Are Turkish Non-Case-Marked Objects with and without bir Interpreted and Acquired Differently?

Abstract

Whether non-case-marked objects with and without the numeral bir in Turkish have the same structural properties or not has been a topic of discussion for decades. This study aims to contribute to this discussion with experimental data that compares the comprehension of these object types along with their accusative-marked indefinite counterparts in terms of their scope with respect to negation by four-, five-, and six-year-old children as well as adults. The results suggest that both non-case-marked objects with and without bir contrast with accusative-marked indefinite objects and have a narrow scope with respect to negation in adults’ speech. However, bir can still have a main effect on the interpretation of the objects, just like the accusative case, and unlike non-case-marked objects without bir, objects with bir may scope over negation. Children treat all object types alike at age four and distinguish objects with and without bir at age five. These findings are compatible with an account that assumes different structures for two types of non-case-marked objects.

1. Introduction

Modern Turkish has two types of non-case-marked objects: bare objects (Bare N) and objects with the numeral bir ‘a/one’ (bir N), as illustrated in (1) and (2), respectively. They contrast with objects that bear the accusative case morphology (3–4). Whether these two kinds of non-case-marked objects have the same structural properties or not has been a topic of interest for a couple of decades (see, for example, Erguvanlı 1984; Aydemir 2004; Öztürk 2005, among others).

(1)	Hayriye kitap oku-yor.	(bare N)
	Hayriye book read-prog1
	‘Hayriye is book reading.’
(2)	Hayriye bir kitap oku-yor.	(bir N)
	Hayriye a book read-prog
	‘Hayriye is reading a book.’
(3)	Hayriye bir kitab-ı oku-yor.	(bir N-acc)
	Hayriye a book-acc read-prog
	‘Hayriye is reading a (particular) book.’
(4)	Hayriye kitab-ı oku-yor.	(N-acc)
	Hayriye book-acc read-prog
	‘Hayriye is reading the book.’

In this study, the comprehension of these non-case-marked object types by four-, five-, and six-year-old children and adults was tested through a truth value judgment task (Crain and Thornton 1999) that focused on their scope with respect to negation. Their comprehension was contrasted with the comprehension of accusative-marked indefinite objects (3) in the same context. The study contributes to the discussion of two non-case-marked object types with experimental data from both children and adult speakers of Turkish by showing that when tested in a context that imposes a wide scope reading of the object with respect to negation, bir N objects can have scope over negation to some extent and behave like accusative-marked objects, while bare objects are not affected.

Before the details of the experiment (Section 4) and the presentation and discussion of the results (Section 5 and Section 6), the following sections (Section 2 and Section 3) lay out the properties of the two non-case-marked objects in Turkish and how they are analyzed in the literature.

2. Bare N vs. bir N Objects

Bare N objects are those objects that do not have a case morphology or the numeral bir. They have been analyzed as object-incorporations (Mithun 1984; Nilsson 1985, 1986; Kornfilt 1997, 2003; Schroeder 1999; Aydemir 2004, among others), compounds (Swift 1963; Knecht 1986), complex predicates (Ketrez 2003, 2005, 2015), pseudo-incorporation (Öztürk 2005; Kamali 2015; Jo and Palaz 2018; Driemel 2023), non-definite (Dede 1986) or non-referential (Erguvanlı 1984) in the sense that the nominal lacks the definiteness or referentiality features. They are contrasted with not only the overtly case-marked direct objects but also non-case-marked bir N objects. The difference between overtly case-marked versus non-case-marked objects seems to be widely accepted, although there have been different proposals to account for their distinction (see, for example, Dede 1986; Nilsson 1985; Enç 1991; Taylan and Zimmer 1994; Zidani-Eroğlu 1997; Kelepir 2001; von Heusinger 2002; von Heusinger and Bamyacı 2016; von Heusinger and Kornfilt 2005; Kılıçaslan 2006; Özge 2011, among others). Whether there is a structural difference between the two non-case-marked objects has not been clear. While some accounts propose that two types of non-case-marked objects have different structures and different relations to the verb (e.g., Aydemir 2004; Kamali 2015), some attribute the same structure and interpretation to both, based on their shared case- and scope-taking properties (e.g., Öztürk 2005).

The evidence for the difference between the two object types comes from a number of properties, such as adjacency, modification by adjectives and relative clauses, and their pronominalization behavior. The argument for their similarity mostly relies on their scope with respect to the other constituents and the absence of overt case morphology. Both the series of analyses that focus on their differences and those proposals based on their similarities run into problems due to counterexamples and conflicting empirical evidence. Therefore, the issue concerning the structure of bare N vs. bir N objects remains unresolved.

It has been assumed that non-case-marked direct objects have restrictions in terms of their distance to the verbs, in contrast to the overtly case-marked objects (e.g., Erguvanlı 1984). While overt case morphology provides flexibility of movement, both bare N and bir N objects have more restrictions, and their flexibility is more discourse-dependent. Recent research revealed that movement of bare objects is possible when certain discourse conditions are met (see, for example, Uygun 2006; Gürer 2020; Göksel 2013; Öztürk 2009; İşsever 2006; Gračanin-Yüksek and İşsever 2011, among others). Examples in (5) and (6) show that bare N objects can move to the sentence-initial position for the purpose of topicalization and also to the postverbal position to establish a backgrounding reading of the object. In such examples, bare N objects have a generic reading.

(5)	Kahve Ali de iste-miş-ti.	(Uygun 2006)
	coffee Ali also want-perf-past
	‘Ali too wanted coffee.’
	(The glosses of example (5) have been somewhat changed from the original.)
(6)	Gör-dü-n mü hayat-ın-da hiç film?	(Göksel 2013)
	see-past-2SG Q life-2SG.poss-loc ever film
	‘Have you ever seen a film/films in your life?’
	(The glosses of example (6) have been somewhat changed from the original.)

This is where the two non-case-marked object types behave differently. Kamali (2015) observes that while both non-case-marked objects are possible in the right periphery of the sentence final position, only the bare N objects can occur in sentence initial position. As exemplified in (7) and (8), taken from Kamali (2015), bir N objects are not acceptable or at least less acceptable in the sentence initial position. The examples show that bare N objects are more flexible in terms of their movement, while there are restrictions on bir N objects. Kamali (2015) attributes the flexibility of the bare N objects to their generic reading and the restrictions that indefinites have on their mobility2. When the objects have bir they have an indefinite interpretation, which is lacking in the bare N objects.

(7)	A:	Bir aslan-ın boy-u ne kadardır acaba?
		one lion-gen height-poss.3SG how.tall I.wonder
		‘I wonder how tall a lion is.’
	B: ??	Bir aslani ben ti gör-dü-m. 2 metre var.
		one lion I see-past-1SG 2 meter exists
		‘A lion, I’ve seen one. It’s about 2 meters.’
(8)	A:	Aslan-lar-ın boy-u ne kadardır acaba?
		lion-plu-gen height-poss.3 how.tall I.wonder
		‘I wonder how tall lions are.’
	B:	Aslani ben ti gör-dü-m. 2 metre var.
		lion I see-past-1SG 2 meter exists
		‘Lions, I’ve seen some. They’re about 2 meters.’

Another difference between the two non-case-marked object types is observed in their modification pattern. Modification of bare N objects by an object relative clause is not possible, as seen in (9), and modification by a subject relative clause is possible only when the object is animate (10–11). These restrictions apply only to the bare N objects.

(9)	Ayşe [[herkes-in sev-eceğ-i] *(bir) kitap/*(bir) öğretmen] ar(a)-ıyor.
	Ayşe [[everybody-gen like-rel-poss] a book/teacher] look.for-prog
	‘Ayşe is looking for *(a) book/teacher that everybody will like.
(10)	Ayşe [İngilizce bil-en] sekreter/bir sekreter ar(a)-ıyor.
	Ayşe [English speak-rel] secretary/a secretary looking for
	‘Ayşe is looking for (a) secretary who can speak English.’
(11)	Ayşe [[başka ülke-ler-i anlat-an] *kitap/bir kitap] oku-du.
	Ayşe [[other country-plu-acc tell-rel] book/a book] read-past
	‘Ayşe read a book that are about other countries.’

In such examples, where both object types are possible, there is a difference between their interpretations. When the object has bir, the speaker is looking for a single secretary (10); when the object is a bare object, the number of secretaries is not specified. This is compatible with the analysis of bir as a referentiality marker (Schroeder 1999). Nilsson (1985), too, argues that bir establishes a greater independence of the noun vis-à-vis the predicate (Nilsson 1985, p. 47). Bir N nominals and accusative-marked objects contrast with bare N objects, which undergo incorporation in her analysis. In a more recent study, Aydemir (2004) proposes a similar account where bir N nominals are arguments and bare Ns are incorporated into verbs.

Another difference between the two non-case-marked objects is observed in the modification by simple adjectives. When a bare N object is preceded by an adjective, which can have an adverb interpretation, the adverbial reading is preferred, where the adjective/adverb modifies the event (bare object + verb complex) rather than the object alone. Therefore, in (12), the preferred interpretation is that ‘Ayşe is driving fast.’ The other interpretation, ‘Ayşe is driving fast cars’ is also possible, but the adverb interpretation is more salient.

(12)	Ayşe hızlı araba kullan-ıyor.
	Ayşe fast car drive-prog
	‘Ayşe is driving fast.’
	‘Ayşe is driving a fast car.’

When the adjective cannot have an adverb reading, as is the case with yeşil ‘green’ in (13), the only interpretation available is the adjective interpretation, where the adjective modifies the bare N object.

(13)	Ayşe yeşil araba kullan-ıyor.
	Ayşe green car drive-prog
	‘Ayşe is driving a green car.’

When the object has bir, either before or after the adjective, we do not have an adverbial reading, and hızlı modifies only the direct object (14).

(14)	Ayşe hızlı bir araba/bir araba-yı kullan-ıyor.
	Ayşe fast a car/a car-acc drive-prog
	‘*Ayşe is driving fast.’
	‘Ayşe is driving a fast car.’

The possibility of the adverb reading suggests that bare N objects and verbs can behave as one single unit. The possibility of the adjective reading suggests that the noun can be treated as one single word, independent of the verb. In the case of the bir N objects, only the adjective interpretation is possible, so the two object types contrast in terms of their modification behavior.

Restriction in pronominalization and object drop is another property that shows that bare N objects and bir N objects are different (Erguvanlı 1984; Aydemir 2004; Ketrez 2005; Arslan-Kechriotis 2009). Only objects that are referential and that have a DP projection can be pronominalized and undergo pro-drop. The ungrammaticality of bare N object pronominalization and object drop is shown in (15). The same structure is grammatical with bir, as seen in (16).

(15)	*Dün filmi seyret-ti-m, on-ui/on-lar-ıi sen de seyret-meli-sin. (Aydemir 2004)
	yesterday film watch-past-1SG that-acc that-pl-acc you too watch-nec-2SG
	‘I watched movies/did movie watching yesterday, you should watch them too.’
	(The glosses of example (15) have been somewhat changed from the original.)
(16)	Dün bir filmi seyret-ti-m, on-ui sen de seyret-meli-sin. (Aydemir 2004)
	yesterday a film watch-past-lSG that-acc you too watch-nec-2SG
	‘I watched a movie yesterday, you should watch it too.’
	(The glosses of example (16) have been somewhat changed from the original.)

Another example is seen in (17) and (18). When the first sentence has a bare N object, what is recommended is interpreted as the action (‘book-reading’ versus ‘swimming’). When the object has the numeral bir, what is recommended is the object, namely the book. In other words, when the object is in its bare form, it behaves like a single unit with the verb. When the object has a numeral, it behaves as an independent unit.

(17)	Ayşe tatil-de [bir kitap]i oku-du, on-ui herkes-e tavsiye et-ti. (Ketrez 2005)
	Ayşe break-loc a book read-past, it-acc everybody-dat recommend-past
	‘Ayşe read a book during the break and recommended it (book) to everybody.’
(18)	Ayşe tatil-de [kitap oku]i-du, on-ui herkes-e tavsiye et-ti. (Ketrez 2005)
	Ayşe break-loc book read-past, it-acc everybody-dat recommend-past
	‘Ayşe read a book during the break and recommended it (book reading) to everybody.’

These judgments, however, were questioned (e.g., Bliss (2004)) and reinvestigated more recently with a quantitative study by Seidel (2019). It was found that bare objects can be taken up by overt pronouns and also pro in experimental settings. Özge (2011, p. 116) also notes that bare objects can behave like bir N objects in examples such as (19).

(19)	Eğer bir gün çocuki yap-ar-sa-m,	on-ui kesinlikle kolej-e yolla-ma-yacağ-ım.
	if one day child make-aor-cond-1SG s/he-acc definitely college-dat send-neg-fut-1SG
	‘If I have (a) childi one day, I will definitely not send him/heri to college.’

In summary, two types of non-case-marked objects were contrasted with the accusative-marked objects and assumed to have a unified behavior due to their common property, which is the lack of overt case morphology. Recent research challenges not only the accounts that rely on their similarities but also the ones that assume that they have different behaviors.

3. Scope of Bare N and bir N Objects

Öztürk (2005) shows that both bare objects (20) and bir N objects (21) take narrow scope in contrast to (accusative-marked) indefinite objects (22) that can take both narrow and wide scope. Based on these examples, it was concluded that only the accusative-marked indefinites are “real” indefinites in Turkish, while the non-case-marked ones appear in complex predicates and cannot interact with other quantifiers.

(20)	Her çocuk kitap oku-du.	every > kitap; *kitap > every
	every child book read-past
	‘Every child book-read.’
(21)	Her çocuk bir kitap oku-du.	every > bir kitap; *bir kitap > every
	every child a book read-past
	‘Every child read a book.’
(22)	Her çocuk bir kitab-ı oku-du.	every > bir kitab-ı; bir kitab-ı > every
	every child a book-acc read-past
	‘Every child read a book.’

As seen in examples (23–24), both bir N objects and bare N objects take narrow scope with respect to negation as well. Non-case-marked objects contrast with accusatively marked indefinite objects (25) where the wide-scope reading of the object (‘there is an apple that is not eaten’) is possible.

(23)	Ayşe elma ye-me-di.	neg > elma; *elma > neg
	Ayşe apple eat-neg-past
	‘Ayşe did not eat an apple.’
(24)	Ayşe bir elma ye-me-di.	neg > bir elma; *bir elma > neg
	Ayşe a apple eat-neg-past
	‘Ayşe did not eat an apple.’
(25)	Ayşe bir elma-yı ye-me-di.	bir elma-yı > neg; neg > bir elma-yı
	Ayşe a apple-acc eat-neg-past
	‘Ayşe did not eat an apple.’
	(or there is an apple such that Ayşe did not eat it).

The examples above show that the two non-case-marked objects behave similarly in terms of their scope-taking properties and both contrast with overtly case-marked indefinites. However, the following examples from Dede (1986) show that direct objects can have a wide scope reading even in the absence of overt case morphology (26)3. Note that bare objects cannot have this reading in the same context (27). Therefore, in terms of scope-taking properties as well, two non-case-marked objects can behave differently.

(26)	Bir kitap arı-yor-um.	Bul-a-mı-yor-um.
	a book look.for-prog-1SG	find-abil-neg-prog-1SG
	i.‘I’m looking for a book. I cannot find find one.’
	ii. ‘I’m looking for a book. I cannot find it.’
(27)	Kitap arı-yor-um.	Bul-a-mı-yor-um.
	book look.for-prog-1SG	find-abil-neg-prog-1SG
	i.‘I’m looking for a book. I cannot find find one.’
	ii. ‘*I’m looking for a book. I cannot find it.’

Ketrez (2015) tested the scope of accusative-marked and non-case-marked objects with respect to negation in a comprehension experiment. Adults as well as four- to six-year-old children participated in the experiments. The results suggested that adults assigned wide scope to accusative-marked indefinite objects, and children acquired the relationship between scope and case morphology gradually. Four-year-old children treated case-marked and non-case-marked objects alike. Although the accusative case is one of the earliest acquisitions in Turkish children’s speech, emerging before two years of age in early productions (Ketrez and Aksu-Koç 2009), its comprehension in complex structures, such as the ones where it interacts with the scope of negation, is mastered later.

Based on the same set of data analyzed in Ketrez (2015), the present study focuses on the interpretation of the bare N versus the bir N objects and contrasts them with each other, as well as with the accusative-marked indefinite objects. The scope of objects with respect to negation is one of those areas where non-case-marked objects are predicted to be the same. A unified behavior of non-case-marked objects is expected, especially in (pseudo-)incorporation accounts where objects without case morphology are dependent on the verbs, although they do not necessarily have movement and modification restrictions (e.g., Öztürk 2005; Driemel 2023). On the other hand, as the review and examples above show, bir N objects, although they do not have case morphology, may receive a specific interpretation. The goal of the present study is to provide examples of children’s and adults’ treatment of two types of non-case-marked objects in experimental settings.

4. Method

4.1. Participants

The participants included 49 four-year-old (age range: 3;0–4;5), 51 five-year-old (age range: 4;6–5;6), 48 six-year-old (age range: 5;7–6;6) children, and 139 university student adults. They were all monolingual native speakers of Turkish and belonged to middle- or upper-middle-class families living in two big metropolitan cities (İstanbul and İzmit) in the northwest of Turkey. The children were drawn from four different kindergartens and had normal language development, as reported by their teachers. Upon their completion of the test, they were awarded a color sticker. Adult participants were undergraduate students from various majors at Boğaziçi University in İstanbul and either participated voluntarily upon an announcement in a class they were enrolled in or were recruited through a subject pool and received an extra credit in a course. Informed consent forms were signed by the parents of the children and the adult participants prior to the test. The participants were divided into four age groups, as shown in

Table 1. Participants (mean age, age-range and number).

Object Type	4-Year-Olds	5-Year-Olds	6-Year-Olds	Adults
N V-neg	3;8 (3;5–4;3) n = 9	5;0 (4;8–5;4) n = 10	6;0 (5;7–6;4) n = 10	N.A. n = 25
bir N V-neg	4;0 (3;0–4;4) n = 10	5;0 (4;8–5;6) n = 11	6;1 (5;8–6;6) n = 8	N.A. n = 32
bir N-acc V-neg	4;0 (3;3–4;4) n = 10	5;0 (4;7–5;4) n = 9	6;1 (5;7–6;5) n = 11	N.A. n = 32
bir N-acc XP V-neg	4;2 (3;9–4;5) n = 10	5;0 (4;6–5;5) n = 10	6;2 (5;10–6;5) n = 10	N.A. n = 24
V-neg bir N-acc	3;9 (3;5–4;4) n = 10	5;1 (4;6–5;6) n = 11	6;0 (5;10–6;6) n = 10	N.A. n = 25

. Each object type was tested on a different group of participants to avoid a possible bias resulting from the contrast between object types.

4.2. Procedure

A truth-value judgment task was conducted by two experimenters in a quiet room in kindergartens (Crain and Thornton 1999). One of the experimenters acted out the scenes with small toys on a table in front of the children and told their stories. See (28) for an example scenario. See Appendix A for the test material. The other experimenter used a hand puppet that was introduced to the children prior to the game. The puppet listened to the story told by the experimenter and then commented on what happened at the end of the story. The voice of the puppet was played on a computer to control or maintain the proper intonation of the sentences. If the puppet understood what happened at the end of the story, that is, if it stated what happened at the end correctly, the child rewarded the puppet with candy. When the puppet’s statement was judged as “wrong” by the child, the experimenter asked the child what really happened. The task of the child in this game was to determine whether or not the puppet gave the correct answer and, when it was wrong, to correct the answer. With adult participants, the same setting was used without a puppet. The experimenter played the test sentences as a conclusion to the stories, and the participants marked their responses on a response sheet, where they had the names of the scenarios (e.g., “the goat story”).

Prior to the test, participants heard three training sentences accompanied by three scenarios. The aim of the training stories was to teach the participants the rules of the game; therefore, they were simpler than the actual test sentences. During the experiment, after the training session, the participants heard six test sentences and scenarios (all the same object type) and five filler/control sentences and scenarios in a randomized order. The test items and the filler/control items had similar stories, but they contained intransitive verbs or transitive verbs with a definite object. The aim of the filler/control items was to control any response biases participants might have, so they had a mixture of clear reject or accept responses. Each participant received 14 scenarios in total (3 training, 6 test sentences, and 5 filler/control sentences), and the entire session lasted about 15–20 min. Those children who failed in two or more training sentences were not allowed to continue the test. Those children who failed in more than one filler sentence and those who gave irrelevant responses to the follow-up questions (what really happened) were excluded as well.

4.3. Test Items

The test sentences were accompanied by a wide scope context in which there were two objects (e.g., two flowers), and the agent’s task was to act upon both objects (e.g., to eat both flowers). Due to some unexpected problem (e.g., he was not tall enough), one of the objects was not acted upon. Test sentences that were used along with the scenario in (28) were presented in (29). In this condition, the wide scope reading of the object with respect to negation was forced so that we could see whether non-case-marked objects can have a wide scope interpretation.

(28)

The hungry goat story: Karnı aç olan keçi bahçede dolaşıp yiyecek arıyormuş. Bakmış, otların arasında iki çiçek görmüş. Ama keçi çiçek yemeyi hiç sevmiyormuş. Çiçeklerin yanından geçmiş, etrafa bakmış, yiyecek başka bir şey bulmaya çalışmış ama başka hiçbir şey yokmuş. O da aç kalmaktansa en iyisi çiçekleri yiyeyim demiş. Gidip yakındaki çiçeği afiyetle yemiş. Karnı hala çok açmış. Biraz yukarıda bir yerde bir çiçek daha varmış. Ama o çiçeğe bir türlü boyu yetişmemiş. O da onu bırakmış eve dönmüş. O gittikten sonra bir at gelmiş, kalan çiçeği de o at yemiş. Atın boyu uzun olduğu için yukarıdaki çiçeklere bile yetişebiliyormuş. ‘The goat, which was still hungry, was searching the garden to find something to eat. He saw two flowers among the grass. But he didn’t like flowers. So he kept looking for some other things to eat. But he could not find anything. So he had to eat the flowers. He ate one of the flowers. He was still so hungry. So he wanted to eat the other one too. But the other flower was somewhere high. He tried to reach it, but no matter how hard he tried he could not get it. In the end he gave up and returned home. A horse that was passing by found the flower after the goat left and ate it. He could reach the flower because he was a very tall horse.’

Along with two non-case-marked objects with and without bir, accusative-marked indefinite objects were included in the experiments to see the scope contrast. In the sentences that included accusative-marked objects, objects occurred in three different surface positions. Each of these sentences, along with five other sentences that had the same type of object, was presented to a different group of participants.

(29)	Test sentences
	PUPPET:	Keçi çiçek ye-me-di.	bare N V-neg
		goat flower eat-neg-past
		‘The goat did not flower-eat.’
	PUPPET:	Keçi bir çiçek ye-me-di.	bir N V-neg
		goat a flower eat-neg-past
		‘The goat did not eat a flower.’
	PUPPET:	Keçi bir çiçeğ-i ye-me-di.	bir N-acc V-neg
		goat a flower-acc eat-neg-past
		‘The goat did not eat a flower.’
	PUPPET:	Bir çiçeğ-i keçi ye-me-di.	bir N-acc XP V-neg
		a flower-acc goat eat-neg-past
		‘The goat did not eat a flower.’
	PUPPET:	Keçi ye-me-di bir çiçeğ-i.	V-neg bir N-acc
		goat eat-neg-past a flower-acc
		‘The goat did not eat a flower.’
	CHILD:	Accept (wide scope reading)/Reject (narrow scope reading)

4.4. Scoring

Each wide scope reading of the test sentence was recorded as one point. The maximum wide scope reading recorded for each participant was six, so each subject had a score of wide scope reading calculated out of six. The mean scores of each object type were compared within age groups. In addition, the participants were examined in terms of their response patterns. Those participants who accepted the readings in five or six items were grouped as accept (wide scope) pattern; those who accepted only one, two or no scenarios were grouped as reject patterns. The ones who rated three, or four items as acceptable were then the mixed-pattern participants.

5. Results

Figure 1 below shows the comprehension of five object types by four different age groups. As seen in Figure 1 and also in

Table 2. Mean wide scope (accept) scores in four age groups and five object types.

Object Type	4-Year Olds	5-Year Olds	6-Year Olds	Adults
Bare N V-neg	0.56 (9%, SE = 0.242)	0.30 (5%, SE = 0.213)	0.60 (10%, SE = 0.40)	0.24 (4%, SE = 0.119)
bir N V-neg	1.20 (20%, SE = 0.80)	2.09 (34%, SE = 0.63)	2.00 (33%, SE = 0.91)	2.04 (34%, SE = 0.413)
bir N-acc V-neg	1.10 (18%, SE = 0.586)	3.11 (52%, SE = 0.754)	3.45 (57%, SE = 0.666)	5.82 (97%, SE = 0.092)
bir N-acc XP V-neg	1.50 (25%, SE = 0.637)	3.10 (52%, SE = 0.862)	4.80 (80%, SE = 0.291)	5.83 (97%, SE = 0.078)
V-neg bir N-acc	2.00 (33%, SE = 0.577)	2.73 (41%, SE = 0.752)	3.90 (65%, SE = 0.640)	5.60 (93%, SE = 0.208)

, the adult participants assigned wide scope to accusative-marked indefinite objects in more than 90% of the instances, regardless of the position of the object with respect to the verb (negation) or their adjacency condition. Adults rejected the wide-scope reading of the bare objects. The non-case-marked objects that have the numeral bir patterned in between. Their wide-scope interpretation is not close to the accusative-marked counterparts (90–97% vs. 34%). Although their wide-scope interpretation was actually significantly below the chance level (t(27) = −2.335, p < 0.05), which could be interpreted as a narrow scope reading, it was obvious that they did not pattern like bare objects either. A factorial ANOVA was conducted on the comprehension scores of adults to see what determined the scope assignment of the objects. Overt case morphology, the numeral bir, word order (preverbal vs. postverbal), and adjacency to the verb were entered into the analysis as possible factors. The results indicated that case morphology had a main effect (F(1, 131) = 200.8, p < 0.001) on the interpretation of indefinite objects, i.e., adults treated the case-marked vs. non-case-marked objects differently. The numeral bir had a main effect (F(1, 131) = 31.741, p < 0.001), as well, i.e., bir N(-acc) objects and bare N objects were treated differently by adults. Word order (preverbal and postverbal occurrence of the object) and adjacency to the verb did not have any main effects (F(1, 131) = 0.653, n.s., F(1, 131) = 0.002, n.s.).

The response patterns of the participants confirmed the results. While the accusative-marked objects and bare objects had relatively distinct accept or reject patterns, participants in the bir N group had more variable behavior. 94% (n = 31, bir N-acc V-neg), 96% (n = 24, V-neg bir N-acc) and 100% (n = 24, bir N XP V-neg) of the adult participants accepted the wide scope reading of the accusative-marked objects. No participants (n = 25) accepted the wide scope reading of the bare N objects. 92% of them (n = 23) rejected it in all their responses, while 8% (n = 2) had a mixed pattern. In the case of the bir N objects, 58% (n = 21) of the participants rejected the wide scope reading, 21% of the participants (n = 6) accepted it, and 21% of them (n = 6) had a mixed pattern.

The majority of the four-year-old children rejected the wide scope reading of all the objects and, more importantly, treated all the object types alike (F(4, 44) = 0.745, p = 0.567) regardless of their case, adjacency, word order properties, and the numeral bir. This is revealed in their response patterns as well. In contrast to the adult participants, 10% (n = 1, bir N-acc V-neg), 10% (n = 1, V-neg bir N-acc), and 20% (n = 2, bir N XP V-neg) of the four-year-old participants accepted the wide scope reading of the accusative-marked objects. These object types were rejected by 80% (n = 8), 40% (n = 4), and 60% (n = 6) of the participants, respectively, and the rest had a mixed pattern. Similar to the adult participants, none of them (n = 10) accepted the wide scope reading of the bare N objects. In the case of the bir N objects, 80% (n = 8) of the participants rejected the wide scope reading, 20% of the participants (n = 2) accepted it, and no participant had a mixed pattern.

The differentiation of the objects was observed around age five. The five-year-olds had a wide scope of the accusative-marked indefinite objects in 41–52% of the cases, while their bir N objects had a wide scope over negation in 34% of the cases (the same as the adult participants’). Their bare objects had the wide scope assignment in around 5% of the instances (4% in adult speech). A factorial ANOVA conducted on five-year-old participants’ scores suggested that case morphology did not yet have a main effect (F(1, 47) = 0.297, n.s.), and bir had an effect trend F(1, 47) = 3.627, p = 0.06). Therefore, the first differentiation of object types was observed between the bare N objects and the objects that had bir around age five. In five-year-old children, bir N objects patterned similarly to the bir N-acc objects, although they were not case-marked overtly. It is important to note, however, that although bir N-acc and bir N objects patterned similarly, neither of them had a wide scope above the chance level. Therefore, the objects that have bir and bare N objects were differentiated, but bir did not result in an early wide scope assignment. Rather, the five-year-old participants had random behavior in their reactions to bir N and bir N-acc objects, regardless of their case morphology. In the case of the bare N objects, in contrast, wide scope reading of objects was rejected. Therefore, the five-year-olds differentiated bare objects, although this differentiation is a very weak one.

The response patterns of the participants confirmed the results. 33% (n = 3, bir N-acc V-neg), %36 (n = 4, V-neg bir N-acc) and 50% (n = 5, bir N XP V-neg) of the five-year-olds accepted the wide scope reading of the accusative-marked objects. No participants (n = 10) accepted the wide scope reading of the bare objects. 90% of them (n = 9) rejected it in all their responses, while 10% (n = 1) had a mixed pattern. In the case of the bir N objects, 45% (n = 5) of the participants rejected the wide scope reading, 18% of the participants (n = 2) accepted it, and 36% of them (n = 4) had a mixed pattern.

In the six-year-old group, accusative-marked objects in the preposed position had 80% wide scope with respect to negation. No increase was observed in the adjacent position, which was recorded to have around 57% wide scope reading. Postverbal accusative-marked objects had a wide scope in around 65% of the cases. At this age, bir N objects did not pattern like the bir N-acc objects anymore. An increase was observed in the preverbal non-adjacent objects in the six-year-old participants’ speech, but even with this increase, the wide scope interpretation of accusative-marked objects was not adult-like. In contrast to the adult participants, six-year-olds had neither a main effect of case (F(1, 45) = 9.797, n.s.) nor bir (F(1, 45) = 2.46, n.s.) in their interpretation of the objects. Note that children’s wide scope scores for the bir N objects and bare N objects were the same as adults, but they had a different interpretation of the case-marked objects at age six. Therefore, the difference between adults and children was due to children’s different interpretations of the accusative case.

The response patterns of the six-year-old participants confirmed the results. 36% (n = 4, bir N-acc V-neg), 60% (n = 6, V-neg bir N-acc), and 50% (n = 5, bir N XP V-neg) of the adult participants accepted the wide scope reading of the accusative-marked objects. One participant accepted the wide scope reading of the bare objects. 80% of them (n = 8) rejected it in all their responses, while one participant had a mixed pattern. In the case of the bir N objects, 62% (n = 5) of the participants rejected the wide scope reading, 20% of the participants (n = 2) accepted it and 13% of them (n = 1) had a mixed pattern.

6. Discussion

The experimental results suggested that case morphology played a major role in scope assignment to direct objects in adult participants’ speech, and the child participants were in the process of acquiring the relationship between overt case morphology and scope interpretation. We observed that bare N objects predominantly had narrow scope readings for all age groups. Crucially, that was not the case for the other non-case-marked object type, bir N objects. Adults had a wide scope interpretation of bir N objects in 34% of the cases. Although there was a big enough difference in the proportion of wide scope reading in case-marked and non-case-marked objects in adults’ responses (34% vs. 90–97%) and the wide scope reading of bir N objects is lower than chance (50%), the factorial ANOVA test still detected an effect trend of bir. This is not expected in the models that treat all non-case-marked objects alike, or in those accounts that make a distinction between case-marked and non-case-marked objects as two categories only. The experimental results here rather suggest that non-case-marked objects are further divided into two categories based on their scope-taking properties. These results are in line with the proposal in Kamali (2015), for example, which suggests that only the bare objects are pseudo-incorporated in Turkish, and instead of a two-way distinction (case-marked vs. non-case-marked), there is a three-way distinction of direct objects.

The experimental results of von Heusinger et al. (2019), based on a grammaticality judgment task on case-marked and non-case-marked partitive constructions, showed that the accusative-marked objects were ranked equally acceptable in specific and non-specific contexts. Unmarked direct objects, in contrast, were rated clearly more acceptable in non-specific contexts. Based on their findings, they conclude that the lack of case-marking is incompatible with a specific reading, whereas case-marked objects can be interpreted as specific or non-specific indefinites. The difference between the adult results of the present study and theirs can be attributed to the types of structures that were tested. They had partitive constructions with and without case-marking, and they did not test the scope with respect to negation. The procedure for data collection was also very different. In the present study, the participants responded to the audio material and had limited time to respond. The participants in the other study had the test sentences on the screen of a Google document, had more time and more options, not only the binary choice of accept or reject, and they did not have access to the intended prosody of the test sentences. They were also exposed to different types of objects and had a chance to contrast the object types, whereas the participants in this study received only one object type. Whether any of these methodological differences played any role in the differences observed between the two studies needs to be investigated further. The results on the unmarked direct objects and their compatibility with non-specific contexts do not necessarily contradict the present results. In the present study, too, the wide scope reading of non-case-marked objects was lower than that of accusative-marked objects, and they were not interpreted as specific objects in the majority of the cases by the majority of the participants. The contrast that was emphasized here was between the bare N objects and the non-case-marked objects with bir. The study reported by von Heusinger et al. (2019) did not include bare N objects, and the contrast they had was between case-marked versus non-case-marked objects with bir.

It is important to note that the similarities between bare N objects and bir N objects that were reported in the literature (e.g., Öztürk 2005) are mostly based on their similar scope-taking properties. Both non-case-marked objects have a narrow scope with respect to other quantifiers and negation, and in this study too, that is what was observed. Despite the difference between bare objects and bir N objects, we see that neither of them takes a wide scope with respect to negation. Their similar scope behavior, however, cannot be attributed to the same exact structure. They may both fall under the scope of negation, but this may be due to different reasons or relations to the verbs. The crucial result of this study is that, although neither had a wide scope with respect to negation, they had different degrees of scope assignment. While bir N objects could have some wide scope reading, and in five-year-old children’s speech, they even behaved like accusative-marked objects when a wide scope reading was forced, bare N objects were not affected at all; that is, they did not have a wide scope reading, unlike the objects with bir. This small but crucial difference shows that the numeral bir makes a difference in the interpretation, and the results of the present study can be compatible with an account that recognizes the contribution of bir to the interpretation.

Having said that, it is also important to acknowledge the difference between the impact of the accusative case and bir on the interpretation of the objects. Although there is a distinction made between the two types of non-case-marked objects, the difference between them is not as salient as the difference between case-marked and non-case-marked objects in adult speech. Such a trend is only observed at age five, where bir N objects, regardless of their case morphology, patterned alike and a differentiation based on bir was recorded. The availability of such a distinction at any phase of language development suggests that it is a possible grammatical distinction, although it does not necessarily have any implications for adult Turkish grammar.

It is important to note that this result cannot be attributed to the lexical properties of the numeral, that is, its cardinal reading. According to the Lexical Factor Hypothesis formulated by Su (2001), in those languages where an indefinite interpretation is expressed with a numeral, children have earlier access to the wide scope reading of the indefinites. Therefore, the Lexical Factor Hypothesis predicts that, as is the case in Mandarin, those structures where we have the numeral bir will have a wide scope reading due to its cardinal interpretation. First of all, the cardinal reading, or the ‘exactly one’ reading, was controlled by the number of objects (e.g., the number of flowers in the hungry goat story). For the reason that one out of two objects was acted upon, the cardinal reading of the object (the goat did not eat one flower) led to a rejection response as ‘the goat did not eat one flower’ was false. The goat ate exactly one flower in the story. So having a cardinal reading could not result in a higher wide scope reading of the caseless objects, even if the participants in the experiments had access to such a reading. Therefore, we cannot attribute the relatively higher wide scope reading of bir N objects to the cardinal reading of bir.

Secondly, although bir N-acc and bir N objects without case patterned similarly, neither of them had wide scope above the chance level in children’s responses. Therefore, the objects that had bir and bare N objects were differentiated, but bir did not result in an early wide scope assignment. Rather, bare objects had narrow scope, and the other object types had a random scope assignment in the middle age groups. In the earlier age group, however, they clearly lacked wide scope. These results are similar to the results reported for other languages, such as Dutch or English (Schaeffer 1997; Krämer 2000; Musolino 1998; Lidz and Musolino 2002). Children, then, around the age of five, stop assigning default scope (narrow scope) to bir N objects, but they do not assign wide scope either. Another important point is that children’s scope assignment to bir N objects is not different from adults (around 30%). Therefore, the picture shows that bir N-acc objects behave like bir N objects (which are adult-like). The cardinal number hypothesis predicts it to be the other way around, namely, bir N objects are expected to behave like bir N-acc objects having wide scope. Therefore, the trend observed in bir N objects in the five-year-old group cannot be accounted for by the Lexical Factor Hypothesis (Su 2001).

One possible explanation for the wide scope reading of the bir N objects in this particular experimental setting could be their potential infelicity in negative contexts. When speakers would like to negate an entire event and say that no object was acted upon (wide scope of negation), they typically use a bare object. So the availability of the bare object option in the language makes the use of the bir N object infelicitious and results in a wide scope assignment to the object. Interestingly, in those languages that have a bare object option for narrow scope reading with respect to negation, wide scope assignment to objects with the numeral or indefinite determiner is acquired earlier (see Su (2001) for Mandarin and Miller and Schmitt (2004) for Spanish). These languages contrast with those that do not have bare objects, where indefinites are used to express both wide and narrow scope readings (see Krämer (2000) for Dutch).

In the description of the two object types above, we have seen that the two non-case-marked object types are different in terms of their behavior in adjacency, modification, and pronominalization, but these traditional accounts were challenged not only by the more recent observations but also by empirical findings based on the experimental results of adults. It is clear that an analysis that is based on case morphology or the absence of case morphology alone is not compatible with the results that are presented here or elsewhere in other experimental studies.

7. Conclusions

This study compared and contrasted two types of non-case-marked direct objects in Turkish and provided experimental results based on adult speakers’ and children’s comprehension of these object types. The discussion and the experimental results suggested that bare N objects and objects with bir were comprehended differently by adults in the experimental settings, and they were acquired differently by children between ages four and six. While objects with bir can have some wide scope with respect to negation, similar to the accusative-marked indefinite objects, bare N objects did not interact with negation at all. This contrastive behavior is consistent with an analysis that treats these two non-case-marked object types differently (e.g., Kamali 2015). This study does not provide any detailed theoretical explanation for the way adults or children treat objects with and without bir, or objects with or without case morphology. It only presents the empirical results with the hope that they will be used to develop the relevant parts of the theory further.