and TP according to Pollock (1989) such that TP dominates AgrP, there might be the possibility to say that the person-gender and number affixes are generated under Agr° and the verbal stem moves there to join them. Furthermore we can create the Mode Phrase (MP) which would follow AgrP and where we could attach the mode affix to. The resulting structure would then be the following:

illustration not visible in this excerpt

This analysis, however, is only quickly proposed. It is not examined in detail and lacking the consideration of other phrase structures and sentence conditions.

Fassi-Fehri continues by posing the question of what the Arabic stem or root looks like. He compares the English stem work to the Arabic consonantal root ‘ktb’ observing that in none of the examples below is there an Arabic stem which could be inserted under V°.

[...]

---

^[1] The explanation of these notions I will give later in the text.

^[2] Example taken from Haegeman, Liliane. Introduction to Government and Binding Theory. Blackwell Publishers Ltd., 1994, p.455

^[3] Examples taken from Haegeman Liliane. Introduction to Government and Binding Theory. Blackwell Publishers Ltd., 1994, p.456

^[4] taken from Koopman, Hilda. The Position Of Subjects. 1988, and Roberts, Ian. Comparative Syntax. 1997

^[5] I will use IP instead of Roberts´ AgrSP for simplicity´s sake.

^[6] Example taken from Roberts, Ian. Comparative Syntax. 1997, p.79

^[7] quoted from Roberts, Ian. Comparative Syntax. 1997, p.109

^[8] taken from ibid., p.109

^[9] quoted from Roberts, Ian. Comparative Syntax. 1997, p.79

^[10] The content of the following section summarizes and analyzes Fassi_Fehri, Abdelkader. Issues In The Structure Of Arabic Clauses And Words. Kluwer Academic Publishers, 1993