3.1 Method – The Headturn Preference Paradigm
3.1.1 Reliability of the HTP paradigm and range of application
In their evaluation of the reliability of the HTP paradigm, Kemler Nelson and her colleagues (Kemler Nelson, Jusczyk, Mandel, Myers, Turk & Gerken, 1995) compared the online coding of the OT during the experiment and data obtained from offline coding by a second experimenter. Both experimenters are blind with respect to the kinds of stimuli presented and to the randomised order of presentation of these stimuli in order to prevent influences on the data recording. During online coding, the experimenter controls the experiment with a response box which is connected to a computer. She starts the flashing of the centre lamp by pushing a button on the box, and when the infant orients towards this lamp, one of the side lamps is activated by a second push of the same button. The side on which the side lamp starts to flash is computer controlled. When the infant orients towards this lamp, a third push of the button starts the presentation of the auditory stimuli. The time the infant spends looking towards the side on which the stimulus is presented is recorded by the computer. If the infant turns away from the presentation side, the time recording is stopped by pushing another button on the box. If the infant returns her attention within a two second predefined cut-off window, the recording of the time is continued. If the infant looks away for more than two seconds, the presentation of the stimulus is stopped and a new trial is started. The experiment is videotaped to allow for offline coding by a second experimenter.
For this offline coding the second experimenter views the videotape of the experiment and also uses the push-button box in the same way as described above, as if online coding the experiment. The data obtained by this second coding is compared to the data obtained from the original coding. Kemler Nelson et al. (1995) report that the inter-coder reliability was very high (r = 0.94 to r = 0.99; ibid., p. 113f). Other studies
confirmed the high inter-coder reliability (cf., Seidl, 2007). Thus, the HTP paradigm is a credible tool in the research of infants’ language acquisition and development.
Compared to two other methods used in the research on language acquisition in infants, namely the High-Amplitude-Sucking procedure (HAS) (Eimas, 1974; Mehler et al., 1988) and the Conditioned Headturn technique (CHT) (Morgan, 1994; Werker & Tees, 1984), the HTP has the advantage of being applicable to a wider age range when testing infants’ language acquisition development.
The HAS paradigm is the only one of these three methods which can be used with newborns, but it can only be applied up to the age of 4 to 6 months, the age when the natural sucking reflex in the baby starts to decay. The HAS is a habituation paradigm in which the natural sucking reflex is used to test infants’ discrimination abilities. When presented with a stimulus, the infant usually starts to suck faster and / or more intense, that is, with a high amplitude, on a blind nipple connected to a transducer and computer. The longer the infant listens to the stimulus, the more habituated she becomes, reflected in a decrease in sucking rate. When the sucking rate falls below a predetermined criterion amplitude, the next high amplitude suck will result in the presentation of a new stimulus type (experimental group) or the same stimulus type (control group). If the infant is able to detect the stimulus change, an increase in sucking rate can bee seen compared to infants of the control group. This paradigm has been used to test infants’ sensitivities, for instance, to rhythmic properties of language (cf., chapter 2.2).
The CHT technique has been successfully used with infants between 5.5 and 18 months, but it seems to be most suited to test infants between 6 and 10 months, because at these ages “the infant is a ‘captive audience’ and not easily bored with the reinforcer” (Werker, Polka & Pegg, 1997, p. 173).34 In the first phase of an experimental session, the infant is trained to turn her head towards a loudspeaker on the detection of a stimulus change compared to the background speech or sound stream. If the detection was correct a reinforcer, usually a mechanical toy, starts to move. A second experimenter is needed in
34 However, with modifications this method may be used with older children and even adults. These
modifications include using a button which is pressed instead of the conditioned headturn, when a stimulus change is identified (Werker et al., 1997).
this kind of experiment to divert the infant from the reinforcer, so that a headturn can be uniquely interpreted as a response to the detection of a change in the auditory material. During the test phase the auditory material is changed and the recognition of the stimulus on which the infants was trained can be observed. The CHT technique “may provide a more sensitive index of infants’ speech processing abilities [than the HTP]” as it is “time- locked to particular stimuli” (Gout et al., 2004, p. 555).35 This technique was used in studies on word recognition (cf., chapter 2).
However, unlike CHT, which can only be used to test the discrimination and categorisation of relatively short speech items, like syllables, words or short melodic patterns (Werker et al., 1997, p. 176), the HTP paradigm also allows to study infants’ knowledge about more complex structures in language like dependency structures (Santelmann & Jusczyk, 1998; Höhle, Schmitz, Santelmann & Weissenborn, 2006), or clausal structures, which are the main interest of this study (Hirsh-Pasek et al., 1987; Hayashi et al., 1996; Hayashi & Mazuka, 2002). In addition, the HTP has been successfully used to study infants’ language acquisition development from 4 months36 to about 20 months of age (Höhle, Weissenborn, Schmitz & Ischebeck, 2001; Höhle, 2002; Höhle et al., 2006; Santelmann & Jusczyk, 1998).