Navigation – Plan du site

AccueilNuméros36A Textual Study of Turkish Overt ...

A Textual Study of Turkish Overt and Null Singular Personal Pronouns in Novels

Étude textuelle des pronoms personnels singulier dans les romans en turc
Buse Şen Erdoğan et Emine Yarar

Résumés

Cette étude examine l’usage des pronoms sujets singuliers, exprimés et non exprimés (pronoms nuls), en turc, dans des romans. Une attention particulière est portée aux pronoms en relation avec leurs antécédents, dans le cadre de la théorie de l’accessibilité de Mira Ariel. Le corpus d’étude comprend 300 phrases extraites de 12 romans écrits en turc. Les antécédents, ainsi que les formes exprimées et nulles des pronoms personnels au singulier ont été codés selon les dimensions de récence, de saillance discursive (givenness) et de proéminence syntaxique. Les données recueillies ont été analysées à l’aide des tests U de Mann-Whitney et du khi carré. Les résultats indiquent que les dimensions de récence, de saillance discursive et de proéminence syntaxique impactent différemment l’usage de ces pronoms. Par exemple, l’emploi des pronoms exprimés et nuls à la troisième personne est influencé à la fois par la récence et la proéminence syntaxique, tandis que la dimension de saillance discursive affecte les pronoms de première personne, mais pas ceux de deuxième ou de troisième personne. Les deux formes des pronoms de deuxième personne ne sont pas déterminées en fonction de ces dimensions, ce qui constitue un résultat notable. Les résultats de l’étude suggèrent que tous les pronoms personnels ne répondent pas de la même manière aux contraintes du discours. Il apparaît également que, parmi les trois pronoms personnels au singulier, le pronom de troisième personne est le plus sensible à ces contraintes. Ces résultats confirment que les dimensions de récence, de saillance discursive et de proéminence syntaxique sont sujettes à des variations interlinguistiques, ce qui devrait être pris en considération lors de l’analyse de la cohérence discursive.

Haut de page

Texte intégral

1. Introduction

  • 1 This article is based on the MA thesis of the first author. The data that support the findings of t (...)

1This article1 aims to explore how overt and zero singular pronouns are employed and structured in novels written in Turkish. It focuses on the use of these pronouns to see if there are any clear patterns of their organization. In fiction, pronouns function as a key device for maintaining coherence and, therefore, the way in which they function can provide insights into how readers make sense of information. Regarding the discourse anaphors, there have been many views and models but in languages with flexible word order such as Turkish, little attention has been paid to the specific ways in which these pronouns are used in novels. Another characteristic of Turkish which is directly related to discourse anaphors is its pro-drop or null-subject feature.

2The effects of recency, givenness, and syntactic prominence on the choice of overt or zero pronouns are not examined in Turkish. Ariel’s Accessibility Theory (1988) accounts for the use of pronouns in discourse providing an opportunity to reveal this understudied topic about Turkish pronominal system. The aim of the study is to explore to what extent saliency variables of recency, givenness, and syntactic prominence impact the choice of overt or zero personal pronouns in novels. By looking at how these pronouns work, this study seeks to shed more light on Turkish discourse anaphora offering insights into how these variables affect the choice of singular pronouns in a syntactically very flexible and a pro-drop language.

3The structure of the paper is as follows: Section 2 summarizes views on the accessibility of personal pronouns and Turkish personal pronouns. The materials and method of the study are introduced in Section 3. Section 4 reports the results. In Section 5 the findings of the study are discussed. Section 6 presents the conclusion of the study.

2. Background information related to the study

4In this section some information about Accessibility Theory and Turkish personal pronouns is given.

2.1. Accessibility of personal pronouns

5Ariel’s accessibility hierarchy (1988, 1990, 2001) is a fundamental theory on pronoun resolution that emphasizes the role of such variables as recency, saliency, and syntactic prominence. Arnold (2010) states that cognitive load and referential predictability have an impact on the processing of pronouns. While such studies establish fundamental principles of accessibility, there has been limited research on these variables in Turkish, particularly in fictional texts. The present work attempts to bridge this gap by looking at overt and null singular pronouns in Turkish novels to uncover how these accessibility factors function in a pro-drop language with rich morphosyntactic features.

6The antecedent of personal pronouns is generally a full noun phrase introduced earlier in the discourse, designating the intended representation of the pronoun (Lehecka, 2023). Ariel (1988) defines pronouns as high-accessibility markers that are generally placed closer to their antecedents. In her framework, four factors affect the accessibility rate of pronouns: distance, competition, saliency, and unity (Ariel, 1990). It is argued that accessibility increases when there is less distance between the antecedent and the anaphor and when there are fewer competitors. In addition, when the antecedent is the topic, there is a higher degree of accessibility, and when the referring expression and the antecedent are unified, the likelihood of accessing the referent increases.

7In a similar vein, Arnold (2010) identifies four discourse properties that impact the accessibility of referring expressions: (1) givenness, (2) recency, (3) syntactic prominence, and (4) thematic prominence. Of these, recency, givenness, and syntactic prominence are considered as factors of saliency. Recency refers to the proximity of the antecedent to the referring expression in terms of words, sentences, or paragraphs (Ariel, 2001). Following Givón (1983), Arnold (2010) suggests that a closer antecedent is more likely to be referred to by a pronoun than one that is further away. Similarly, Gundel et al. (1993) argue that the recency of an antecedent enhances its focus within a given context. This increased focus influences its cognitive status, making the antecedent more accessible and, consequently, more likely to be referenced by a pronoun.

8There are sentence processing studies which highlight the importance of recency in pronoun resolution. For instance, Duffy and Rayner (1990) demonstrate using an eye-tracking experiment that pronouns are comprehended more quickly and accurately when their antecedents are mentioned nearby compared to when they are placed farther away in the text. Similarly, Garrod and Sanford (1994) report that recency reduces processing costs significantly in that more recent antecedents remain more active in memory and require fewer retrieval resources. Furthermore, the study by O’Brien (1987) about searching for antecedents in the memory reveals that recent antecedents are comprehended faster than earlier antecedents.

9In this study, givenness, as a discourse property, refers to how frequently an antecedent has been mentioned before a pronoun in the text. It determines whether a referent is considered given or new information. While some studies treat givenness as a binary distinction, others adopt a more nuanced perspective. For instance, Baumann and Riester (2013), building on Prince (1981), propose four levels of givenness: (1) new, (2) inferable/accessible/bridging, (3) situationally evoked/indexical/deixis, and (4) textually evoked/discourse-old/given/active entities.

10According to Prince (1981), new referents are those not previously mentioned in discourse. Inferable referents are discourse-new but linked inferentially to previously mentioned expressions. Evoked referents encompass both endophoric and exophoric references – some are explicitly mentioned in discourse, while others exist in the conversational context. Finally, given referents are those overtly mentioned earlier in discourse. Despite this complexity, experimental studies often consider givenness as a binary category for the sake of easier coding. Similarly, in this study, givenness is operationalized as a binary distinction (discourse-given vs. discourse-new) but the degree of givenness is also considered by tracking prior mentions of both overt and null pronouns.

11Typically, pronouns do not refer to new entities; instead, new entities are introduced using definite descriptions (Gundel et al., 1993). However, exceptions exist where pronouns may refer to new information. Jaeger and Wasow (2005) propose the concept of “derived accessibility,” suggesting that a referent’s accessibility is not solely intrinsic but is largely influenced by contextual factors.

12The syntactic role of the antecedent significantly influences the accessibility of pronouns, particularly regarding subjecthood. Syntactic prominence refers to an antecedent’s status as the subject and its sentential position. Crawley et al. (1990) argue that the antecedent of a pronoun is typically interpreted as the subject in the preceding context. Similarly, Arnold (2010) argues that when syntactic prominence is controlled, speakers prefer to co-refer to a referent stimulus using a pronoun rather than a referent experiencer in transitive sentences. However, it is also stated that in languages like English, where subjects are generally positioned sentence initially, this initial placement may contribute to the tendency for pronouns to co-refer with subjects. To separate these effects, Gordon et al. (1993) conducted a reading-time experiment to determine whether it is subjecthood or sentence-initial position that accounts for a pronoun’s co-referentiality with the subject. Their findings indicate that both factors independently influence the prominence of antecedents.

13Crawley et al. (1990) examine the subject antecedent strategy with the parallel function strategy which suggests that a pronoun is interpreted as co-referential with a preceding noun phrase occupying the same grammatical position. For example, a subject pronoun is linked to a preceding subject noun phrase, while an object pronoun is co-referential with a preceding object noun phrase. In contrast, the subject antecedent strategy posits that the antecedent of a pronoun is typically a noun phrase in the subject position of the preceding sentence, regardless of the grammatical position of the pronoun. The results of this study provide stronger support for the subject antecedent strategy over the parallel function strategy.

14Various studies have been conducted on the accessibility of personal pronouns. For instance, Cameron (1997) analyses personal pronouns and their null counterparts in Spanish using Accessibility Theory. The study concludes that split second-person plural pronouns are not as common as non-split antecedents, which are considered less prominent. These findings are not consistent with the assumption of Accessibility Theory that if the antecedent is less prominent, speakers would prefer a marker with a lower degree of accessibility. Kronrod and Engel (2001) investigate newspaper headlines written in Hebrew using Accessibility Theory. They conclude that accessibility principles are not the only determinants of pronoun choice in the sample, as other factors, such as brevity and curiosity-arousal, interfere with accessibility factors in newspaper headlines. Gutman (2004) studies null subjects in Hebrew, Finnish, and Romanian – languages that all allow pro-drop – to test the applicability of Accessibility Theory to such subjects. She reports that the factors of salience and unity have partial effects on pronoun choice in these languages.

15Dimension of givenness is found to have distinct effects in each language. Carminati (2002) analyses the effect of syntactic prominence related to the pronoun choice in Italian using a self-paced reading experiment. It is found that reading time increases when the antecedent of a null pronoun is a noun phrase in the object position in that null pronouns require more prominent antecedents. Similarly, reading time is longer when an antecedent in the subject position is linked with an overt pronoun since overt pronouns are mostly co-referential with less prominent antecedents.

16Portele and Bader (2016) examine German personal pronouns and demonstrative pronouns based on the principles of recency, givenness, and syntactic prominence. Based on corpus findings, they conclude that personal pronouns and demonstrative pronouns significantly differ in terms of syntactic prominence and givenness. However, recency is found not to distinguish between them. They also administer a sentence-completion test, the findings of which indicate that only syntactic prominence makes a significant difference between personal pronouns and demonstrative pronouns.

17All these findings suggest that there are cross-linguistic variations regarding the three principles of givenness, recency, and syntactic prominence. This study extends these findings to Turkish, a null-subject language with an agglutinative structure.

2.2. Turkish personal pronouns

18Turkish is a null-subject language allowing the omission of subject pronouns. This is typically explained by its rich verbal morphology, which identifies the grammatical subject with a high level of specificity, making explicit subject pronouns generally unnecessary. Turkish subjects, whether null or overt, agree with the verb in terms of person and number and are not marked for gender or animacy. Turkish personal pronouns and the related verbal endings are illustrated in Table 1.

Table 1 – Subject pronouns and verbal personal endings in Turkish (adapted from Göksel, Kerslake, 2005: 230)

Turkish Personal Pronouns Verbal Endings
1st person singular ben (“I”) -m oku-du-m (“I read.”)
read-PST-1SG
1st person plural biz (“we”) -k oku-du-k (“We read.”)
read-PST-1PL
2nd person singular sen (“you”) -n oku-du-n (“You read.”)
read-PST-2SG
2nd person plural siz (“you”) -nIz oku-du-nuz (“You read.”)
read-PST-2PL
3rd person singular o (“he/she/it”) oku-du-∅ (“he/she/it read.”)
read-PST.3SG
3rd person plural onlar (“they”) /-lAr oku-du-(lar) (“They read.”)
read-PST-(3PL)

19As illustrated in Table 1, most pronouns correspond to specific verbal endings, which indicate both the doer of the action and the identity of the referent. Because these verbal endings make the subject easily identifiable, pronominal subjects in Turkish are often omitted. However, 3rd person singular and plural pronouns differ from others in this regard. Neither form of the 3rd person pronouns has an explicit personal ending. The plural form may include an optional plural marker -lAr.

20The use of overt or null pronouns in Turkish does not alter the basic referential meaning of a sentence. However, they differ pragmatically. Enç (1985) argues that sentences with overt pronominal subjects are more marked than those with null pronominal subjects, as the former can convey additional pragmatic information. Enç (1986) further claims that overt subjects indicate a topic shift in discourse, whereas null pronouns typically maintain the existing topic.

21Göksel and Kerslake (2005: 241–242) state that overt personal pronouns in Turkish are required in finite clauses under the following conditions:

  1. where a subject contrasts with that of the preceding sentence;
  2. where the subject is focused;
  3. where a 1st or 2nd person subject is one of a set of people actually or potentially involved in some action or situation;
  4. where a 3rd person subject is an entity that was introduced in a non-subject role in the previous sentence;
  5. where, despite the continuance of the same subject, there is a shift from a statement about a specific event to a generalization about the person involved;
  6. in the opening sentence of a conversation, or a sentence in which the speaker introduces a new topic of discussion.

22The use of overt pronominal subjects is mainly shaped by the discourse context, serving pragmatic functions such as contrast, similarity, emphasis, or the introduction of new information (Enç, 1986; Göksel, Kerslake, 2005). In Turkish, the discourse topic is typically introduced with a full noun phrase or an overt pronoun, while referential continuity is maintained with null pronouns. However, when there is a shift in the topic, this is marked through the utilization of an overt pronoun (Öztürk, 2001), as null pronouns cannot fulfil this role.

3. Materials and methods

23This section outlines the materials used in the study and details the procedure of the analysis. Additionally, the methods employed for data analysis are discussed in this section.

3.1. Selection of the texts and procedure

24A corpus was constructed from twelve novels written by different Turkish writers across distinct time periods to eliminate any potential personal bias of the authors in the use of personal pronouns. The novels were downloaded from an open-access website. Full references for the novels are given in bibliography. Sentences containing overt and null singular pronominal subjects were extracted, yielding a total of 300 sentences. All sentences were simple constructions, each comprising a subject and a predicate. For each pronoun type, 50 sentences with overt and 50 sentences with null singular pronominal subjects were included in the sample.

25Following the extraction, the antecedents of the pronominal subjects were identified. The sample was then analysed based on three saliency-related properties: recency, givenness, and syntactic prominence. These properties of the antecedents were manually coded.

26Recency was determined by counting the number of sentences between the sentence containing the anaphor (overt or null pronoun) and the sentence containing its antecedent. If the antecedent was in the immediately preceding sentence, the recency value was coded as 0.

27Givenness was evaluated by counting the number of noun phrases in the discourse that referenced the antecedent, including the antecedent noun phrase itself.

28Syntactic prominence was analysed by considering whether the antecedent occupied a subject position.

29The data coding process is illustrated as follows:

[1] Kardeşim bir aktar dükkanı önünde durdu. θ bir bebeği seyretti.
My sister stopped in front of an herbalist shop. She watched a baby.
(Eroğlu, 2006: 88)

30In this example, the null form of the 3rd person pronoun is used in the second sentence. The recency (the number of sentences between the anaphor and its antecedent) is 0 in that the antecedent, kardeşim (“my sister”), appears in the immediately preceding sentence. Concerning the givenness dimension, the antecedent is coded as 1. Because it is mentioned once. Regarding syntactic prominence, the antecedent is the subject. Then, the final coding for this sentence is as follows: recency = 0, givenness = 1, syntactic prominence = subj.

[2] “Dert etmeyin,” dedi Antonio. “Ben sizi arabama alabilirim.”
“Don’t worry!” said Antonio. “I can take you to my car.”
(Aysu, 2011: 201)

31In [2] the pronoun ben (“I”) is overtly stated, and its nearest antecedent is Antonio. Since there is no intervening sentence, recency is coded as 0. The code for the givenness is 1, as there is only one mention of the antecedent. The antecedent Antonio is the subject of the sentence, yielding a coding of recency = 0, givenness = 1, syntactic prominence = subj.

[3] Az sonra Efe odaya geliyor.
“N’aber?” “Eh işte, senden n’aber?”
“Çok sıkıldım, zor kaçtım içerideki hatundan.”
“Kadına katlanamıyorsan onunla neden yatıyorsun ki?”
“Hadi ama gene başlama. Senin de bir kadına ihtiyacın var, inan bana.”
“Doğru diyorsun aslında.”
Soon after, Efe comes to the room.
“What’s up?” “So so, what about you?”
“I’m too bored; I have barely escaped from the woman inside.”
“If you cannot stand the woman, why are you sleeping with her?”
“Come on, don’t start again. You also need a woman, believe me.”
“You are actually right.”
(Keskin, 2015: 120)

32In this example, the null 2nd person pronoun is utilized as the subject. Its nearest antecedent is the proper name Efe. There are six sentences between the subject and its antecedent, so the recency dimension is coded as 6. Concerning givenness, there are two references to the antecedent: the antecedent Efe itself and the dative personal pronoun bana (“to me”), yielding a value of 2. Regarding syntactic prominence, the antecedent Efe functions as the subject of the sentence. The final coding for this example is as follows: recency = 6, givenness = 2, syntactic prominence = subj.

33Exophoric references, such as 1st and 2nd person pronouns, typically do not rely on textual antecedents in the same way as 3rd person pronouns. These pronouns are primarily used to refer to the speaker (1st person) or the listener (2nd person). However, in narrative genres such as novels, 1st and 2nd person pronouns can occasionally take on a textual antecedent. In these contexts, the pronouns may refer to characters within the narrative but the identity of the referent must be established by the author earlier in the text to ensure that readers can correctly interpret the reference.

34Consider the following excerpt:

[4] Sonra mahzun mahzun müşteriye döndü ve annesinin iyi bir kadın olduğuna onu tanık getirmek, ondan yardım dilemek istedi:
– Öyle değil mi efendim?
– Ben ne bileyim?
Later, he sadly turned to the customer, and he wanted to call him in evidence that his mother was a good woman; he wanted to ask his help:
– Isn’t it sir?
– How could I know?
(Celal, 2001: 72)

35In this example, the pronoun “I” in the response “How could I know?” does not refer to the author but rather to the character speaking in the context, the customer. Understanding the identity of the speaker is challenging unless the author has previously provided sufficient cues. Here, the character is introduced through earlier mentions, such as the use of the pronouns onu (“him”) and ondan (“from him”), which guides the comprehension of the identity of the referent. These mentions contribute to the givenness of the 1st person pronoun “I”, which yields a givenness value of 3.

36It should be also noted that 1st and 2nd person pronouns in novels are often used in dialogues which are marked by quotation marks.

3.2. Data analysis

37The Mann-Whitney U test was utilized to analyse the recency and givenness factors between overt and zero subject pronouns. This non-parametric test is appropriate for comparing two independent groups when the data do not have a normal distribution (Field, 2013). It assesses whether the ranks of one group differ significantly from those of the other, making it suitable for evaluating differences in the number of intervening sentences or mentions between antecedents and pronouns.

38The Chi-square test was employed to investigate the relationship between the form of the pronoun (overt vs. zero) and the syntactic role of its antecedent (subject vs. non-subject). Given that both variables are categorical, the Chi-square test is appropriate for determining whether the distribution of pronoun types varies significantly based on the antecedent’s syntactic role.

39To ensure the reliability of the findings, an independent rater analysed all sentences extracted from the novels. Interrater reliability was assessed using the Kappa coefficient, which measures the level of agreement between coders. In the Kappa analysis, values ≤ 0 indicate no agreement, values between 0.01 and 0.20 indicate slight agreement, values between 0.21 and 0.40 represent fair agreement, values between 0.41 and 0.60 reflect moderate agreement, values between 0.61 and 0.80 denote substantial agreement, and values between 0.81 and 1.00 indicate almost perfect agreement (StATS, n.d.).

40The mean Kappa coefficient in this study was calculated as 0.72, demonstrating substantial interrater agreement.

4. Results

41The results of the study are presented below in relation to the principles of recency, givenness and syntactic prominence.

4.1. Results based on recency

42As stated earlier, recency refers to the number of sentences between the antecedents and the pronouns. The data on recency were analysed using the Mann-Whitney U test. The results of the analysis on the 1st person singular pronoun ben (“I”) are presented in Table 2 below:

Table 2 – Results of the Mann-Whitney U test concerning the number of sentences between the antecedent and overt or null subject ben (“I”)

Recency Number of Sentences Mann-Whitney U (U) p
Mean rank SD Median (mdn) Min. Max.
“I” Overt subject 4.1 4.8 2.5 0.0 1.0 1189.500 0.673
Zero subject 3.2 4.0 2.0 0.0 23.0

43Table 2 shows that the recency is slightly higher for overt 1st person pronouns (mdn = 2.5) than for null 1st person pronouns (mdn = 2.0), although the difference between these two pronouns is not statistically significant (p = 0.673). There is no statistically significant difference between overt and null 1st person singular subjects in terms of recency (p > 0.05). Based on this finding, it can be argued that the number of sentences between antecedents and 1st person pronouns in novels does not play a significant role in determining whether overt or null forms are selected.

44The findings also indicate that there may be two or three sentences between 1st person singular subjects and their antecedents. This suggests that both overt and null forms of 1st person singular subjects can refer to antecedents that may not appear in the immediately preceding sentence. It is not consistent with the recency assumption. Because it predicts that zero subjects are used immediately after their antecedent sentence due to the difficulty in accessing them. However, in Turkish, such a difficulty does not appear to exist, likely due to the verbal morphology. As mentioned above, verbal morphology in Turkish contains a person marker facilitating the accessibility of null pronominal subjects.

45The recency of the overt and null 2nd person singular pronoun sen (“you”) was also analyzed using the Mann-Whitney U test. The results of the test are presented in Table 3:

Table 3 – Results of the Mann-Whitney U test concerning the number of sentences between the antecedent and overt or null subject sen (“you”)

Recency Number of sentences

Mann-Whitney U (U)

p
Mean rank SD Median (mdn) Min. Max.
“You” Overt subject 4.4 4.5 3.0 0.0 21.0 1223.500 0.854
Zero subject 5.0 5.5 3.0 0.0 22.0

46As shown in Table 3, the recency factor is nearly equal for both overt and null 2nd person singular pronoun subjects (mdn = 3.0), U = 1223.500, p = 0.854. The median values for the overt and null forms are identical. In other words, there is no statistically significant difference between these two forms in terms of recency (p > 0.05). Therefore, the number of sentences between the antecedent and the 2nd person singular subjects is approximately the same for both overt and null forms.

47In most cases, there are two or three sentences between 2nd person singular subjects and their antecedents. This suggests that these pronouns regardless of being overt or null tend to occur at some distance from their antecedents. This finding is not consistent with the assumption of the recency dimension. As mentioned above, this assumption posits that there should be no intervening sentence between an antecedent and a zero subject. It is possible to argue that Turkish verbal morphology containing person markers allows zero 2nd person subjects to appear in non-immediate sentences.

48The recency of overt or null 3rd person singular pronouns, namely o (“he/she/it”), was analysed using the Mann-Whitney U test. The results of the test are illustrated in Table 4:

Table 4 – Results of the Mann-Whitney U test concerning the number of sentences between the antecedent and overt or null subject o (“he/she/it”)

Recency Number of sentences Mann-Whitney U (U) p
Mean rank SD Median (mdn) Min. Max.
“He/she/it” Overt subject 1.9 3.7 0.0 0.0 18.0 1159.000 0.490
Zero subject 1.0 1.6 0.0 0.0 7.0

49Table 4 indicates that the recency factor is again nearly equal for both overt and null 3rd person pronouns (mdn = 0.0), U = 1159.000, p = 0.490. There is no statistically significant difference between overt and null 3rd person singular subjects in terms of recency (p > 0.05). In most cases, there is only one sentence between the 3rd person singular subject and its antecedent.

50This finding differs from the findings on the overt and null 1st and 2nd person singular pronouns. Because the overt and null 3rd person singular pronouns tend to be positioned much closer to their antecedents. This could be attributed to the fact that Turkish 3rd person singular pronouns lack a distinct personal marker. Furthermore, the finding on the 3rd person singular pronouns partially aligns with the assumption of recency which predicts that zero pronouns should be closer to their antecedents compared to overt pronouns. More specifically, it successfully predicts that zero forms of 3rd person singular pronoun appear closer to their antecedents in the sample. However, overt forms are also found to be closer their antecedents. In other words, there is no difference between overt and null forms of the 3rd person singular pronoun concerning their closeness to the antecedents.

4.2. Results based on givenness

51This section presents the results on the effects of the givenness on the use of personal pronouns. In this study, givenness is defined as the number of mentions of the referent in the sentences between the antecedent and the pronoun. This was analysed using the Mann-Whitney U test concerning the overt and null subject ben (“I”). The results of the test are presented in Table 5.

Table 5 – Results of Mann-Whitney U test on the number of mentions of the referent in the sentences between the antecedent and overt or null subject ben (“I”)

Givenness Number of sentences Mann-Whitney U (U) p
Mean rank SD Median (mdn) Min. Max.
“I” Overt subject 1.5 0.8 1.0 1.0 4.0 932.500 0.015
Zero subject 2.2 1.4 2.0 1.0 6.0

52As shown in Table 5, the number of mentions of the referent in the sentences between the antecedent and the null subject 1st person pronoun (mdn = 2.0) is higher than that for the overt subject ben (“I”) (mdn = 1.0), U = 932.500. This difference is found to be statistically significant (p < 0.05). Therefore, overt and null subject forms of this pronoun significantly differ in terms of the number of mentions of their referents. Given that there is higher number of mentions for the null form, it can be argued that these frequent mentions facilitate the use of null forms for reference continuity. These results are consistent the givenness assumption, which predicts that the givenness feature of null forms is more prominent than that of overt forms.

53The number of mentions of the referent between the antecedent and the overt or null pronominal subject sen (“you”) was also examined using the Mann-Whitney U test. The results of the test are presented in Table 6.

Table 6 – Results of Mann-Whitney U test on the number of mentions of the referent in the sentences between the antecedent and overt or null subject sen (“you”)

Givenness Number of sentences Mann-Whitney U (U) p
Mean rank SD Median (mdn) Min. Max.
“You” Overt subject 2.2 1.3 2.0 1.0 6.0 1154.500 0.486
Zero subject 2.4 2.0 1.0 1.0 8.0

54As shown in Table 6, the number of mentions of the referent for overt 2nd person singular subjects (Mdn = 2.0) is higher than for their null counterparts (mdn = 1.0), U = 1154.500, p = 0.486. However, this difference is not found to be statistically significant (p > 0.05). Therefore, the number of mentions in the sentences between the antecedent and the pronominal has no effect on their choice for the overt and null forms of this pronoun. Both overt and null 2nd person singular subjects appear to have similar numbers of mentions in the sample, typically two or three times. These findings do not support the givenness assumption, which argues that zero pronouns refer to older referents while overt pronouns refer to newer ones.

55The number of mentions of the referent between the antecedent and the overt or null 3rd person singular subject o (“he/she/it”) was also examined using the Mann-Whitney U test. The results of the test are presented in Table 7.

Table 7 – Results of Mann-Whitney U test on the number of mentions of the referent in the sentences between the antecedent and overt or null subject o (“he/she/it”)

Givenness Number of sentences Mann-Whitney U (U) p
Mean rank SD Median (mdn) Min. Max.
“He/she/it” Overt subject 1.3 0.6 1.0 1.0 3.0 1199.500 0.641
Zero subject 1.3 0.6 1.0 1.0 4.0

56Table 7 illustrates that the number of mentions between the antecedent and the 3rd person pronominal subjects is nearly equal (mdn = 1.0), U = 1199.500, p = 0.641. It is also seen that there is no statistically significant difference between the overt and null forms of the 3rd person singular subjects in terms of the number of mentions of the referents (p > 0.05). This result suggests that for the overt and null subject o (“he/she/it”), the number of these mentions has no impact on their choice. In most cases, both overt and null subjects have only one mention in the sample, indicating that the 3rd person singular pronouns, whether overt or null, typically refer to new entities rather than given ones. In short, these findings are not consistent with the givenness assumption, which states that zero pronouns refer to older referents while overt pronouns refer to newer ones.

4.3. Findings based on syntactic prominence

57This section presents findings regarding the effects of the syntactic prominence on the personal pronouns. Here, syntactic prominence refers to the subjecthood of the antecedents. More specifically, it focuses on the choice of the overt and null forms of the personal pronouns based on the grammatical roles of their antecedents. These effects were analysed using the Chi-square test. The results of the Chi-square test about the effects of the syntactic prominence on the choice of overt and null subject ben (“I”) are presented in Table 8.

Table 8 – Results of Chi-square test concerning the syntactic role of the antecedent of overt or null subject ben (“I”)

Syntactic prominence Overt subject Zero subject Total
Antecedent of the subject “I” Subject n 36 39 75
% 48.0% 52.0% 100%
Non-subject n 14 11 25
% 56.0% 44.0% 100%
Total n 50 50 100
% 50.0% 50.0% 100%

χ2 = 0.480

p = 0.488

58As seen in Table 8, there is no significant difference between overt and null 1st person pronouns concerning the syntactic role of their antecedents (χ2 = 0.480, p = 0.488). Therefore, the subjecthood of the antecedents does not significantly influence the choice between overt and null forms of the 1st person singular pronoun. In other words, regardless of whether the antecedent is a subject or non-subject, both overt and null forms may be used in the sample. It should be added that the antecedents of both overt and null 1st person pronouns are generally found to be subjects. The syntactic prominence assumption posits that overt pronouns may have either subject or non-subject antecedents while the antecedents of null pronouns are mostly subjects. However, the results reported above do not support this assumption.

59The Chi-square test was also utilized to examine the syntactic prominence effect for the overt and null subject sen (“you”). The results of the test are presented in Table 9.

Table 9 – Results of Chi-square test concerning the syntactic role of the antecedent of overt or null subject sen (“you”)

Syntactic prominence Overt subject Zero subject Total
Antecedent of the subject “you” Subject n 35 29 64
% 54.7% 45.3% 100%
Non-subject n 15 21 36
% 41.7% 58.3% 100%
Total n 50 50 100
% 50.0% 50.0% 100%

χ2 = 1.563

p = 0.211

60As shown in Table 9, there is no statistically significant difference between overt and null forms of the 2nd person singular pronoun sen (“you”) concerning the syntactic role of their antecedents (χ2 = 1.563, p = 0.211). The results indicate that the antecedents of both overt and null subject sen (“you”) may or may not be subjects. These results do not support the view that overt pronouns mostly have non-subject antecedents compared to null pronouns.

61The results of the Chi-square test for the overt and null subject o (“he/she/it”) concerning the syntactic prominence of their antecedents are presented in Table 10.

Table 10 – Results of Chi-square test concerning the syntactic role of the antecedent of overt or null subject o (“he/she/it”)

Syntactic prominence Overt subject Zero subject Total
Antecedent of the subject “he/she/it” Subject n 26 36 62
% 41.9% 58.1% 100%
Non-subject n 24 14 38
% 63.2% 36.8% 100%
Total n 50 50 100
% 50.0% 50.0% 100%

χ2 = 4.244

p = 0.039

62As shown in Table 10, overt and null forms of the 3rd person singular pronoun significantly differ in terms of having subject or non-subject antecedents (χ2 = 4.244, p = 0.039). These results suggest that null forms of the 3rd person singular pronoun are more likely to have subject antecedents (58.1%) compared to overt forms (41.9%). Conversely, overt forms are more likely to have non-subject antecedents (63.2%) compared to null forms (36.8%). This finding is consistent with the syntactic prominence assumption. Because it argues that null pronouns are more likely to refer to antecedents with higher prominence such as subjects, whereas overt pronouns are more likely to refer to less prominent antecedents such as non-subjects.

5. Discussion

63The findings of the study suggest that the saliency factors of recency, givenness, and syntactic prominence do not exclusively govern the use of Turkish overt and null singular personal pronominal subjects in novels. Each saliency factor interacts differently with the pronominal subjects in the study. The recency and syntactic prominence factors appear to be relevant only for the 3rd person singular pronoun o (“he/she/it”). Conversely, the givenness factor significantly affects the choice of the 1st person singular pronoun ben (“I”) but it does not play a role in the use of the 2nd and 3rd person pronominal subjects.

64As mentioned earlier, unlike other pronouns, the 3rd person singular pronoun in Turkish does not have a specific personal ending on verbs. The correlation between the recency factor and the 3rd person singular pronominal subjects is unsurprising due to its lack of a personal verbal ending. In other words, the authors of the novels sampled tend to use 3rd person pronominal subjects immediately following sentences containing their antecedents to facilitate comprehension. While the recency dimension has traditionally been associated only with null pronominals, it is found that both overt and null forms of the 3rd person pronominal subjects frequently occur immediately after the sentences in which their antecedents appear. Unlike the 1st and 2nd person singular pronouns, the 3rd person pronouns seem to require more proximal antecedents when they function as the subject of a sentence.

65Syntactic prominence refers to the subjecthood of the antecedents of pronominal subjects. It predicts that zero pronouns are more likely to have subject antecedents than overt pronouns due to their lower accessibility. The findings of the study confirm that the null form of the 3rd person singular pronoun is predominantly used with subject antecedents rather than non-subject antecedents. It is reported that the 3rd person pronominal subjects are inherently ambiguous in Turkish. The reasons for this ambiguity stem from the fact that these subjects do not mark the gender of the referent and that the null forms of the 3rd person pronominal subjects lack verbal morphology (Kılıçaslan et al., 2009; Küçük, Yöndem, 2007). All these morphological characteristics indicate that Turkish 3rd person pronouns require more salient antecedents to ensure clarity. This requirement is also documented in the findings of the study.

66The findings further reveal that the null 1st person pronouns are more commonly used with given antecedents than with new ones. This aligns with the givenness factor, which predicts that null pronouns are more likely to have given antecedents than their overt counterparts. Previous findings indicate that given referents are significantly more accessible than new referents (Portele, Bader, 2016: 5). Therefore, the given antecedents of null 1st person pronominal subjects likely enhance their accessibility to readers. This observation also supports the view that Turkish sentences with pronominal subjects are more marked than those with null subjects (Enç, 1985). The presence of given antecedents appears to reinforce the association of null 1st person pronominal subjects with their givenness properties.

67A possible explanation for the greater sensitivity of the 1st person pronouns to givenness is that givenness is a form of derived accessibility. Put differently, it is largely shaped by contextual factors rather than the intrinsic accessibility of the referent (Ariel, 1990; Jaeger, Wasow, 2005). As noted earlier, 1st person singular subjects typically refer to the speaker in discourse. In written texts, identifying the speaker depends entirely on contextual cues. This may explain why the pronoun ben (“I”) is more frequently subject to the effects of the givenness dimension in contrast to the 2nd and 3rd person singular subjects.

68The partial effects of recency, givenness, and syntactic prominence factors on Turkish singular pronominal subjects seem to stem from various factors. It is observed that the 2nd person singular subjects are largely insensitive to these accessibility factors, whereas the 1st person singular subjects are only influenced by the givenness factor. The 3rd person singular subjects, on the other hand, are affected by both recency and syntactic prominence. This variation may be attributed to the hierarchical ranking of pronouns. In this hierarchy the 2nd person singular pronoun, representing the addressee, is the highest-ranked personal pronoun and thus less subject to discourse-level factors. However, the 1st and 3rd person pronouns, being lower in rank, are more open to the effects of these factors.

69The findings of the study indicate that Turkish personal pronominal subjects exhibit distinct discourse and accessibility characteristics in novels. Ariel (2001) argues that while the relative ranking of referential expressions is universal, crosslinguistic variation can occur in two major ways. First, the inventory of referential expressions differs across languages. For instance, null-subject languages include empty pronominal categories in their system, whereas non-null-subject languages do not. Second, the relative positioning of expressions on the accessibility scale may vary across languages. Even in languages with similar referential inventories, morphologically or semantically equivalent expressions – such as personal and demonstrative pronouns – may retrieve antecedents with different degrees of accessibility. This variation are relevant to the findings of the study.

70Even typologically similar languages, like Spanish and Italian, show differences in their sensitivity to accessibility factors. For example, Spanish overt pronouns are more influenced by syntactic prominence than their Italian counterparts (Filiaci et al., 2014). Similarly, within a single language, different personal pronouns may exhibit varying degrees of sensitivity to accessibility factors. In Turkish which is a null-subject and morphologically rich language like Italian, this difference may stem from the phi-features of pronouns, which are encoded morphologically.

6. Conclusion

71This study examines singular overt and null pronominal subjects in Turkish, focusing on their antecedents through a sample of 300 sentences taken from novels. These constructions were analysed based on three saliency factors: recency, givenness, and syntactic prominence. The findings reveal that each factor has distinct effects on the overt and null forms of singular pronominal subjects in Turkish novels. Notably, the 3rd person singular subjects are found to be more influenced by discourse factors, likely due to their unique grammatical features, such as the absence of personal endings. In addition, the study highlights the distinct grammatical characteristics of pronouns in literary texts (Hartung et al., 2016; Jin, Liu, 2022). It appears that pronominal subjects behave differently in response to discourse factors because of their distinct roles in novels.

72The findings presented in this study provide valuable insights into the use of singular pronominal subjects in Turkish novels. However, as suggested by Hoop and Hogeweg (2014), literary communication differs significantly from everyday communication. In novels, the narrator often acts as the speaker, but there is no clearly defined addressee. Unlike the bidirectional pattern of everyday conversations, literary communication is unidirectional, with the communicative act occurring from a fixed narrator to the reader. Moreover, the writer and the reader are typically separated in space and time, with limited shared experiences. These differences highlight the need for caution when generalizing the findings of this study to other forms of communication in Turkish.

73Despite these limitations, this study makes important contributions to our understanding of Turkish pronominal subjects in a literary context. By demonstrating how discourse factors such as recency, givenness, and syntactic prominence interact with singular pronominal subjects, it sheds light on the nuanced ways pronouns function in written narratives. These findings could inform future research on discourse-pragmatic features of pronouns in Turkish, as well as studies on pronominal use in other null-subject languages like Turkish. Moreover, the study offers a framework for examining how linguistic and grammatical features adapt to different communicative contexts, particularly in literary works. This framework could inspire further cross-linguistic and interdisciplinary investigations into the intersection of grammar, discourse, and narrative structure.

74Although the findings reported in this study are specific to Turkish novels, they emphasize the importance of considering the context-dependent nature of pronominal use. Understanding how pronouns operate in literature not only enriches our knowledge of Turkish but also contributes to the wider field of linguistics by highlighting the intricate relationship between grammar, discourse, and communication.

Haut de page

Bibliographie

Example sources

Açar, M. 2005. Siyah Hatıralar Denizi. Istanbul: İletişim Yayınları.

Altan, A. 2002. Aldatmak. Istanbul: Can Yayınları.

Aysu, O. 2011. Kanlı Pazar. Istanbul: Alfa Yayınları.

Baykurt, F. 1973. Tırpan. Istanbul: Remzi Kitabevi.

Celal, M. 2001. Bir Kadının Hayatı. Istanbul: Anka Yayınları.

Demirkıran, E. 2007. Sen Şimdi Gidecen ya Cehennemin Dibine Git: Ben Bağrıma Taş Basarım. Istanbul: Kashna Kitap Ağacı.

Eroğlu, M. 2006. Belleğin Kış Uykusu. Istanbul: Agora Kitaplığı.

İzgören, A.Ş. 2015. Kara Oklar Çetesi 1: Büyük Macera. Ankara: Elma Yayınevi.

Keskin, H. 2015. Barikat. Ankara: Sage Yayınları.

Kulin, A. 1997. Adı: Aylin. Istanbul: Remzi Kitabevi.

Kulin, A. 1999. Sevdalinka. Istanbul: Remzi Kitabevi.

Özdeş, O. 1975. Vatan Borcu. Istanbul: Tekin Yayınevi.

Studies

Ariel, M. 1988. Referring and Accessibility. Journal of Linguistics 24 (1): 65–87.

Ariel, M. 1990. Accessing Noun-Phrase Antecedents. London: Routledge.

Ariel, M. 2001. Accessibility Theory: An Overview. In T. Sanders, J. Schilperoord and W. Spooren (ed.), Text Representation. Linguistic and Psycholinguistic Aspects. Amsterdam – Philadelphia: J. Benjamins: 29–87.

Arnold, J.E. 2010. How Speakers Refer: The Role of Accessibility. Language and Linguistics Compass 4 (4): 187–203.

Baumann, S. and Riester, A. 2012. Referential and Lexical Givenness: Semantic, Prosodic and Cognitive Aspects. In G. Elordieta and P. Prieto (ed.), Prosody and Meaning. Berlin – Boston: De Gruyter: 119–161.

Cameron, R. 1997. Accessibility Theory in a Variable Syntax of Spanish. Journal of Pragmatics 28 (1): 29–67.

Carminati, M.N. 2002. The Processing of Italian Subject Pronouns. Ph.D. dissertation. Department of Lingusitics, University of Massachusetts Amherst.

Crawley, R.A., Stevenson, R.J. and Kleinman, D. 1990. The Use of Heuristic Strategies in the Interpretation of Pronouns. Journal of Psycholinguistic Research 19: 245–264.

Duffy, S.A. and Rayner, K. 1990. Eye Movements and Anaphor Resolution: Effects of Antecedent Typicality and Distance. Language and Speech 33 (2): 103–119.

Enç, M. 1985. Temporal Interpretation. Unpublished manuscript.

Enç, M. 1986. Topic Switching and Pronominal Subjects in Turkish. In D.I. Slobin and K. Zimmer (ed.), Studies in Turkish Linguistics. Amsterdam – Philadelphia: J. Benjamins: 195–208.

Field, A. 2013. Discovering Statistics Using IBM SPSS Statistics. Los Angeles: Sage.

Filiaci, F., Sorace, A., and Carreiras, M. 2014. Anaphoric Biases of Null and Overt Subjects in Italian and Spanish: A Cross-Linguistic Comparison. Language, Cognition and Neuroscience 29 (7): 825–843.

Garrod, S.C. and Sanford, A.J. 1994. Resolving Sentences in a Discourse Context: How Discourse Representation Affects Language Understanding. In M.A. Gernsbacher (ed.), Handbook of Psycholinguistics. London: Academic Press: 675–698.

Givón, T. 1983. Topic Continuity in Discourse: A Quantitative Cross-Language Study. Amsterdam – Philadelphia: J. Benjamins.

Göksel, A. and Kerslake, C. 2005. Turkish: A Comprehensive Grammar. London: Routledge.

Gordon, P.C., Grosz, B.J. and Gilliom, L.A. 1993. Pronouns, Names, and the Centering of Attention in Discourse. Cognitive Science 17 (3): 311–347.

Gundel, J.K., Hedberg, N. and Zacharski, R. 1993. Cognitive Status and the Form of Referring Expressions in Discourse. Language 69 (2): 274–307.

Gutman, E. 2004. Third Person Null Subjects in Hebrew, Finnish and Rumanian: An Accessibility-Theoretic Account. Journal of Linguistics 40 (3): 463–490.

Hartung, F., Burke, M., Hagoort, P. and Willems, R.M. 2016. Taking Perspective: Personal Pronouns Affect Experiential Aspects of Literary Reading. PLoS ONE 11 (5).

Hoop, H. de and Hogeweg, L. 2014. The Use of Second Person Pronouns in a Literary Work. Journal of Literary Semantics 43 (2): 109–125.

Jaeger, T.F. and Wasow, T. 2005. Processing as a Source of Accessibility Effects on Variation. Annual Meeting of the Berkeley Linguistics Society 31 (1): 169–180.

Jin, J. and Liu, S. 2022. An Analysis of the Linguistic Features of Popular Chinese Online Fantasy Novels. Discourse Processes 59 (4): 326–344.

Kılıçaslan, Y., Güner, E.S. and Yıldırım, S. 2009. Learning-Based Pronoun Resolution for Turkish with a Comparative Evaluation. Computer Speech & Language 23 (3): 311–331.

Kronrod, A. and Engel, O. 2001. Accessibility Theory and Referring Expressions in Newspaper Headlines. Journal of Pragmatics 33 (5): 683–699.

Küçük, D. and Yöndem, M.T. 2007. Automatic Identification of Pronominal Anaphora in Turkish Texts. In E.G. Schmidt, İ. Ulusoy, N. Çiçekli, U. Halıcı and M. V. Atalay (ed.), 22nd International Symposium on Computer and Information Sciences (ISCIS 2007). Ankara: METU: 181-186.

Lehecka, T. 2023. Antecedents of a Novel Gender-Neutral Pronoun: A Corpus-Based Approach to Pronoun Reference in Swedish. Corpora 18 (3): 329–356.

O’brien, E.J. 1987. Antecedent Search Processes and the Structure of Text. Journal of Experimental Psychology: Learning, Memory, and Cognition 13 (2): 278–290.

Öztürk, B. 2001. Turkish as a Non-Pro-Drop Language. In E. E. Taylan (ed.), The Verb in Turkish. Amsterdam – Philadelphia: J. Benjamins: 239-259.

Portele, Y. and Bader, M. 2016. Accessibility and Referential Choice: Personal Pronouns and D-Pronouns in Written German. Discours 18.

Prince, E.F. 1981. Toward a Taxonomy of Given-New Information. In P. Cole (ed.), Radical Pragmatics. London: Academic Press: 223–255.

StATS: What is a Kappa coefficient? (Cohen’s Kappa). N.d. Available online: http://www.pmean.com/definitions/kappa.htm.

Haut de page

Notes

1 This article is based on the MA thesis of the first author. The data that support the findings of this study are available upon request.

Haut de page

Pour citer cet article

Référence électronique

Buse Şen Erdoğan et Emine Yarar, « A Textual Study of Turkish Overt and Null Singular Personal Pronouns in Novels »Discours [En ligne], 36 | 2025, mis en ligne le 04 juillet 2025, consulté le 20 janvier 2026. URL : http://journals.openedition.org/discours/13761 ; DOI : https://doi.org/10.4000/14a62

Haut de page

Auteurs

Buse Şen Erdoğan

Hacettepe University, Department of Linguistics, Ankara, Turkey

Emine Yarar

Hacettepe University, Department of Linguistics, Ankara, Turkey

Haut de page

Droits d’auteur

CC-BY-4.0

Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search