• Keine Ergebnisse gefunden

Chapter 6 Focus

6.2 Speech production

6.2.2 Method

6.2.2.1 Participants

The study was conducted with 16 students from the University of Bielefeld with Standard Turkish as their native language, 16 students from the Univer-sity of Bielefeld with Russian as native language as well as with 16 native speakers of Caucasian Urum in Tbilisi, Georgia. The 16 native speakers of Turkish (10 female, 6 male) ranged in age from 20 to 25 with a mean age of 22.10 years. The 16 native speakers of Russian (9 female, 7 male) ranged in age from 20 to 31 with an average of 24.93. Due to the fact that the study took place at a German university, the Russian and Turkish speakers in the study were partly bilingual to different degrees. Therefore, all participants were asked to rate the frequency with which they use their native languages on a range from 1 (=rarely) to 5 (=very frequently/several hours a day), which resulted in an average of 4.25 for the Russian speakers and 4.44 for the Turkish speakers. The 16 native speakers of Urum (9 female, 7 male) ranged in age from 16 to 73 with an average of 39.94 years. All of them

were born in Georgia and considered themselves as native speakers of Urum.

However, all of them are also fluent in Georgian and Russian.

6.2.2.2 Material and design

The applied method is the elicitation of semi-spontaneous answers to several different questions. By contrast to previous elicitation studies on the effect of focus on word order, the present study minimizes the risk of too many invalid tokens due to the tendency of using pronouns instead of full lexical NPs by presenting visual stimuli with more than one entity. Within the study the participants were presented 16 target pictures, which were designed with the online comic-making toolPixton Comics. All pictures were colored and depicted a scene with two animate entities (=agent), one of them involved in an action with an inanimate entity (=patient), see for instance the sample picture in Figure 6.1.

Figure 6.1: Example of visual stimuli used in elicitation task.

The experiment used a 2x2 factorial design with the factorsFOCUS TYPE

(two levels: non-identificational vs. corrective) andFOCUSED ARGUMENT

(two levels: subject vs. object). The permutation of the levels lead to four experimental conditions, see the design in Table 6.1.

Table 6.1: Experimental design of focus-elicitation study

F-TYPE

non-identificational corrective

F-ARGUMENT subject N/SBJ C/SBJ

object N/OBJ C/OBJ

Each of the four conditions of the experiment was matched with one specific question type. In order to test the two non-identificational conditions (N/SBJ;N/OBJ) I used simplewh-questions that trigger an answer with either a narrow focus on the subject or the object. In order to examine the two corrective conditions (C/SBJ;C/OBJ) I created questions that trigger answers

which involve a correction of either the subject or the object argument. The questions were translated into Turkish, Russian and Urum and recorded by native speakers of the respective language. See (218) for the Turkish translations of the four sample questions.

(218) a. Turkish: N/SBJ, OSV Muz-u

banana-ACC

kim who

yi-yor?

eat-PROG[3]

‘Who is eating the banana?’

b. Turkish: N/OBJ, SOV Adam

man ne what

yi-yor?

eat-PROG[3]

‘What is the man eating?’

c. Turkish: C/OBJ, SOV Kadın

woman muz-u banana-ACC

yi-yor eat-PROG[3]

mi?

Q

‘Is the woman eating the banana?’

d. Turkish: C/SBJ, SOV Adam

man

elma-yı apple-ACC

yi-yor eat-PROG[3]

mi?

Q

‘Is the man eating the apple?’

The examples in (218) show that the word order of the Turkish questions is not consistent among the four conditions. This results from the fact that the most unmarked position of a wh-word in Turkish is the immediately preverbal position (Kornfilt 1997: 10). Hence, the most natural order for a wh-question with a subject focus is OSV, whereas it is SOV if the focus is on the object. By contrast, the most natural order of a question which triggers a correction of either the subject or the object is the basic order SOV. Moreover, it must be noted that all direct objects in the target questions are marked with the accusative suffix(-y)I, which can be attributed to the fact that bare direct objects in Turkish are restricted to the immediately preverbal position, whereas the position of marked direct objects is free (Erguvanlı 1984: 27).

Consequently, Scrambling objects over subjects is not possible with bare objects, though it is felicitous with marked direct objects. Furthermore, the examples in (218c)-(218d) show that the corrective focus questions are formed with the particle mI, which is attached at the end of the questions and has scope not only over the focused argument but also over the whole question (Kornfilt 1997: 5).

For the Russian translations of the four sample questions, consider (219).

(219) a. Russian: N/SBJ, SVO Kto

who yest’

eat:IPFV[3]

banan?

banana:ACC

‘Who is eating the banana?’

b. Russian: N/OBJ, OVS Chto

what yest’

eat:IPFV[3]

muzhchina?

man

‘What is the man eating?’

c. Russian: C/SBJ, OVS Banan

banana:ACC

yest’

eat:IPFV[3]

zhenshchina?

woman

‘Is the woman eating the banana?’

d. Russian: C/OBJ, SVO Muzhchina

man

yest’

eat:IPFV[3]

yabloko?

apple:ACC

‘Is the man eating the apple?’

Similar to the Turkish questions, the word order of the Russian questions is not consistent among all conditions. The difference in the order of the wh-questions is due to the fact that the basic position of interrogative words in Russian is sentence-initial (Wade 2011: 525). By contrast, the most natural position for a corrective focused argument is considered to be the postverbal position, which leads to an OVS order for a corrective subject focus question and a SVO order for a correct object focus question.

The Urum translations of the four sample questions are finally illustrated in (220).

(220) a. Urum: N/SBJ, SVO Kim

who i-er

eat-IPFV[3]

banan-i?

banana-ACC

‘Who is eating the banana?’

b. Urum: N/OBJ, OVS Nä-i

what-ACC

i-er

eat-IPFV[3]

ärgishi?

man

‘What is the man eating?’

c. Urum: C/SBJ, SVO Gari˘

woman i-er

eat-IPFV[3]

banan-i?

banana-ACC

‘Is the woman eating the banana?’

d. Urum: C/OBJ, SVO Ärgishi

man

i-er

eat-IPFV[3]

alma-yi?

apple-ACC

‘Is the man eating the apple?’

The translations of the questions were done by an Urum native speaker.

It is interesting that she used verb-medial orders among all conditions, which corresponds to the Russian word order. However, though the position of the verb is similar to Russian, the orders of the questions are in turn resembling the Turkish questions. Like in Turkish, wh-words in Urum are generally realized left adjacent to the predicate (Skopeteas 2013: 349). Due to the fact that the native speaker used a verb-medial construction, the left-adjacent position is identical with the sentence-initial position, which explains the SVO order in the non-identificational subject focus question and the OVS order in the non-identificational object focus question. Moreover she used the canonical (SVO) order for both types of corrective focus questions.

In sum, the translations of the sample questions show that the word orders of the different questions types are not consistent within and across languages. This results from the fact that the questions should sound as natural as possible to the participants. This fact has to be kept in mind when analyzing the results.

To ensure that each of the 16 target pictures is presented to every par-ticipant with only one of the four different question types, the study used a Latin square design. Hence, every participant within the study produced a total of 16 answers, i.e., 4 answers per each condition. The experiment was conducted together with another elicitation task on topics (cf. Chapter 7).

The items of both experiments thus functioned as distractors to each other.

Moreover, the order of all items was pseudo-randomized for each participant.

Consider Appendix A for a list of all experimental stimuli used in the focus elicitation study.

6.2.2.3 Procedure

The participants were told that the study investigates the effect of visual stimuli on speech production and that their descriptions are going to be

audio-recorded for this purpose. If they agreed to the prerequisites for attending the study, the participants were asked to go carefully through the instructions which were presented in their respective native languages.

Within the instructions the participants were told that they would see a set of different pictures and picture pairs. They were instructed to pay attention to these pictures, because after five seconds the pictures disappear and they should either give an answer to a question or in case of the topic elicitation study (cf. the description of the procedure of the topic study in Section 7.2) a short description of the presented scene. The participants were requested to avoid elliptical answers such as ‘yes’ or ‘no’, but to produce full sentence answers. The stimuli were presented to the participants with the experimental software DMDX. The participants were listening to the questions through high performance headphones. The presentation stopped after every stimulus. After giving their answers, the participants had to press the space bar on the computer keyboard to proceed with the next item.

Before the experiment started, four practice trials illustrated the procedure of the study. All participants completed the experiments individually. The participation was voluntary and paid. After the completion of the study, all recordings were transcribed in order to have a written record of the speakers’

descriptions.