An Investigation into Lexical Aspect of Persian Light Verbs: zadan, dǎdan and xordan Based on Huddleston and Pullum (2002)

The verb and its complement have their specific contributions to the meaning of predicate. Sometimes, the verb takes a more modest share than the complement in expressing the predicate; such verbs are referred to as light verbs. However, the degree of the lightness of the verbs is neither systematic nor predictable in verb constructions. We aim to investigate the lexical aspects of the Persian light verbs “zadan” (to hit), “dǎdan” (to give), and “xordan” (to eat) based on Huddleston and Pullum’s Model (2002). There are many verbs that lend thems elves to light verb constructions. The data are collected from the Updating Persian Corpus consisting of 48M words. To this end, we extracted all light verbs in the sentences stored on the Updating Persian Corpus using the AntConc software based on Mansour i’s approach (2013). The sampling was done using random sampling method that has scientific validity and is generalizable. In random sampling, every entity of the population enjoys equal chance of inclusion in the sample. Then we analyzed the sentences containing light verbs in terms of the state and activity lexical aspects. The results were plotted in tables and figures with annotations of the verb features. For similar patterns, only one pattern was selected for further analysis. The light verb patterns were listed and analyzed in an inventory one by one. The study adopted a descriptive-analytical method and a corpus-based survey design. Descriptive research has both applied and basic aspects so that it leads to the discovery of facts and bodies of knowledge in an inductive way. An important characteristic of descriptive studies is that the researcher may not manipulate or control the status and role of variables. The researcher merely tends to study and describe what already exists. In this study, we used library research and content analysis methods.


Introduction 1
Light verbs have always been a topic of interest for literati, syntacticians and linguists because of their more applicability, productivity and effectiveness compared with simple verbs in Persian. In most syntactic compositions, the verb is the most important element as it affects other sentence elements. Unlike many other languages, Persian language has more compound verbs than simple ones. According to capacity theory, a verb is the structural center of gravity of a sentence as it determines the number of complements that may or should appear in a sentence. That is, the verb is often the core of the sentence while other elements are either verb complements or complements of verb complements (Tabibzadeh, 2006: 25). All compound verbs constitute a base and a verbal element. In compound verbs, the base is a grammatical role that manifests as a noun phrase (NP), adjective phrase (AP) or prepositional phrase (PP). Verbal component, however, is almost bleached of meaning. These verbs are commonly referred to as light verbs (LV). Simple verbs that constitute the verbal element of a compound verb are verbs such as "kardan" (do), "šodan" (become), "dǎdan" (give), "zadan" (beat), "gereftan" (get) and "xordan" (eat). Sometimes, light verbs are interchangeable. For instance, "faryăd zadan" and "faryăd kardan" express the same meaning, i.e. to shout (Lazard, 2014). The base and verbal element are two distinct syntactic constructions that make up together to create the LV construction. There has been extensive research on compound and light verbs in Persian. However, a glance at previous studies proves the need for new research on the semantics and patterns of LVs. We hope to shed light on the lexical aspects of some Persian LVs.
The results of the present study may be used by linguists, syntacticians, and natural language processing (NLP) specialists to create Persian language tools and resources, translation machines, and syntactic and syllabic fragmentation software. The results may also prove useful in teaching Persian to non-Persian language learners.

Research objective
A glance at the syntax literature reveals that different researchers have addressed LVs from different aspects. However, some dimensions of Persian LVs have remained untouched especially in terms of lexical aspects. LV lexical aspects in Persian have not yet been addressed in research coherently. When studied in the light of state and activity concepts, LVs will reveal novel dimensions of their roles that are currently unnoticed.
Our cursory study shows that LVs "zadan", "xordan", and "dǎdan" are different specific features so that they may be used in different constructions. This study aims at an investigation into lexical aspects of constructions carrying three Persian light verbs: "zadan", "xordan", and "dǎdan".

Significance of the study
As many scholars including Huddleston and Pullum (2002:290) contend, LVs have a smaller share than the verb complements in producing the predicate meaning. In example A below, the meaning of the predicate is completely conveyed by the lexical verb "did" (saw). In example B, however, the predicate meaning is conveyed through the NP "negăhi" (a look) while the verb "andăxt" (threw) has a smaller semantic share.
The instructor had a look at my homework.
Accordingly, the verbal construction in B is considered a light verb. The major goal of the present study is to such verbs in terms of their aspect. Since, tense and aspect are inter-related, we will also address aspect along with tense.

Research method
The research adopts a descriptive-analytical method and a survey design. Descriptive studies assume both applied and conceptual roles so that they can lead to the discovery of scientific facts and add to the body of knowledge in an inductive way.
A key feature of such studies is that the researchers make no intervention in the status and role of variables. That is, they do not manipulate or control the variables but merely describe what already exists. In the present study, we adopted a library research approach to examine the texts and contents. We drew on the data stored in the Updating Persian Corpus to extract the sentences containing the verbs "zadan", "dǎdan", and "xordan" as LVs. The sampling was done using random sampling method that has scientific validity and is generalizable. In the random sampling method, all members of the population have an equal chance of inclusion in the sample. Subsequently, the sentences containing LVs were analyzed in terms of the state, activity, achievement, durative, and punctual lexical aspects. The results were illustrated for each LV in tables and figures containing values and annotations. The data were analyzed both quantitatively and qualitatively. We extracted and analyzed the sentences containing LVs. Overall, the study was conducted in six stages to address the lexical aspects of Persian LV constructions containing "zadan", "dǎdan", and "xordan": 1. Extraction of the applicable samples of "zadan", "dǎdan", and "xordan" from the Updating Persian Corpus 2. Classification of the samples 3. The structural patterns of each LV were analyzed in terms of lexical aspects. 4. All verbs were studied in terms of their lexical aspects. 5. The data were entered into Excel software. 6. The data were statistically analyzed in Excel software.
The research instrument included the Updating Persian Corpus that is freely available online at http://www.peikare.ir. The sample size was calculated using the Krejcie and Morgan's table (1970). Following the data collection stage, we acquired a large bulk of data to use in the analysis process. Thus, we extracted and tabulated the data manually. 5. How many LV constructions with "zadan", "dǎdan", and "xordan" have simple counterparts?
6. How many LVs are expandable?

Research Hypotheses
1. All LV constructions containing the verb "xordan" are intransitive.
2. All state verbs are transitive, and their complements are necessarily direct objects.
4. LV constructions with "xordan" have more simple verb counterparts than other LV constructions.
5. Persian verbs often appear as intransitive verbs in the sentence. In other words, intransitive verbs are more frequently used than intransitive ones.
6. Activity verbs are used more frequently than state verbs.

Conceptual foundations of the research 2.1 Theoretical framework
The first and foremost objective of the research is to describe the principles of the lexical aspects of Persian LVs constructed with "zadan", "dǎdan", and "xordan" based on Huddleston and Pullum (2002). However, we do not aim to showcase or defend a theory of syntax. Since human languages are all complex, it is not viable to describe them without adhering to a theoretical framework. Any attempt at describing the Persian language without adhering to a theory of syntax poses a major problem because an effective description of the Persian language is only viable though generalization. In turn, generalization is only workable through following a theoretical framework.
Nowadays, it is necessary to follow a certain theoretical framework in scientific research; otherwise, the researcher may deviate from the standard path of research. Thus, theoretical framework is a signpost in the risky path of research. A study of the Cambridge Grammar of the English Language by Geoffrey K. Pullum and Rodney Huddleston (2002) fueled the motivation for this research. Although the book is written on English grammar, it has a certain research tendency both in general and in particular. That is, as the authors contend, the discussions of the book on syntax do not solely apply to the English language but address the universal aspects of syntax as well. Therefore, the book has a capacity to function as a research model. In the following section, we will provide a brief discussion on the theoretical aspects of the book as the conceptual framework of this research. In the discussion, we will review the distinction between universal and specific definitions of language as well as our theoretical perspective towards LV.
So far, linguists have provided various models and classifications of the basic concepts in syntax. For example, Falk (1993) addressed some of the fundamental concepts as achievements of the contemporary linguistics research in chapters 2, 3, 5, and 6 of her book. Akmajian et al. (1996) discussed some of fundamental syntactic concepts in sentence construction. Aarts (1997), Huddleston (1988), Fromkin et al. (1990), Rahimian (2002) and Katamba and Stonham (2012) have also discussed such concepts. Among all works on syntax, Huddleston and Pullum (2002) seems the best source to be used as a framework for the present study LVs in this study. Shariat (2005) defines verb as a word that denotes something or somebody's action or state in the past, present, and/or future or at all times. A verb is a word that stands alone or in combination with its dependents to denote an action in a given time and that is an action that refers to another action (Farshidvard, 2003). Ahmadi Givi and Anvari (2006) write that a verb, usually coming at the end of the sentence, is inflectional so that it adopts various forms. The verb stands alone or in combination with dependents to signify an action, a state, an attribute, or imputation. The verb contains the concepts of person (first, second, third), number (singular and plural), and tense (past, present, future).
Meshkatodini (2011:86) categorizes verbs as simple and linking verbs. He further distinguishes three types of linking verbs including prefixal, compound and copulas.

Literature review
The literature review section can be divided into two lines of discussion. First, there will be a discussion on the studies that have directly addressed LVs. Second, a body of literature will be discussed that has addressed other issues such as compound verbs but has marginally referred to LVs either directly or indirectly. In this section, we review some of the most important studies on the issue.

LV
According to Huddleston and Pullum (2002), LVs either bear no meaning or have a small share in denoting the predicate meaning. In these constructions, the main meaning of the predicate lies in the element that appears either as an NP in the form of a verb complement or as a non-verbal element of the compound verb.
LVs are typically known as a distinct type of syntactic categories that have a small semantic share in verb phrases. Jesperson (1940) coined the term LV. Mohamad and Karimi (1992) believe that LVs constitute the verbal element of Persian compound verbs. The verbal element is thought to be empty of meaning. They argue that only a noun -that has a theta relation to the verb -can appear in the complement structure, i.e. the structure that constitutes an NP core and its theta themes. Brugman (2001) contends that LVs are constructions that bear little or no meaning in verb constructions. LVs are often regarded as a special category that, despite other language categories, has no clear syntactic and sematic features. This has prompted many linguists to overlook the syntax and semantics of LVs in their analyses. They often assign LVs the roles that contain the syntactic requirements of the sentence (Soltani et al., 2017).
Mansouri (2013) asserts that a verbal construction is an LV whenever the non-verbal element takes a bigger share than the verb in expressing the predicate. Like other verbs, LVs can account for different interpretations depending on other contextual elements. According to Mansouri (2013), the sematic lightness of LVs is variable so that it falls on a continuum. Neither is an LV semantically rich enough to be called a lexical verb, nor is it too weak to entail only non-semantic features. He believes that LVs stretch on a continuum of sematic emptiness such that they approximate lexical verbs on the one end and become almost completely semantically bleached on the other (Mansouri, 2013:95). Sometimes, the semantic content may become so weak that the verb denotes more of time and aspect than meaning. However, the bleaching sometimes is not too strong to completely deplete the verb of meaning. Some researchers believe that the verbal element may become too empty of content to have a syntactic function only. However, as we shall see, an LV may never become completely empty of content. Karimi Doostan (2003) conducted an extensive study on the dependence or independence of LVs and non-verbal elements on a special argument structure. He cited examples such as "sokut kardan" (remain silent), "rahnamăee kardan" (guide), and "ehda kardan" (donate) to emphasize that Persian LVs are not dependent on a specific argument structure. He maintains that a light verb is not merely a carrier of tense, person, and number features. Considering the polysemy of LVs in the Brugman's theoretical framework, Khanbaz (2013) studied the verbs "zadan" and "kešidan" whereas Amuzandeh and Bahrami (2012) examined the verbs "kardan" and "dăštan". They studied the polysemic relationship between LVs and their counterparts and showed how the pictorial schemas of heavy verbs are projected to their LV counterparts in compound verb constructions. Vahidian Kamyar and Omrani (2000) believe that other constituents may not be inserted between the verbal and non-verbal elements of the LV". Karimi (2005) explains that the verbal element (LV in his words) and the complement (non-verbal element in his words) are generated separately but are combined later in the language syntax. Following their combination, they function as a sematic unit as does a lexical unit. He describes this as a conflicting property of Persian complex predicates.

Aspect
Aspect refers to sematic features that reflect the events in the real world. Two types of aspects may be distinguished: grammatical and lexical.

Grammatical aspect
The grammatical aspect denotes the difference between perfective and imperfective forms of the verb as follows: A) "mădar-am qază ră poxt" [perfective] [My mother cooked the food.] B) "Mădar-am qază ră mi-poxt" [imperfective] [my mother used to cook food.] The difference between grammatical and lexical aspects is that the latter exists within the verb as a sematic feature while the former adopts these features through external and inflectional elements. Situations denoting a state entail a static situation (e.g. the house is big, I know her). Event indicate dynamic situations (e.g. My father walked in the park). Achievements occur instantaneously and lack any lack any continuation. (e.g. fall down). Achievements can also be repetitive such as cough.

Research method
There are a large number of verbal constructions that constitute LVs. The present study, focuses on the verbs "zadan", "dădan", and "xordan". To this end, we extracted all verbal constructions carrying "zadan", "dǎdan", and "xordan" from the Updating Persian Corpus using AntConc software based on Mansouri's approach (2013). We used the random sampling method that has scientific validity. This sampling method lends itself to generalizability and gives equal chance of selection to all members of the research population. Then we analyzed all sentences containing LVs in terms of the lexical aspects such as state and activity. The results were illustrated in tables and figures with descriptions of every verb. Only was a single pattern reported from among similar patterns. The LV patterns were listed and analyzed in an inventory. Both descriptive and inferential statistics were used to analyze the data. The study adopted a descriptive-analytical method and a corpus-based survey design. Descriptive studies have both applied and fundamental facets. They may lead to the discovery of scientific facts and attainment of a body of knowledge in an inductive way. In descriptive studies, the researcher makes no intervention in the situation, status, and role of variables. We neither manipulate nor control the variables. In fact, we merely describe what already exists. We carried out library studies and did content analysis. We extracted and analyzed the sentences containing LVs from the Updating Persian Corpus. In order to examine the lexical aspects in "zadan", "dǎdan", and "xordan", we carried out the study in five phases as follows: 1. Samples of "zadan", "dǎdan", and "xordan" were extracted from the Updating Persian Corpus.
2. The compound patterns of LVs were analyzed in terms of lexical aspect. 3. All verbs were studied in terms of lexical aspect. 4. The data were entered in Excel application. 5. Statistical analysis was carried out in Excel.

Data collection method and instruments
The research instrument included the Updating Persian Corpus available freely online at http://www.peikare.ir. The sample size was calculated using the Krejcie and Morgan's table (1970). Following the data collection phase, we amassed a large bulk of data to analyze. Thus, we extracted and tabulated the data manually.

Research population
The research population consisted of the Updating Persian Corpus in total. This is a monitor of Persian news stories updated automatically on a daily basis by crawling the political, economic, social, cultural, scientific, and sports news from news agencies. The non-news texts will be added to the corpus in the future. The corpus covers a wide variety of language genres so that there is an equal proportion of genres in the corpus. The data are updated daily; thus, the corpus is representative, balanced, and upto-date. In the first phase of corpus development, the news texts have been collected from ISNA news agency since 2017. In the next phases, however, the data will be added to the corpus from other news agencies. The corpus size amounted to 48M words approximately by the end of December 2018. The corpus has an advanced search tool that can be used to search in the database with high accuracy and speed. Moreover, normalization and stemming are automatically done on the data every day, and the words are tagged for the parts of speech.

Data analysis
The Persian compound verbs constitute an innumerable number of non-verbal and a limited number of verbal elements. In this study, we first extracted all sentences containing the verbs "zadan", "dǎdan", and "xordan". Then we used the Krejcie and Morgan's table (1970) to calculate the sample size. Based on Mansouri's approach (1391), we separated the LVs from heavy, compound prefixed, and incorporated verbs. Eventually, we studied the LVs in terms of their lexical aspects. The results were entered in Excel and analyzed in terms of the number of distinct lexical aspects, patterns, and differences of LVs. The data were managed in Excel and analyzed in SPSS software.

Results and recommendations
The main findings of the research are presented in this section. Before reporting the research findings, it is necessary to elaborate on the research hypotheses. The results confirmed all but the second and sixth hypotheses. The second hypothesis posits, "all state verbs are transitive, and their complements are necessarily direct objects." The findings did not confirm this hypothesis, though. In the LVs made up with "Dǎdan", there were 273 state verbs while there were only 72 transitive verbs. Thus, the second hypothesis was rejected. One, however, should note that we addressed linguistic action and language application such that informal constructions were used in the study. Thus, the direct object marker "Ra" was often missing in the samples.
According to the sixth research hypothesis, "activity verbs are used more frequently than state verbs." The results did not confirm this hypothesis, though. The results of studying a total of 1042 verbs revealed a number 512 state and 512 activity verbs in the sample. That is, activity verbs were not more common than state verbs. Rather, they distributed evenly across the corpus. The results of LV analyses are discussed below.

Results of LV constructions with "zadan"
The LV "zadan" is semantically bleached to a variable degree in different constructions. Some verbs like "kotak zadan" (beat) and "sili zadan" (slap) are used as almost lexical verbs while others like "labxand zadan" (smile) are semantically bleached to some extent. In some verbs such as harf zadan" and "telefon zadan", the verbal element is almost semantically bleached. "zadan" is an LV with high frequency of usage. We extracted about 74000 sentences from the corpus that contained "zadan". The sample size was calculated to be 498 items using the Krejcie and Morgan's table (1970). All LV constructions containing "zadan" were analyzed one by one in terms of the lexical aspect. Only one pattern was explained in case of similar patterns. The selected patterns were then analyzed one by one. The verb "zadan" has 1689 entries in Moin Encyclopedic Dictionary. LVs containing "zadan" may denote various meanings including beating, playing an instrument, talking, speaking, mixing with oxygen, clearing, jumping, biting, and stealing (Homayon Farrokh, 1985:931). There were 498 instances of the verb "zadan" in the research sample including 458 LV constructions (92%) and 40 heavy or incorporated verbs (8%) (see Table 5.1).

"zadan" as state and activity verb
The results of lexical aspects analysis showed that 43% (n=198) of LV constructions were state and 57% (n=260) were activity verbs (see Figure 5.2).  One should note that instantaneous and process genres are only observed in the activity verbs while state verbs lack such categories.

Simple counterparts
With regard to standalone counterparts of LVs, the results showed that only 9% (n=41) of LVs had simple counterparts while 91% (n=417) had no simple equivalents (see Figure 5.3).

Expandability
LVs are expandable if they can accommodate more constituents between the verbal element and complement. For example, "tafa?ol zadan" (to have divination) may be split to accommodate "be qazaliăte pormohtavăye Hăfez" (on the rich sonnets of Hafiz). Therefore, we will have a sentence like "Ali tafa?oli be qazaliăte pormohtavăye Hăfez zad" [Ali had a divination on the rich sonnets of Hafiz]. However, the verbal element and complement cannot split to lodge more constituents in the verbs like "zang zadan" (to ring) and "rekăb zadan" (to pedal). As shown in Figure 5.4, 43% (n=199) LV constructions with "zadan" are expandable while 57% (n=259) are non-expandable.

Results of LV constructions with "dǎdan"
There were a number of 5522 sentences containing LV constructions with "dǎdan" in the corpus. A number of 328 sentences were selected as the sample size based on the Krejcie and Morgan's table (1970). The verb "dǎdan" is repeated 1552 times in Moin Encyclopedic Dictionary (1992). Some of the verb constructions with "dǎdan" are obsolete or scarcely used in modern Persian. LV constructions with "dǎdan" are either achievement or durative verbs. The compound constructions of "dǎdan" in the corpus were mostly transitive.

Simple counterparts and expandability of "dǎdan"
As shown in Figure 5.7, a number of 293 (90%) LV constructions with "Dǎdan" have no simple counterparts while 32 (10%) have simple equivalents.

Figure 5.7 LV constructions with "Dǎdan" by absence or presence of simple counterparts
As illustrated in Figure 5.8, a number of 85 (26%) LV constructions with "dǎdan" are expandable while 240 (74%) are nonexpandable.  Table 5.4, there are a number of 39 achievement verbs including 6 instantaneous and 33 process verbs in LV constructions with "dǎdan". There are also 13 durative verbs including 4 unbounded and 9 bounded verbs. One should note that LV constructions with "Zadan" constituted 52 activity verbs.

"Xordan" as transitive and intransitive verb
All LV constructions with "Xordan" in the sample are intransitive. Therefore, the results are not illustrated in the figure.

"Xordan" as state and activity verb
As shown in Figure 5

Expandability
As shown in Figure 5.11, 225 LV constructions with "xordan" are expandable while 16 are non-expandable.

Transitivity
As shown in Figure 5.13, there are a number of 127 transitive LV constructions with "zadan" so that they outnumber the transitive LV constructions with "dǎdan" by 55 items. Besides, intransitive LV constructions with "zadan" amounted to 331 items that outnumber "Dǎdan" by 78 and "xordan" by 90 items. Intransitive LV constructions with "dǎdan" rank second with 253 items. Intransitive LV constructions with "Xordan" are 12 items short of those with "Dǎdan". As discussed above, the verb "Xordan" lacks transitive genus in the sample corpus.

Sate versus activity verbs
As shown in the figure below, state LV constructions with "Dǎdan" outnumber those with other verbs by far. As illustrated in Figure 5.14, of the total 1024 verbs, 512 items are state and 512 are activity verbs. LV constructions with "Zadan" of the activity type by far outnumber those with other verbs of the same type. In other words, they outnumber LV constructions with "Xordan" by 60 and those with "Dǎdan" by 208 items. Moreover, state LV constructions with "Dǎdan" outnumber by far those with other verbs of the same type. That is, they outnumber LV constructions with "Xordan" by 232 and "Zadan" by 75 items. Figure 5.14 Comparison of LV constructions with "zadan", "dǎdan" and "xordan" as state and activity verbs

Simple counterparts
As shown in Table 5.7 and Figure 5.15, a number of 415, 293 and 80 LV constructions with "zadan", "dǎdan" and "xordan" have no simple counterparts, respectively, which are 788 items in total. However, a number of 161, 32 and 43 LV constructions with "xordan", "dǎdan" and "zadan" have simple equivalents, respectively, which make up 236 items altogether. Table 5.7 Number of simple counterparts of LV constructions with "zadan", "dǎdan" and "xordan" As Shown in Figure 5.15, LV constructions with "dǎdan" have fewer simple counterparts than those with other verbs. Moreover, LV constructions with "xordan" rank first in terms of the number of simple counterparts.

Expandability
A number of 418 LV constructions were expandable and 605 were non-expandable in the research sample (see Table 5.8).  With regard to the absence of simple counterparts, LV constructions with "dǎdan" and "xordan" have a significant difference with other verbs. As shown in Figure 5.16, LV constructions with "xordan" are more expandable than those with other verbs. Besides, LV constructions with "dǎdan" are the least expandable items.

Conclusion
The three verbs looked into in this study have two functions. They are used as both lexical and light verbs. However, their uses as light verbs are far more frequent compared to their functions as lexical verbs. The three verbs revealed a variety of features in different syntactic constructions. Our study confirmed four of the six hypotheses presented at the beginning of the paper. Among all instances of the three verbs appearing in different verbal constructions, Zadan is most frequent while Xordan is the least. The results of the study show that Zadan has most transitive uses than the other two verbs while Xordan never is used as

Recommendations
According to the theoretical domain of the research, the following recommendations may be made for future research: 1. It is recommended that future studies examine larger samples.
2. The present study addressed LV constructions with "zadan", "dǎdan" and "xordan". It is recommended that future studies examine other verbs as well. This will increase the generalizability of the results and allow for comparisons. 3. It is recommended that future studies examine other corpuses as well.