Running head: WORDS PROMOTE CATEGORIZATION AT 3 AND 4 MONTHS Categorization in 3- and 4-month-old Infants: An Advantage of Words Over Tones

Neonates prefer human speech to other non-linguistic auditory stimuli. However, it remains an open question whether there are any conceptual consequences of words on object categorization. Previous research shows that naming a distinct set of objects with a consistent label promotes categorization in 6-month-old infants. The current study examined the influence of words and tones on object categorization in 3to 4-month-old infants. Infants were familiarized to different exemplars of a category accompanied by either a labeling phrase or a tone sequence. In test, infants viewed novel and new within-category exemplars. Infants who heard labeling phrases provided evidence of categorization at test. In contrast, infants who heard tone sequences did not, suggesting that infants as young as three months of age treat words and tones differently vis a vis object categorization. Ferry, A., Hespos, S., & Waxman, S. (in press). Categorization in 3and 4-Month-Old Infants: An Advantage of Words Over Tones. Child Development. 3 Categorization in 3and 4-month-old Infants: An Advantage of Words Over Tones Questions concerning the relation between linguistic and conceptual organization hold a central position in the cognitive sciences. Most of the research investigating the dawning of this relation has focused on early word learning and has documented that by roughly their first birthdays, words support the formation of object categories (e.g., Balaban & Waxman, 1997; Fulkerson & Haaf, 2003; Namy, 2001; Waxman & Booth, 2003; Waxman & Braun, 2005; Woodward & Hoyne, 1999). More recent work has identified a link between words and object categories in infants as young as six months (Fulkerson & Waxman, 2007). Our goal is to consider this phenomenon in infants at three and four months. Background Waxman and Markow (1995) showed that for infants as young as 12 months of age, providing the same name for a set of distinct objects (e.g., four different animals) highlights commonalities among them and supports the formation of an object category (e.g., animal). For half of the infants in their task, the objects (e.g., the four animals) were introduced in conjunction with a novel noun (e.g., See the fauna?); for others, the very same objects were introduced with a general orienting phrase containing no novel word (e.g., See here?). At test, infants viewed two objects: a new member of the now-familiar object category (e.g., another animal) and a member of a novel object category (e.g., a vehicle), and heard, “See what I have?” Infants in the word condition showed a reliable preference for the novel object (e.g., a vehicle); those in the no word control condition revealed no preference. These outcomes document an early link between words and object categories in infants just beginning to build a lexicon. Ferry, A., Hespos, S., & Waxman, S. (in press). Categorization in 3and 4-Month-Old Infants: An Advantage of Words Over Tones. Child Development. 4 Additional evidence documents the specificity of this phenomenon within the first year of life. At 13 14 months, infants distinguish between kinds of novel words, mapping novel words presented as nouns (but not adjectives) specifically to object categories (Booth & Waxman, 2008; Waxman & Booth, 2001). But younger infants’ expectations about words are less precise. At 11 months, they do not distinguish between kinds of words (nouns and adjectives) in categorization tasks (Waxman & Booth, 2003). Nonetheless, their expectations are not so general as to include novel words under any circumstance: if the same novel word is applied to a set of distinct objects, 12-month-olds successfully form an object category, but if a different word is applied to each, they do not (Waxman & Braun, 2005). This indicates that by 12 months, infants successfully cull novel words from the ongoing speech stream, detect whether the same or different words have been applied to a set of objects, and expect that objects named by the same word (but not those named by different words) are members of a shared category. Importantly, although other kinds of auditory stimuli, including tones and melodic sequences, are engaging to infants, these stimuli do not promote object categorization even when they are consistently applied. Fulkerson and Waxman (2007) compared the influence of novel words and tone sequences on object categorization in infants at 6 and 12 months. Infants viewed pictures of exemplars from the same category (e.g., dinosaurs), presented sequentially. Infants in the word condition heard a naming phrase as they viewed each picture (e.g., Look at the toma! Do you see the toma?); those in the tone condition heard a sequence of pure tones, matched precisely to the word condition for timing, duration, and volume. At test, all infants viewed pictures of a new exemplar from the familiar object category (e.g., a new dinosaur) and an exemplar from a novel category (e.g., a fish), presented side-by-side. Infants in the word condition revealed a reliable preference for the member of the novel category, indicating that Ferry, A., Hespos, S., & Waxman, S. (in press). Categorization in 3and 4-Month-Old Infants: An Advantage of Words Over Tones. Child Development. 5 they successfully formed the object category. Infants in the tone condition failed to reveal reliable preferences, suggesting that they had not formed an object category. Thus, by 6 months of age, words facilitate object categorization in a way that well-matched tone sequences do not. It is important to point out that the claim is not that object categorization depends entirely on the presentation of words. Clearly, there are many circumstances in which humans, and even non-human animals, categorize successfully even in the absence of a novel word (e.g., Hespos & Spelke, 2004; Mareschal & Quinn, 2001; Plunkett, Hu & Cohen, 2008; Quinn, 2002; Santos, Sulkowski, Spaepen, & Hauser, 2002). The claim is more measured and more precise: even before infants produce words on their own, naming a set of objects with a consistently applied word promotes the formation of object categories, highlighting commonalities among objects in a way that other well-matched auditory stimuli (e.g., tone sequences) do not (Brown, 1957; Waxman & Markow, 1995). Interpreted in this light, the claim is that words support the ubiquitous process of categorization, and are especially effective in highlighting commonalities among objects that may otherwise have gone undetected (Fulkerson & Waxman, 2007; Gelman & Kalish, 2006; Plunkett, et al., 2008; Waxman & Markow, 1995). In the current study, we sought to examine the precursors to this phenomenon in infants at three and four months of age. We focused our efforts here because although 3and 4-monthold infants share with older infants an interest in human speech, it is not until roughly six months of age that they begin to parse individual words (especially their own names) from the stream of fluent speech (Bortfeld, Morgan, Golinkoff, & Rathbun 2005; Jusczyk & Aslin, 1995). Apparently, then, infants’ treatment of words may differ importantly from three to six months. Although infants prefer human speech to many types of non-linguistic auditory stimuli even at birth (Ramus, Hauser, Miller, Morris, & Mehler, 2000; Vouloumanos & Werker, 2007), whether Ferry, A., Hespos, S., & Waxman, S. (in press). Categorization in 3and 4-Month-Old Infants: An Advantage of Words Over Tones. Child Development. 6 this early preference has any consequences on categorization behaviors in infants younger than six months remains very much an open question. Our goal was to examine object categorization under two different auditory conditions, one involving human speech (the word condition) and the other serving as a non-language control (the tone condition). This design permitted us to discover whether words and tones exert the same influence on infants’ categorization, or whether words enjoy a privileged status vis a vis categorization as early as three or four months of age. Method Participants Participants included 46 healthy, full-term infants divided into two age groups. There were 24 4-month-olds (9 males, 15 females) with an average age of 4 months, 5 days (range: 3 months, 23 days to 4 months, 21 days) and 22 3-month-olds (13 males, 9 females) with an average age of 3 months, 7 days (range: 2 months, 19 days to 3 months, 21 days). Half of the infants within each age group were randomly assigned to the word condition and the others to the tone condition. All infants in the language condition were from families where English was the predominant language spoken in the home. An additional 13 infants were tested but excluded due to parental/sibling interference (2), fussiness (10), or falling asleep during the study (1). An additional 17 infants, who looked at the stage for less than half of the familiarization period, were included in separate analyses (5 were 3-month-old infants and 12 were 4-month-old infants). Other studies have shown that the amount of exposure infants receive during the familiarization period can affect whether infants will show a novelty or a familiarity preference during test trials (e.g., Theissen, Hill, & Saffron, 2005). We suspected that the low exposure to Ferry, A., Hespos, S., & Waxman, S. (in press). Categorization in 3and 4-Month-Old Infants: An Advantage of Words Over Tones. Child Development. 7 the familiarization stimuli may have impacted their ability to form a category and separate posthoc analyses were conducted to examine that possibility. Participant families were recruited from a commercial mailing list from the surrounding area and contacted by mail and telephone. They were given a t-shirt or a book as a gift but were not otherwise compensated for their participation. The participants were predominantly from white, middle-class families. The ethnicity of the sample was 80% non-Hispanic. The racial make-up was 75% white, 5% Asian, 3% Black/African American, 9% multiracial, and 8% chose not to answer. Stimuli The stimuli were identical to the stimuli used in Fulkerson and Waxman (2007). Visual Stimuli: Twenty slides of line-drawn dinosaurs and fish were produced to form two familiarization sets of eight stimuli each and two sets of two test stimuli each (see Figure 1). Stimuli were outlined in black and filled in with a solid color. Within each familiarization set, stimuli varied in color; within each test set, stimuli were matched in color. Auditory stimuli: Two sets of naming phrases (“Look at the toma/modi. Do you see the toma/modi?”) were spoken by a female in the infant-directed speech register and recorded for presentation. Two sequences of pure tones (400 and 800 Hz), matched to the naming phrases in duration (2.2s), pause length, and volume (82 dB) were created. Apparatus Infants sat on a parent’s lap in a small room in front of a wooden puppet stage measuring 243.5 cm high, 128 cm wide and 61 cm deep. The infants faced an opening in the front of the stage (93 cm above the floor, 61 cm high, and 106 cm wide). Visual stimuli were projected onto a white screen via a slide projector from behind the stage. A white projection screen was on the Ferry, A., Hespos, S., & Waxman, S. (in press). Categorization in 3and 4-Month-Old Infants: An Advantage of Words Over Tones. Child Development. 8 stage behind a sheet of black foam core. There were two 18 cm by 18 cm windows cut into the cardboard. There was 11 cm of separation between the two windows, which were centered approximately 75 cm from the infants’ eye level. Auditory stimuli were presented through a central speaker hidden 56 cm below of the screen. Infants were videotaped through a 3 cm hole in the stage located 12 cm below the projection screens. Procedure The novelty-preference task included a familiarization phase and a test phase (see Figure 1). All infants saw the same visual stimuli materials (either dinosaurs or fish). What differed was the accompanying audio: Infants were randomly assigned to either the word or the tone condition. Parents were instructed not to influence their infant’s attention in any way and to close their eyes during test trials. Familiarization phase. Infants were presented with a sequence of eight different exemplars from a single category (either dinosaurs or fish). The category used was counterbalanced across both word and tone conditions. In the word condition, infants heard a naming phrase as they viewed the stimuli; in the tone condition, infants heard a sequence of pure tones. In both conditions, the auditory stimulus was introduced once at the onset of the visual stimulus and again 10s after the onset of the visual stimulus. The trials were a fixed duration of 20 s and the order of presentation was randomized across participants. The position of the image alternated sides from trial to trial. The following factors were counterbalanced across participants: left/right position of the first image, use of toma/modi or 400/800 Hz, and familiarization to a picture of a dinosaur or a fish. Test Phase. All infants saw the same two test stimuli: a new exemplar from the familiar category and an exemplar from a novel category. These were presented side-by-side in silence. Ferry, A., Hespos, S., & Waxman, S. (in press). Categorization in 3and 4-Month-Old Infants: An Advantage of Words Over Tones. Child Development. 9 The test phase began with the infants’ first visual fixation to one of the stimuli and lasted until the infant had accumulated 10 s of looking to the test stimuli. The left-right positioning of the test stimuli was counterbalanced across infants. Coding. Trained observers, blind to the condition assignment, coded infants’ visual fixations online. Later, off-line frame-by-frame coding to determine left and right looks was conducted using ELAN (2007). Thirty-three percent of infants in each age and condition were re-coded by an independent observer. Reliability between observers was 91% during familiarization and 90% during test. To measure categorization, a preference score was calculated for each infant by dividing their looking time to the novel-category exemplar at test by their total looking time (10 s). Chance performance was .50. Preliminary analyses revealed no significant effect of sex, the specific word or tone used during visual stimuli, or left/right position of the test stimuli; the data were therefore collapsed across these variables in subsequent analyses.