AN EXPERIMENTAL APPROACH TO NEW AND OLD INFORMATION IN TURKISH LOCATIVES AND EXISTENTIALS

AN EXPERIMENTAL APPROACH TO NEW AND OLD INFORMATION IN TURKISH LOCATIVES AND EXISTENTIALS Engin ARIK 1, Pınar ÖZTOP 2, and Esen BÜYÜKSÖKMEN 1 Doguş University, 2 Plymouth University enginarik@enginarik.com pinaroztop@gmail.com esenbuyuksokmen@gmail.com 1 Introduction Locatives ('a pen is on the table'), existentials ('there is a pen on the table'), and possesives ('I have a pen on the table') are related to each other (Freeze 1992, Jackendoff 1990, Heine 1997) but differ from one another for syntactic and semantic-pragmatic reasons. Existential sentences may contain a verb related to locative to be and possessive to have. Existentials contain a pivot and a locative expression. Moreover, existential sentences obey definiteness restriction which is discussed below. On the basis of his crosslinguistic study, Freeze 1992 suggests that locative and existential be predicates do not have a [+human] feature while possessive have predicates do. Existential have is composed of a preposition syntactically incorporated into a copula (be) to account for alternations between be (existentials and locatives) and have (possessives). Most syntactic analyses consider existential there in (1) as the subject (Barwise & Cooper 1981, Keenan 1987. McNally 1992, Milsark 1974, Stowell 1978). But the syntactic status of book (pivot) and table (coda) in (1) is open to discussion. (1) There is [a book] pivot [on the table] coda There are some disagreements among the linguists on the status of codas. In some analyses, codas such as table in (1) are part of Noun Phrase (NP) headed by the head of the pivot according to Barwise & Cooper 1981, codas are Verb Phrase (VP)-adjuncts according to McNally 1992. There are yet other factors that differentiate locatives from existentials. For example, the pivot shows the definiteness effect (e.g. Zucchi 1995, Keenan 2003): Definite?'the book' instead of a book in (1) is unacceptable suggesting that existentials do not allow definite pivots. There is an exception for this case. Rando

& Napoli 1978 argue for the listing effect. For example, the definite marker, the, is still acceptable in the existential construction in (2): (2) How could we get there? Well, there is the trolley. (Rando & Napoli 1978: 300) Locatives and existentials also differ from one another for pragmatic reasons. For example, the pivot introduces a new discourse referent (see Heim 1987, Enç 1991). While (3) is acceptable, (4) is unacceptable if the pivot is coreferential: (3) There is a book i on the table. The book i is mine. (4)?The book i is mine. There is a book i on the table. In this study, we focus on pragmatic factors such as new information and old information in locatives and existentials analyzing acceptability judgments of the Turkish speakers. We use an experimental linguistic methodology, i.e. acceptability judgments and forced-choice scales, to manipulate the linguistic context in which the utterances appear. For example, we asked participants to rate an item which has either a locative sentence and an adjectival sentence or an existential sentence and an adjectival sentence. The NPs in the subject position were coindexed, for example, sentences in Turkish corresponding to A/The book is on the table. The book is thick. vs There is a/the book on the table. The book is thick. in English. 2 Turkish locatives and existentials Turkish locatives and existentials can be illustrated in (5) and (6) respectively. (5) Bardak kutu-nun iç-in-de-y-di- Ø. Glass box-gen in-poss-loc-copula-past-3sg The glass was in the box (6) Kutu-nun iç-in-de bardak var-(y)-dı-ø. Table- GEN in-poss-loc book exist-copula-past-3sg (lit.) In the box was a glass In Turkish, the phrasal structure of locatives is NP and NP+loc+(Tense)+Person,number whereas that of existentials is an optional (NP+loc) coda and NP pivot as well as var+(tense)+person,number. In locatives, locative case marked NPs are inflected for tense and person/number. In existentials, var (lit. exist, sometimes translated as there is) is inflected for tense and person/number. In locatives, the NP is the syntactic subject of the sentence and the theme argument semantically whereas, in existentials, NP pivot is in the same position. While the locative marked NP is the locative argument in both, it carries the predicate function in locatives but it is in an adjunct position in existentials.

3 Previous studies We have conducted several studies to explore locatives and existentials in Turkish. In the studies that we have used experimental linguistics methodology, i.e. statistical analysis of acceptability judgments, we have already shown that the Turkish speakers acceptability judgments differ with respect to the structure of NPs in Turkish locatives and existentials. In one study (Arik 2013), we questioned whether the internal structure of NPs in the theme position has an effect on distinguishing locatives from existentials. For this, a 7 point Likert-type scale was created. Twelve token sets (12X6, a total of 72 items) and thirty-six fillers, similar in the number of words, were prepared. All of the sentences were in their canonical orders: Either Subject-Predicate (locative) or Adjunct-Subject-Predicate (existential) depending on the construction type. Two scripts were prepared and randomized according to two blocks. Each participant received only one script with twelve test items and twelve fillers. In the directions (appropriateness to Turkish), the sixty-four native speakers of Turkish participated in that study. They were asked to rate locative and existential sentences in Turkish. In these sentences, NPs in the theme position were modified with number: Bare noun kitap 'book' vs. number quantification iki kitap 'two books' vs. plural marked kitap-lar 'books'. A token set is given below: (7) Locative with Singular Noun Kitap masa-da. Book table-loc A/the book is on the table (8) Locative with Plural Noun Kitap-lar masa-da. Book-PL table-loc (The) books are on the table (9) Locative with Quantified Noun Iki kitap masa-da. Two book table-loc (The) two books are on the table (10) Existential with Singular Noun Masa-da kitap var. Table-LOC book exist (lit.) On the table a/the book exists (11) Existential with Plural Noun Masa-da kitap-lar var. Table-LOC book-pl exist (lit.) On the table (the) books exist

(12) Existential with Quantified Noun Masa-da iki kitap var. Table-LOC two book exist (lit.) On the table (the) two book exist The results showed that there was a significant effect of the sentence type: the ratings for the existential sentences were significantly higher than the ratings for the locative sentences. There was no main effect of quantification type. But there was a significant interaction between sentence type and quantification type. We speculated that this finding could be because of new vs. old information: Compared to locatives, existentials introduce new information therefore participants could have rated existentials higher than locatives because they came across the sentences in the scale for the first time thus interpreted them as new information. We test this speculation in the present study. 4 Method 4.1 Participants One hundred twenty-one native speakers of Turkish participated in this study. They were all first year undergraduate students taking Introduction to Psychology course at Isik University in 2011. They signed the consent forms and gained an extra credit for their participations. 4.2 Materials and Procedure A 2x3 factorial design (Construction: locatives vs. existentials x NP: bare vs. plural vs. numbered NPs) was used. We prepared scripts which had a 7-point Likert scale, ratings between 0-6, where 0 is the least appropriate and 6 is the most appropriate. There were a total of four scripts. Each participant received only one of the scripts and filled it out by using paper and pen/pencil. Directions stated that each sentence should be rated according to their appropriateness to Turkish, there was no right or wrong answer, and no name or signature needed for anonymity reasons. It took approximately 10 minutes to collect data. Each script consisted of thirty-six items/sentences: Twelve testing items (two sets randomly selected from eight token sets), twelve good fillers, and twelve bad fillers. The sentences were randomly ordered. We give two examples from a token set below: Locative, Bare noun: (13) Kitap masa-nın üst-ün-de-ø. Kitap kalın-ø. Book table-gen on-poss-loc-copula-3sg Book thick-copula-3sg A/The book is on the table. The book is thick.

Existential, Bare noun: (14) Masa-nın üst-ün-de kitap var-ø. Kitap kalın-ø. Table-GEN on-poss-loc book exist-copula-3sg Book thick-copula-3sg There is a/the book on the table. The book is thick. We assumed that the second sentence Kitap kalın contains old information because the first sentence (i.e. Kitap masanın üstünde in [13] Masanın üstünde kitap var in [14] have already introduced kitap. Therefore, kitap in the first sentences of (13) and (14) is new information. Note that when the theme of the first sentence was plural (iki kitap 'two books' or kitaplar 'books'), the theme of the second sentence was also plural marked (kitaplar). We hypothesized that existential sentences will receive higher ratings than locative sentences because existential sentences but not locative sentences are expected to carry new information. We conducted this experimental linguistic study to manipulate the linguistic constructions and expose the participants to the same condition. We use repeated measures ANOVA with construction type (locative, existential) and number type (singular, plural, numbered). 5 Results There was no difference among the results obtained from the scripts. Therefore, there was no effect of a particular script. Repeated measures ANOVA indicated a main effect for the construction type of sentence; (F(1,119)= 8.78, p<. 01). Speakers judged the existential sentences (M=3.33, SD=0.15) with higher ratings compared to locative sentences (M=3.61, SD= 0.14). These findings confirmed our hypothesis in that existential sentences received higher ratings than locative sentences. The items in which existential sentences are the first sentences received higher ratings because those sentences present new information which is expected in the first sentence. In the items in which locative sentences are the first sentences received lower ratings than those with existentials because, compared to existential sentences, locative sentences are not expected to introduce new information. There was a main effect of quantification type of noun phrases (F(2,238)=8.83, p<.01), plural type had the highest ratings (M=3.61, SD=0.15) followed by number type (M=3.55, SD= 0.15) and single type (M=3.25, SD= 0.16), respectively. Moreover, there was a significant interaction between sentence type and quantification type (F(2,238)= 14.77, p<.001). These findings confirmed our hypothesis in that existential sentences received higher ratings than locative sentences. These findings provide additional support to a previous study (Arik 2013). In that study, participants rated locative and existential sentences where nouns in the theme position were modified with number. Compared to plural marked nouns such as kitaplar, single type of nouns such as kitap or number modified noun iki kitap particularly in a locative sentence received lower ratings because they most likely provide nonspecific reading especially in the first sentences. These findings are

because of the method, Likert-type (graded) scale. One could wonder what participants would do when they are supposed to choose a particular type of sentence, say locatives, over the other one, existentials. We conducted a further study to investigate whether this was the case, which we report in the next section. 6 A follow-up study In a follow-up study (Arik & Arik 2012), a similar method was used, a forcedchoice scale instead of Likert-type scale to investigate whether the found difference between locatives and existentials held when the participant were asked to pick either locative or existential sentence. Sixty-four native speakers of Turkish participated in the follow-up study. They were all first year undergraduate students taking Introduction to Psychology course at Isik University in 2012. They signed the consent forms and gained an extra credit for their participations. These participants were not the same participants of the experiment reported in the previous section. The same design with a forced-choice scale was used in which participants were asked to choose, for example, either (13) or (14) which were given in a single line. There were two scripts with twenty-four items (twelve testing items with two sets and twelve fillers). Each item had two columns in which either a locative or existential sentence was presented on a single paper. Results showed similar results with our experiment. With a Cochran Q s test, it was found that there was a significant difference (χ2(5)=94.57, p<.001) among the forced choices with respect to locatives vs. existentials. A pairwise comparison between these factors with McNemar test showed significant differences between locatives (existential preference, 95.3%) and existentials (existential preference, 73.3%) due to new vs. old information; between bare nouns (existential preference, 80.4%) and number quantified nouns (existential preference, 91.3%); and, between number quantified nouns and plural nouns (existential preference, 81.2%) but no significant difference between bare nouns and plural nouns. As hypothesized, Turkish speakers preferred existential sentences over locatives sentences. Participants preferred existential sentences over locative sentences because, in the first sentences, the participants expected to get new information which was provided by existential sentences not locative sentences. There were also differences due to quantification in forced-choice preferences. As hypothesized, Turkish speakers preferred nonspecific nouns in the theme position over specific nouns in the same position. That is, paricipants preferred quantified nouns such as iki kitap over bare nouns such kitap or kitaplar. These findings together provide supporting evidence for the one reported in section 5 and for the previous study. 7 Conclusion Mostly based on insight-based research, it has already been shown that locatives and existentials are related to each other but differ from one another for syntactic and semantic-pragmatic reasons. This difference could be because of a DP/NP syntactic

distinction, a definiteness/specificity interpretation, or, maybe, new vs. old information (Milsark 1974, Erguvanli-Taylan 1987, Enç 1991, McNally 1992, Kornfilt 1997, Zucchi 2005, Öztürk 2005, Nakipoglu 2009) in locatives and existentials. From a pragmatics perspective, they differ because existentials but not locatives present new information. In the present study we tested this claim by using a new method, acceptability judgments. This study contributes to this line of research suggesting that new and old information makes a difference when it comes to distinguish locatives from existentials in Turkish. Previous research (Arik 2013) have already shown that the Turkish speakers acceptability judgments differ with respect to the type of quantification in NPs in theme roles and perhaps because of the introduction of new vs. old information in Turkish locatives and existentials. The current study also tested these findings. We have constructed scales in which there were two sentences: The first one was either a locative or an existential sentence and the second one was an adjectival sentence. We have also manipulated the theme NP with number (bare, number-quantified, plural) in these sentences. Providing additional evidence to previous research, results from the present study (acceptability judgments of the Likert-type scale) and the follow-up study (acceptability judgments of the forced-choice scale) have indicated that the Turkish speakers gave higher ratings to existential sentences than locative sentences. We attributed these differences to the new vs. old information in locatives and existentials. Existentials received higher ratings than locatives because the first sentences should introduce the new information, which can be provided by existentials but not locatives. There were also differences because of the number modifications of NPs. We attributed these findings to (non)specificity readings in existentials and locatives. As for any experimental study, this study suffers from several limitations. The scales were based on written language with little input from the participants other than follow-up comments and suggestions. However, we cannot predict what exactly participants are thinking or what kind of contexts the participants are creating at the moment of making acceptability judgments. There would be some differences in the results if the participants were exposed to audiorecorded sentences in addition to written sentences. In addition, we did not record response time of the participants. If we recorded, that would provide additional evidence to whether or not the participants spent longer time to rate locatives sentences than existential sentences due to the new vs. old information. Future research will investigate these issues. One could argue that type of modifiers could make a difference in the ratings of locatives and existentials because of the difference in their semantic category. Future research, therefore, will investigate other types of noun modification such as color and quality to further explore this topic. References Arik, E. 2013. Location and existence in Turkish from an experimental linguistic perspective. In: Arik, E. (ed.) 2013. New research into language learning, teaching and assessment. Ankara: MacroWorld, 1-12. Arik, E. & Arik, B.T. 2012. Location, Existence, and Quantification in Turkish: What do the

results from a forced choice scale show? In Botinis, A. (ed.) 2012. Proceedings of the International Conference of Experimental Linguistics (ExLing2012). 13-16. Barwise, J. & Cooper, R. 1981. Generalized quantifiers and natural language. Linguistics and Philosophy 4(2), 159-219. Enç, M. 1991. The semantics of specificity. Linguistic Inquiry 22, 1-25. Erguvanlı-Taylan, E. 1987. The Role of Semantic Features in Turkish Word Order. Folio Linguistica XXI 2-4, 215-229. Freeze, R. 1992. Existentials and other llocatives. Language 68, 553-595. Heine, H. 1997. Possession: Cognitive sources, forces, and grammaticalization. Cambridge, UK: Cambridge University Press. Jackendoff, R. 1990. Semantic structures. Cambridge, MA: MIT Press. Heim, I. 1987. Where does the definiteness restriction apply? In: Reuland and ter Meulen 1987. The Representation of (In)definiteness. Cambridge: MIT Press. 21 42 Keenan, E. 1987. A semantic definition of indefinite NP. In: Reuland and ter Meulen 1987. The Representation of (In)definiteness. Cambridge: MIT Press. 286 326. Kornfilt, J. 1997. Turkish. Routledge: London. McNally, L. 1992. An interpretation for the English existential construction. [PhD Dissertation, UCSC, CA]. McNally, L. 2011. Existential sentences. In: Maienborn, C., von Heusinger, K. & Portner, P. (eds.) 2011. Semantics: An International Handbook of Natural Language Meaning, Volume: 2. Berlin: de Gruyter. 1829-1848. Milsark, G. 1974. Existential sentences in English. [PhD dissertation, MIT, MA]. Nakipoğlu, M. 2009. The semantics of the Turkish accusative marked definites and the relation between prosodic structure and information structure. Lingua 119, 9, 1253-1280. Öztürk, B. 2005. Case, referentiality, and phase structure (Linguistik Aktuell 77). Amsterdam: John Benjamins. Stowell, Timothy. 1978. What was there before there was there. In: Farkas, D. (ed.) 1978. CLS 14, Proceedings of the Fourteenth Annual Meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society. 458 471 Zucchi, A. 1995. The ingredients of the definiteness and the definiteness effect. Natural Language Semantics 3, 33-7.

Running Head: Turkish locatives and existentials Contact Information: ARIK, Engin Doguş University Department of Psychology Assistant Professor Acıbadem, Kadıköy 34722 Istanbul Turkey enginarik@enginarik.com ÖZTOP, Pınar Plymouth University Faculty of Health Graduate Student pinaroztop@gmail.com BÜYÜKSÖKMEN, Esen esenbuyuksokmen@gmail.com