First Language
r Fo Journal: Manuscript ID:
Keywords:
First Language
FLA-14-0005.R1
Original Manuscript
Re
Manuscript Type:
er
Pe
Effects of prosodic cues on topic continuity in child language production
prosodic cues, sentence continuation, information structure, function assignment, linguistic cues integration
vi
ew
Abstract:
It remains controversial whether children are able to process and integrate specific linguistic cues in their mental model to the same extent as adults. In the present study, a sentence continuation task was employed to determine how Italian speakers (4-, 5-, 6-year-olds, and adults) interpret prosodic cues to decide which referent is more salient in a sentence, and thus more likely to be continued as the subject of a subsequent utterance. Participants as young as five years of age showed the tendency to mention next, as the subject of their sentence continuations, the referent that received prosodic stress in the previous utterance. Conversely, 4-year-olds did not appear to rely on prosodic cues in order to decide which referent to mention next. These findings suggest that sensitivity to prosodic cues and the ability to integrate them in a further linguistic plan evolves with age and that a developmental change occurs between four and six years of age.
http://mc.manuscriptcentral.com/fla
Page 1 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Abstract It remains controversial whether children are able to process and integrate specific linguistic cues in their mental model to the same extent as adults. In the present study, a sentence continuation task was employed to determine how Italian speakers (4-, 5-, 6-year-olds, and adults) interpret prosodic cues to decide which referent is more salient in a sentence, and thus more likely to be continued as the subject of a subsequent utterance. Participants as young as five years of age showed the tendency to mention next, as the subject of their sentence continuations, the referent that
r Fo
received prosodic stress in the previous utterance. Conversely, 4-year-olds did not appear to rely on prosodic cues in order to decide which referent to mention next. These findings suggest that sensitivity to prosodic cues and the ability to integrate them in a further linguistic plan evolves with
Pe
age and that a developmental change occurs between four and six years of age.
er ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
1
http://mc.manuscriptcentral.com/fla
First Language
Page 2 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Introduction In connected utterances, the choice of which referent to mention as the subject of a subsequent utterance is determined by a multitude of variables. For instance, within a sentence, two referents might be equal with respect to the degree of animacy, of giveness, and frequency of the noun phrases (henceforth NPs), etc. that they involve. However, several different linguistic cues, such as syntactic position or prosodic stress, might influence the accessibility of each referent. In the present paper, we investigate whether children as young as four years of age make use of
r Fo
prosodic cues, which we will generically refer to as prosodic emphasis, to decide which referent is more salient in an utterance and thus more likely to be assigned to subject position in a subsequent sentence.
Pe
The syntactic prominence of a referent in an utterance might increase its salience (McKoon
er
et al., 1993). For example, consider a picture showing two characters engaged in a transitive event. In such a case, the event could be described as in 1: (1) The zebra is soaking the hippo.
vi
Re
‘zebra’, the subject, is more syntactically prominent than ‘hippo’, in the direct object
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
position (McKoon et al., 1993), and thus more salient. According to the assumption that people are likely to construe as topic of their sentence what the current focus of attention is (Grosz and Sidner 1986; Grosz, Joshi, and Weinstein, 1983; but see also Brennan, 1995), it is more likely they will make ‘zebra’ the entity the subsequent sentence will be about. As a consequence, a likely continuation of 1 could be 2: (2) And then the zebra is drying it.
2
http://mc.manuscriptcentral.com/fla
Page 3 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
where the referent (i.e., zebra) which was more salient in the previous utterance occurs in the subject position. Note, however, that there are other linguistic and extra-linguistic (i.e., perceptual salience) means that could make ‘hippo’ the most salient referent in a sentence, even though ‘hippo’ is not in the subject position. For instance, consider a scenario in which the same sentence is presented together with a picture that shows an additional character (say, a giraffe) that could get soaked. In such a case, the same sentence can be expressed with prosodic emphasis on ‘hippo’ NP, capitalised in 3, to provide a contrast between hippo that did get soaked, and giraffe, that did not.
r Fo
(3) The zebra is soaking the HIPPO. Thus, one possibility could be that in 3, ‘hippo’ is more salient than ‘zebra’ because it receives prosodic emphasis. Our example suggests that a number of possible linguistic and extra-
Pe
linguistic cues (i.e., the placement of characters in the picture or their perceptual salience) have
er
important implications in determining which element is perceived as more prominent in the speaker’s mind. In the current study we will consider only a linguistic source of salience, namely,
Re
prosodic emphasis.
Under the assumption that the most salient referent within a sentence tends to be mentioned
vi
next as the subject of the consecutive utterance, a possible continuation of 3 could be: (4) And the hippo is getting wet.
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Now consider the case in which the same sentence is presented together with a picture where there is another character besides the zebra (again, a giraffe), that could soak the hippo. In this case, ‘zebra’ would receive prosodic emphasis, as in 5. (5) The ZEBRA is soaking the hippo.
3
http://mc.manuscriptcentral.com/fla
First Language
Page 4 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
After 5, a speaker would presumably tend to mention next ‘zebra’, and make it the subject of the subsequent utterance, as ‘zebra’ is salient with respect to syntactic position and stress placement. As our examples suggest, the placement of prosodic emphasis, as well as the placement of characters in a picture, has important implications for the interpretation of what referent is more salient in a sentence, and crucially more likely to be mentioned next, as the subject of an upcoming utterance. It follows that, by means of a sentence continuation task, one could see which
r Fo
information structure (cf. Vallduvì, 1992) has been assigned to the previous sentence by the speaker, in terms of which referent was perceived as more salient, and thus more likely to be mentioned next. Hence, the use of a sentence continuation task arises from the assumption that
Pe
prosodic prominence is a device that could serve focusing attention towards a specific entity in the sentence.
er
In the present study, we manipulated prosodic emphasis, respectively, on the subject and on
Re
the direct object referent of subject-verb-object (SVO) matrix active sentences, to test to what extent prosodic cues influence sentence continuations of 4-, 5- and 6-year-olds and adult Italian
vi
speakers. Note that in the current report we are using a theoretically neutral term, i.e., prosodic
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
emphasis, to indicate that one referent was bearing prosodic stress in contrast to the other one. As far as we know, no previous study has tested to what extent a manipulation of prosody influences sentence continuation in adult and child language production. Previous psycholinguistic studies have investigated how several different manipulations of the preceding linguistic context can affect sentence continuation in adult language production. For instance, Tannenbaum and Williams (1968) asked adult participants to read a story in which the agent of a transitive event, the patient, or neither of them, was made salient. Then, participants were asked to describe a transitive 4
http://mc.manuscriptcentral.com/fla
Page 5 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
event involving both elements. Participants produced passives more quickly after a preamble that made the patient more salient, since the passive is one way to keep the object referent more salient in a sentence; similarly, participants produced actives more quickly after a preamble that made the agent more salient. A cross-linguistic study manipulated the preceding discourse context to make the object or the subject referent salient both in English and Spanish. Results indicated that, when the object referent was made salient in the preceding context, English speakers tended to prefer passives (e.g., ‘The woman was run over by the train’) to describe transitive events that involved both elements, whereas Spanish speakers also produced ‘dislocated actives’, with object-verb-
r Fo
subject order (e.g., A la mujer la atropelló el tren, literally: ‘To the woman her ran over the train’). In this case, the referent more salient in the previous story was mentioned first, even though it was
Pe
not in subject position (Prat Sala & Branigan, 2000). Overall these studies indicate that a linguistic manipulation of the previous context affects the syntactic encoding of a subsequent utterance. That
er
is, speakers show the tendency to place in subject position, or in the first linear position, the referent that was made salient in the previous context.
Re
But to what extent do speakers make use of prosodic cues to decide which referent in a
vi
sentence is more salient? A previous study investigated sensitivity of adults to prosodic cues in comprehension. By means of a magnitude estimation test (Bard et al., 1996), Keller and
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Alexopoulou (2001) tested the acceptability of six different word orders in Greek declarative sentences. In the spoken sentences, accent position was varied to affect which referent was more salient in a sentence. Results indicated that prosody, in contrast to word order, significantly influenced the acceptability of sentences. The authors concluded that, even in a free word order language such as Greek, prosody played a major role in marking which referent was more salient in a sentence.
5
http://mc.manuscriptcentral.com/fla
First Language
Page 6 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
As for children, experimental research showed that they do have an understanding of the (communicative) function of the prosodic cues at least by the age of two (Grassmann and Tomasello, 2010), although their productive skills outstrip the receptive ones (e.g., Cutler and Swiney, 1987). Additionally, a study on the acquisition of determiner use in French, Austrian German and Dutch pointed out that the early use of prosodic information (noun-length) may vary with respect to the language being acquired (i.e., Romance vs. West Germanic languages) (Bassano et al., 2013).
r Fo
A study of production indicates that English 3- to 4-year-olds exploits contrastive stress to pronounce the word carrying the contrastive information to describe pairs of pictures that differed by one feature (Hornby & Hass, 1970). Wells, Peppe´ and Goulandris (2004) found that English 5-
Pe
year-olds could use accentuation to mark contrast in adjective+NPs, although there was misplacement of accent in adjective-contrast. Chen (2011) examined the intonational realization of
er
focus and topic in SVO sentences in Dutch 4- to 5-year-olds by means of an answer-reconstruction task. Children were adult-like in accenting sentence-initial focus and topic mostly with a falling
Re
accent, but realized sentence-final focus with accentuation and sentence-final topic with no accent
vi
at the age of four or five and only became adult-like in the choice of accent type in sentence-final focus at the age of seven or eight. These results were confirmed in an analysis of German children’s
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
productions from short narrations elicited by a picture-story telling task (de Ruiter, 2009): 5-yearolds consistently marked new information with accentuation, though accent types did not show the same distribution that was found in adults. In contrast to production, the bulk of studies on the comprehension of prosodic cues suggests that child’s sensitivity is not well-established as in adults. For instance, McDaniel and Maxfield (1992) suggested that English 5-year- olds did not show an adult-like competence in 6
http://mc.manuscriptcentral.com/fla
Page 7 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
interpreting contrastive stress in sentences like: ‘Goofy is whispering to Grover. Now YOU whisper to him’. In this example, adults tended to interpret ‘him’ as Grover, whereas children interpreted ‘him’ as Goofy. Another piece of research demonstrated that English children as old as five years did not make use of prosodic information, and specifically of contrastive stress, to determine which referent was associated with the focus operator ‘only’ (Gualmini et al., 2002). The authors manipulated the prosody of a sentence, such that it could be true or false with respect to a preceding story, depending on whether the focus operator ‘only’ was associated with the indirect object or with the direct object. In Experiment 1, 4-year-olds tended to prefer to associate the focus operator
r Fo
‘only’ to the indirect object (e.g., ‘Barney only sold a cake to SNOW WHITE’) more often than to the direct object (e.g., ‘Barney only sold a CAKE to Snow White’). Using sentences analogous to
Pe
those of Experiment 1 (e.g., ‘Tarzan only threw a fish to the penguin’), Experiment 2 revealed that children have access to the alternative interpretation in which the focus operator ‘only’ is associated
er
with the direct object, but only under a condition that makes the indirect object interpretation wrong (e.g., under the case that Tarzan threw a fish and a boat to the dolphin, and not to the penguin).
Re
Another study in a Romance language (Portuguese) revealed that 3- to 5-year-olds could correctly interpret sentences with focus-related word order variation, only when provided with a discourse context (Costa & Szendröi, 2006).
ew
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
A more positive result was reported by Chen (2010) who tested the comprehension of a specific prosodic cue, i.e., narrow focus stress, in a group of Dutch 4- and 5-year-olds. In a question-answer dialogue (‘Look! A chicken! What is the chicken eating? The chicken is eating a PLANT.’), the author varied stress placement respectively on the subject or on the object referent in the answer (‘chicken’ or ‘plant’) and then asked adults and children to judge whether answers sounded correct or not. Interestingly, accent placement played no role in the correct-incorrect judgment task for children. Conversely, RTs to the judgment task revealed that appropriate stress 7
http://mc.manuscriptcentral.com/fla
First Language
Page 8 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
placement triggered shorter RTs in both adults and children. The author concluded that 4- and 5year-olds are able to process prosodic cues, such as narrow focus stress, to a greater extent than was expected given the literature. A series of eye-tracking studies offered further evidence that children do show some sensitivity to prosodic cues during the on-line processing of a sentence, but they use them less effectively than adults. For instance, Zhou et al. (2012) tested the sensitivity to prosodic cues in ambiguity resolution in 4- and 5-year-old Mandarin children, using eye-tracking methodology and a
r Fo
judgment task. Results indicated that children’s eye movements were sensitive to stress, i.e., children were able to direct their eyes to the target element, according to the pitch, whereas their accuracy to the judgment task was poor. The authors concluded that children younger than five
Pe
were sensitive to prosodic cues but that they were not yet fully able to integrate them in their mental model during the judgement task. In another eye-tracking study involving the manipulation of pitch
er
accent on Adj+NP phrases, English-speaking children revealed that pitch accent exerted facilitative effects in reference resolution (Ito et al., 2008). However, the authors observed that children
Re
exploited anticipatory referential processing at the adjective (‘blue’) only when the preceding
vi
sentence explicitly implied a contrast set (i.e., ‘First hang the green ball. Now hang the blue ball’). Additionally, this anticipatory processing was enhanced by the presence of an appropriate pitch accent on the second adjective (e.g., ‘the BLUE ball’).
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Sekerina and Trueswell (2012) recorded Russian 5- to 6-year-olds’s eye-movements as they processed sentences involving the Russian-specific ‘Split Constituent’ construction (e.g., RED put butterfly in the paper bag) vs. NonSplit construction (i.e., red butterfly put…). Split constituent construction is a signal of contrastive focus when the pitch accent is placed on the leftmost split component of an Adjective+NP phrase. Pitch accent on the adjective did facilitate children’s 8
http://mc.manuscriptcentral.com/fla
Page 9 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
identification of the target referent, but only for the NonSplit constituents. Additionally, children showed some anticipatory processing only when the inference about the target referent was provided by the preceding utterance. The bulk of the previous studies using on-line methodologies like eye-tracking, reveals that children as young as four do show some sensitivity to prosodic cues, though their ability to integrate them in their mental model and use them effectively is still lacking. Interestingly, none of the previous studies has directly addressed to what extent the processing of a prosodic cue is integrated
r Fo
in the hearer’s linguistic representation, affecting the planning of a subsequent utterance. Importantly, a continuation task will allow to assess whether children: i) rely on prosodic cues to interpret what referent is more salient in a sentence; ii) know that the more salient entity should
Pe
more likely appear as the subject of the next utterance; iii) use this information in their own subsequent production. To achieve these goals, we asked 4-, 5-, 6-year-olds, and adult speakers of
er
Italian to generate a continuation of three types of sentences. Sentences could present a default stress on the last NP (‘Then the zebra is soaking the hippo’), as assigned by the nuclear stress rule
Re
(Chomsky & Halle, 1968; Cinque, 1993); prosodic emphasis on the subject (‘Then the ZEBRA is
vi
soaking the hippo’); prosodic emphasis on the direct object (‘Then the zebra is soaking the HIPPO’).
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Under the assumption that the most salient referent in a sentence is more likely to be mentioned next and appear as the subject of a subsequent utterance, we can make a number of predictions. First, in a sentence with neutral stress, assuming that both referents are equal with respect to animacy, predicability (Bock & Warren, 1985), etc., the subject referent is more salient than the direct object, as it appears in the most prominent syntactic position. In this case, as no other cue contributes to making the other referent salient, the subject referent should more likely appear 9
http://mc.manuscriptcentral.com/fla
First Language
Page 10 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
as the first mentioned referent of the continuation and crucially in subject position. Second, in a sentence with prosodic emphasis on the subject, both syntactic position and prosodic emphasis promote the salience of the same referent, the subject; thus participants should be likely to produce sentence continuations with subject referent in subject position. Third, in a sentence with prosodic emphasis on the object, the subject referent is salient because of its prominent syntactic position, whereas the object referent is salient because it receives prosodic emphasis. In this case, prosodic emphasis counteracts the effect of syntactic position. If prosodic emphasis on the object is strong enough to contrast the effect of syntactic position, we should observe more continuations with
r Fo
object referent in subject position, than after the other two sentences. Given the fact that sensitivity to prosodic cues develops with age, we expected prosodic
Pe
emphasis to affect sentence continuations to a different extent across age groups. Importantly, note that in order to substantiate the claim that there are across-age differences in how children respond
er
to stress, we would need to demonstrate that there is an interaction between age group and stress manipulation. In such a case our results will possibly allow us to make interesting claims about the
Re
development of children’s sensitivity to prosodic cues and crucially about their ability to
vi
incorporate them in their speech plan. Differences would provide evidence about how sensitivity to prosodic cues and its influence on sentence continuation proceeds with age. Method
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Participants Children were recruited from kindergarten and primary schools in Lecco (Lombardy) and Milan and placed into either 4- (12 children; 6 M; age-range 4;0 to 4;8, M = 4;3), 5- (13 children; 5 M; age-range 5;0 to 5;9, M = 5;5) and 6-year-old (12 children; 6 M; age-range 6;0 to 6;8, M = 6;4)
10
http://mc.manuscriptcentral.com/fla
Page 11 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
groups. Children had no personal history of language delay. Parental consent was collected for each participant. The adult group consisted of eight students of the University of Milano-Bicocca, Italy (one male; age range 19;1 to 23;1, M = 21;7). Materials The experimental materials consisted of 18 sets of sentences describing events involving two animals partaking in a transitive action (see Appendix A). Every set included a preamble such as: Questi pesci nuotano nel mare; ‘These fishes are swimming in the sea’. The preamble was
r Fo
followed by three possible transitive sentences involving two characters of different genders (in Italian) such as:
Pe
(1) neutral-stress: a SVO sentence with default nuclear stress on the last NP (E poi lo squalo insegue la balena; ‘And then the shark is chasing the whale’);
er
(2) subject-stress sentence: a SVO sentence with prosodic emphasis on the subject referent
Re
(e poi lo SQUALO insegue la balena; ‘And then the SHARK is soaking the whale’); (3) object-stress sentence: a SVO sentence with prosodic emphasis on the object referent (E
vi
poi lo squalo insegue la BALENA; ‘And then the shark is chasing the WHALE’).
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Thus, we manipulated whether the sentence presented: (i) default sentence stress (neutralstress condition); (ii) prosodic emphasis on the subject referent (subject-stress condition); (iii) prosodic emphasis on the object referent (object-stress condition). The independent variable was manipulated within-participants and within-items. Sentences were recorded by a trained female with the intended pitch accent patterns and presented acoustically to the children.
11
http://mc.manuscriptcentral.com/fla
First Language
Page 12 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
In Italian, the acoustic correlates of stress are mainly the duration, the intensity and the high pitch on the stressed syllable of the critical word (Avesani & Vayra, 2010). An acoustic analysis of our stimuli revealed that, in the conditions involving a prosodic manipulation, the emphasised word contained the highest pitch of the sentence with respect to the fundamental frequency of the utterance. We calculated the logarithmic difference between the highest F0 (which occurred on the stressed critical word) and the lowest F0 value in the utterance, measured in semitones (Snow & Balog, 2002). In the subject-stress condition, the stressed syllable of ‘SQUALO’ had a higher pitch with respect to the fundamental frequency of the sentence as compared to ‘shark’ in the neutral-
r Fo
stress, t(17) = 2.34, p<.04, and in the object-stress condition, t(17) = 2.33, p<.04. In the object-stress condition, the stressed syllable of ‘BALENA’ showed a higher pitch as compared to the
Pe
fundamental frequency of the utterance in the object-stress condition as compared to the neutralstress, t(17) = 10.81, p<.001, and the subject-stress, condition, t(17) = 17.45, p<.001.
er
Each set of sentences (including the preamble) was matched with two colour drawings. The first picture depicted three animal characters (e.g., a shark, a whale, a dolphin) and appeared
Re
together with the preamble (see Fig. 1). The second one presented the characters named in the
vi
sentence (i.e., a shark and a whale), and appeared together with the sentences of the same set (1-3) (see Fig. 2). That is, the first scene (depicted in Figure 1) provided the child with the relevant extra-
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
linguistic information that allowed the use of a marked prosodic structure in the experimental sentence (e.g., subject- and object-stress sentences). Note, nonetheless, that the presence of a third element in Fig. 1 did not affect the felicity of a canonical structure in the experimental sentence (e.g., neutral-stress). INSERT FIGURES 1 & 2 ABOUT HERE
12
http://mc.manuscriptcentral.com/fla
Page 13 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
We constructed three experimental lists such that each condition occurred six times in each list. The experimental items appeared in random order for each participant, to avoid any effect of order. The perceptual and prosodic salience of the experimental items was normed as indicated in Appendix B. As the dependent variable, we considered whether, in the sentence continuation, the object referent of the previous utterance (e.g., whale) appeared or not as the subject of a grammatical sentence (e.g., active, passive or intransitive); if so, we coded it as 1, otherwise as 0. Thus, we
r Fo
treated it as a categorical variable. We will refer to this variable as ‘object next mention’. Procedure and Scoring
Pe
The experiment was presented on a laptop computer using E-Prime software. The participants were tested in individual sessions, lasting approximately 15 minutes. The procedure
er
was as follows. First, children received training to get familiar with all the characters and the actions. Then, in three practice trials participants were trained to invent a continuation of an
Re
(intransitive) sentence, by producing a simple utterance. The experiment started when the experimenter pressed the spacebar. Afterwards, a picture showing three animals appeared on the
vi
screen, while participants heard the preamble (e.g., ‘These animals are swimming in the sea’). Next,
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
a picture appeared on the screen showing two characters (a shark and a whale) and children heard the experimental sentence (e.g., ‘And then the SHARK is chasing the whale’). Afterwards the recorded voice asked the child to invent a continuation of the sentence they had just heard. Sentence continuations were recorded and scored off line. All the continuations were coded by two naïve coders (percentage of agreement was 95%; kappa = .84). Continuations were qualified as ‘object next mentions’ depending on whether the
13
http://mc.manuscriptcentral.com/fla
First Language
Page 14 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
object (e.g., whale) of the previous utterance was expressed as the subject of an active (4-year-old: Ma poi la balena fa un grande schizzo e gli spruzza l’acqua; ‘But then the whale makes a big splash and soaks it’), passive, intransitive (i.e., 6-year-old: La balena scappa via; ‘The whale is swimming away’) or copular sentence (i.e., 4-year-old: La povera balena è triste; ‘The poor whale is sad’). Conversely, to be qualified as ‘subject next mentions’, sentences had to express in subject position, the subject referent of the previous sentence (e.g., shark). All other continuations that involved as subject an element that was not mentioned in the sentence (even though it appeared on the first picture), or both characters (i.e., 4-year-old: Poi però fanno pace; ‘But then they make it
r Fo
up’), or that was irrelevant (i.e., 5-year-old: Faceva freddo in acqua; ‘It was cold in the water’) were scored as Other. Most of the continuations expressed the subject as a definite NP.
Pe
Continuations involving an overt subject pronoun were disambiguated by gender marking on the pronoun (i.e., lui, lei). The sentence continuations that presented the (singular) subject expressed by
er
means of a null pronoun were coded as subject or object next mentions depending on whether the underlying subject could be easily inferred (i.e., La vuole mangiare; ‘Ø wants to eat it FEM-SING’;
Re
only eight continuations). Note that the limited use of null pronouns in an experimental task by Italian children is in sharp contrast with the production of null pronominal expression in
vi
spontaneous speech (nearly 60%-70% from 2 years of age; Guasti, 2002). However, in a previous
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
study (Serratrice, 2007), it was reported that for the monolingual Italian children null subjects accounted only for 20% of referent reintroductions in a narrative task, overt pronominal subjects made up 6%, while the largest share was represented by noun phrases (74%). As our data involved a categorical dependent variable (i.e., object next mention), we analyzed the data by means of a series of mixed-effects logit model (Jaeger, 2008), which provide greater statistical robustness than an ANOVA for such data (e.g., Blom & Baayen, 2012; Cherubini et al., 2013). In the analysis, prosodic emphasis and age (expressed as a categorical variable in the 14
http://mc.manuscriptcentral.com/fla
Page 15 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
cross-age comparison, i.e., 4-year-olds vs. 5-year-olds vs. 6-year-olds vs. adults) were introduced as potentially significant fixed effects. The participants and items were modeled as random-effect factors. Fixed effects inclusion in the model was evaluated on the basis of likelihood ratio tests; that is, we included an effect only if it significantly increased the model goodness of fit (Gelman & Hill, 2006; p. 549). To do so, we began with a full factorial model, which was progressively simplified by removing the predictor (i.e., prosodic emphasis or age group) that did not significantly contribute to the goodness of fit of the model. All the models were run in the statistical programming environment R (R Development Core Team, 2008). Fixed effects were computed using the
r Fo
languageR package (Baayen, 2012). Using the Means and Standard Deviations in each stress condition, we computed the Cohen’s d to determine effect size. Results
er
Pe
The proportions of object next mentions and standard deviations for each condition in all age groups is listed in Table 1. In addition, Figure 3 graphically contrasts the proportions of object
Re
next mentions in neutral-stress vs. subject-stress vs. object-stress conditions in all age groups. INSERT TABLES 1-2 AND FIGURE 3 ABOUT HERE
ew
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Cross-age Comparison
Out of a total of 216 responses, 4-year-olds produced 199 valid responses (92%), 5-yearolds 213 out of a total of 234 (91%), 6-year-olds 196 out of 216 (90%) and adults 135 (70%) out of a total of 144 responses. First, we compared the proportions of object next mentions across age groups in each condition by means of a mixed effects logit model with age groups (considered as a categorical variable), prosodic emphasis and their interaction as fixed factors and participants and items as random effects (see Table 2). 15
http://mc.manuscriptcentral.com/fla
First Language
Page 16 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Speakers tended to produce more object next mentions along with age (4-year-olds: 32%; 5year-olds: 53%, 6-year-olds: 60%; adults: 76%), and this was demonstrated by the fact that age groups had to be included in the final model as indicated by the likelihood ratio test, χ2(3) =15.02, p<.001. Similarly, the effect of prosodic emphasis did contribute significant fit to the model, χ2(2) =16.31, p<.001, and the interaction between Prosodic Emphasis and Age Groups too, χ2(6) =25.06, p<.001. The final model, reported in Table 2, revealed a number of significant interactions between age groups (as a categorical variable) and prosodic emphasis. Note that, as we predicted in the Introduction, the presence of significant interactions between age groups and prosodic emphasis,
r Fo
besides main effects, indicated that sensitivity to prosodic cues significantly changed along with age (Nieuwenhuis et al., 2011). In the object-stress and subject-stress conditions, 4-year-olds
Pe
consistently produced less object next mentions than 5-, 6-year-olds and adults. In addition, 4-yearolds were more likely to produce object next mentions after a neutral-stress sentence than after an
er
object-stress sentence as compared to 6-year-olds, that showed the opposite tendency (see Figure 3).
Re
Second, to test further the proposition that there is a continuous change in the sensitivity to
vi
prosodic emphasis, we have also conducted an analysis including age (in month) as a continuous variable within each age group. Again, in this case age contributed significant information to the
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
model and significantly predicted the tendency to produce more object next mention after an objectstress sentence, χ2(1) =14.02, p<.001; B=.004; SE=.001; Wald Z=3.96, p<.001, indicating that the tendency to produce object next mentions increased with development. Analyses by Age-group Then, we submitted the data of each age group to a series of mixed effects logit models. As for fixed factors, prosodic emphasis (neutral-stress, subject-stress, object-stress) did not contribute 16
http://mc.manuscriptcentral.com/fla
Page 17 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
to the model’s fit χ2(2) = .70, p= .70, in the 4-year-olds group, whereas it did in the 5-year-olds χ2(2) = 8.86, p<.011, 6-year-olds χ2(2) = 24.71, p<.001, and adults χ2(2) = 6.05, p<.04 groups. As for the random effects, in this and all the subsequent models, only the inclusion of the byparticipants and by-items random intercepts contributed to the fit of the model. The 4-year-olds produced 10% less object next mentions after an object-stress sentence in comparison with a subject-stress one and 2% more object next mentions following a neutral-stress sentence as compared to a subject-stress one (see first column of Table 1). Even though in the 4-
r Fo
year-olds group prosodic emphasis did not contribute to the model fit, we nevertheless ran a mixed logit model: Proportions of object next mentions did not significantly differ across conditions. That is, at four years of age, prosodic emphasis did not predict which referent was mentioned next in the
Pe
sentence continuation, and did not provide significant information to explain the data.
er
The 5-year-olds produced 23% more object next mentions following an object-stress sentence, in comparison with a subject-stress sentence, and 7% more object next mentions after a
Re
neutral-stress sentence in comparison with a subject-stress one (see second column of Table 1). That is, the object-stress condition elicited the greatest proportion of object next mentions, the
vi
subject-stress condition the least, whereas proportions of object next mentions in the neutral-stress
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
fell midway between the other two conditions. The mixed logit model revealed that object-stress sentences enhanced the production of object next mentions to a significantly greater extent than subject-stress and neutral-stress ones. In the neutral-stress condition there were more object next mentions than in the subject-stress condition, but this difference was not significant (see Table 2). To sum up, in the 5-year-olds group, prosodic emphasis influenced the tendency to mention next the referent that received prosodic emphasis in the previous utterance.
17
http://mc.manuscriptcentral.com/fla
First Language
Page 18 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
In the 6-year-olds group, there were 27% more object next mentions after an object-stress sentence as compared to a subject-stress one (see Table 1, third column). In contrast to the 5-yearolds group, there were 13% more object next mentions in the subject-stress condition in comparison with the neutral-stress condition. The mixed logit model indicated that proportions of object next mentions in the object-stress condition were significantly higher than in the subject-stress condition (see Table 2). In addition, in the neutral-stress condition there were significantly less object next mentions than in the object-stress condition, whereas the difference between the neutral-stress and the subject-stress conditions was marginally significant (p=.06). Thus, at six years of age prosodic
r Fo
emphasis significantly predicted the likelihood to mention next, as the subject of the sentence continuation, the referent that received prosodic emphasis.
Pe
Adults produced 25% more object next mentions after an object-stress sentence as compared to a subject-stress one. In addition, there were 8% more object next mentions in the neutral-stress in
er
comparison with the subject-stress condition (see fourth column of Table 1). Again, the mixed logit model revealed that object-stress sentences enhanced the production of object next mentions
Re
significantly more than subject-stress sentences (see Table 2). In addition, object-stress sentences
vi
promoted the production of object next mentions to a greater extent than neutral-stress sentences, but the difference was only marginally significant (p=.07).
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
General Discussion
The present study tested the prediction that the referent that receives prosodic emphasis is more salient and thus more likely to be mentioned next, as the subject of a subsequent utterance. Therefore, if a speaker is aware of the fact that prosodic emphasis signals which referent is more salient, s/he should more likely place it in the most prominent position, i.e., subject, in a subsequent utterance. We tested this claim on 4-, 5-, 6-year-olds and adult speakers. Crucially, the manipulation 18
http://mc.manuscriptcentral.com/fla
Page 19 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
of prosodic emphasis had no effect on the continuations of 4-year-olds. Conversely, in 5-, 6-yearolds and adults, the placement of prosodic emphasis on a referent significantly enhanced the likelihood to make that referent the subject of the sentence continuation. Hence, the basic pattern which holds for the 5-year-olds and the adults is that there are more object next mentions in the object-stress condition than in either the subject-stress and the neutral-stress conditions. Additionally, the cross-age comparison revealed that the number of object next mentions increased along with age.
r Fo
In reply to the experimental questions that motivated the current study, our results indicated that 4-year-olds mostly relied on structural cues (syntactic and/or linear position), to determine which referent was more salient, and produced overall more subject next mentions. 5-year-olds
Pe
appeared to rely more on prosodic cues in comparison with younger children. Such findings support the argument that at five, children appeared to master to some extent the ability to incorporate
er
sensitivity of prosodic cues in their mental model, in order to produce a non-default continuation, i.e., an object next mention, when it was more appropriate. Thus, 5- and 6-year-olds appeared to
Re
have an understanding that prosodic emphasis increases the salience of a referent and, most
vi
importantly, they were able to use this information to plan their own subsequent production.
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Our results are in line with previous studies, indicating that the ability to rely effectively on prosodic cues in assigning an interpretation to a sentence is not fully mastered before five years of age (e.g., Zhou et al., 2012; ; Sekerina & Trueswell, 2012). The existing literature involving the use of off-line tasks, indicates a later development in understanding prosodic cues, not evident before five years of age (e.g., McDaniel & Maxfield, 1992; Szendröi, 2003; Wells et al., 2004; Gualmini, 2002). Studies that involved both on-line (i.e., eye-tracking) and off-line techniques (i.e., judgement tasks) offered interesting insights about the fact that children are sensitive to prosodic cues during 19
http://mc.manuscriptcentral.com/fla
First Language
Page 20 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
the on-line processing of the sentence, but they are not able to use them effectively when building up the final interpretation of a sentence (e.g., Zhou et al., 2012). In all these studies, the pattern of results indicates that the discrepancy between the off-line task data (that was poor in children as compared to adults) and the eye-movements pattern (that was similar in the two groups) indicated that up to four years of age, children do process the information about prosody in the moment-bymoment elaboration of the sentence, but they lack the ability to integrate the prosodic cues to assign an interpretation to the sentence.
r Fo
With respect to the previous studies, our experiment extended the existing literature offering insights about the cues that are processed by children when producing connected utterances, i.e., segments of a discourse. Recall that our study moved from the assumption that prosodic emphasis
Pe
signals the most salient referent in a sentence. Then, if children rely on this information when planning their own subsequent production, they should make the most prosodically prominent
er
referent in a sentence the subject of the following utterance. Therefore, our task might be considered an off-line measure of children sensitivity to a prosodic cue, namely prosodic emphasis.
Re
The sentence continuation task appeared to be an effective test to investigate children’s sensitivity
vi
to such prosodic cue. First, it enabled us to see which was the interpretation children gave to the sentence, in term of which element they felt as more salient. Second, it provided information about
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
the processes underlying sentence production and specifically those that are involved in the choice of the topic (i.e., the subject) of a subsequent utterance. Our findings therefore indicate that child’s reliance on prosodic cues increases later along the course of development with respect to structural cues (i.e., linear order and/or grammatical assignment). An additional difference in comparison with previous research results refers to the type of prosodic cues that has been manipulated here. Most of the studies we have mentioned involved the 20
http://mc.manuscriptcentral.com/fla
Page 21 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
manipulation of pitch accent on a focus operator particle (i.e., Gualmini et al., 2004; Zhou et al., 2012), of narrow focus (i.e., Chen, 2010) or contrastive pitch (e.g., Ito et al., 2008). For instance, in Ito et al. (2008) the preceding sentence explicitly implied a contrast set (i.e., ‘First hang the green ball. Now hang the BLUE ball’), whereas in our stimuli only the visual context (i.e., Fig. 1) explicitly evoked a contrast set (showing three characters out of two). Hence, the prosodic manipulation we have used here cannot be accounted as contrastive pitch (and for this reason we did refer to our manipulation as ‘prosodic emphasis’). We deliberately avoided a verbal expression that evoked the contrast set (as it would have been in 'The shark is not washing the giraffe, the shark
r Fo
is washing the WHALE'), in order to keep the experimental sentences as simple as possible in all the conditions, avoiding (more complex) negative sentences, hard to process not only for children
Pe
(Morris, 2003), but also for adults (Kaup et al., 2006, 2007; Tian et al., 2010). Therefore, we opted for a more basic manipulation that we could control quantitatively (i.e., the most salient element
er
contained the highest pitch of the sentence with respect to the fundamental frequency of the utterance).
Re
Another issue about our data that remains to be addressed is the greater proportion of object
vi
next mentions that we found in adults’ continuations (62%) in contrast to children (36%, 36% and 38% in all children age-groups). We observed that several verbs in our experimental sentences (e.g.,
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
scold, scratch, push, praise, catch, chase, etc.) prompted a ‘because’ continuation (e.g., McKoon et al., 1993). One possibility is that these verbs exhibited a particular semantic strong cue, known as Implicit Causality (Garvey & Caramazza, 1974). Implicit causality cue creates an expectation for an upcoming explanation that focuses on the object (as in ‘John is scolding Mary. She was late’), and involves a next-mention bias to the object (Pickering & Majid, 2007). Such a finding is in line with the claim that sensitivity to semantic aspects of the language is acquired later in the course of development (Gleitman, 1990). 21
http://mc.manuscriptcentral.com/fla
First Language
Page 22 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Overall, our findings have an important implication for child language production theories. Under a well-accepted view of (adult) language production (Levelt, 1999), during utterance planning speakers integrate all non-syntactic (e.g., semantic, discourse, prosodic) cues in a preverbal message in order to determine the most salient concept that will undergo syntactic encoding first. One might suggest that adults are able to integrate them in their speech plan, whereas in children this ability appears to be still lacking at four and progressively mastered from five years of age on. More generally, our findings are compatible with the view that, when planning a sentence, children access the same mechanisms as adults, but integrate specific cues (semantic,
r Fo
prosodic, etc.) to a different extent (Trueswell et al., 1999). Even though our study focuses on micro aspects of syntax and prosody, recall that the
Pe
cognitive procedure underlying our task involves integrating linguistic information (prosodic emphasis) in a speech plan across utterances, and this ability is entailed in a broader linguistic
er
competence, namely, communicative and narrative development. In line with our findings, recent research has found that specific linguistic skills (i.e., reference adequacy, ambiguity detection, etc.)
Re
are associated with different developmental stages. For example, a study on the development of
vi
narrative structure in Finnish children aged between four and eight revealed significant differences between four and five year olds in productivity and between five and six in referential cohesion
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
(Makinen et al., 2014). Similarly, Colozzo and Whitely (2014) observed a change between kindergarten children (up to 5 years of age) and first and second graders (6 and 7 years of age) in referential adequacy. In their study, these authors observed that Canadian English kindergartners were not as able at maintaining reference to characters using pronominals or at switching reference using either nominals or pronomimals compared to children in the two higher grades. Additional evidence extends those findings demonstrating that a reciprocal relation exists between cognitive skills development and linguistic competence. For instance, cognitive flexibility has been reported 22
http://mc.manuscriptcentral.com/fla
Page 23 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
to significantly predict development of communicative abilities, such as ambiguity detection only in preschoolers’, but not in school-age children (Gillis & Nilsen, 2014). To sum up, these shifts in linguistic development could reflect changes in cognitive resources capacity that occur between four and six years of age (Whitely & Colozzo, 2013; Orsolini et al., 1996). In future research, we propose to replicate this study using an on-line methodology. This could be done by setting up a visual-word experiment, where eye-movements on the three referents could be collected while the critical sentence is spoken. By doing so, we would be able to collect
r Fo
not only off-line measures, provided by the continuation task, but on-line data on the moment-bymoment construction of the interpretation. In this way, more specific conclusions could be made about the (on-line) process of integrating prosodic cues in the linguistic representation.
Pe
In conclusion, our study provides an interesting insight into the ability of children to
er
interpret and rely prosodic cues in order to choose which referent to mention as the subject of an upcoming utterance. Such results are in line with a well accepted view under which sensitivity to
Re
prosodic cues is not yet mastered at four, and crucially with the notion that the ability of relying on prosodic cues develops significantly between four and six years of age.
ew
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
23
http://mc.manuscriptcentral.com/fla
First Language
Page 24 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Author notes We would like to thank Marina Nespor, Marianella Carminati and Mirco Fasolo for their insightful comments. This work has been supported by a FIRB grant ‘La ricerca fondamentale sul linguaggio al servizio della lingua italiana: documentazione, acquisizione monolingue, bilingue e L2, e ideazione di prodotti multimediali.’ Fundamental research on language in the service of the Italian language: documentation, monolingual, bilingual and L2 acquisition, and the conception of multimedia products. - FIRB Project (2008).
r Fo er
Pe ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
24
http://mc.manuscriptcentral.com/fla
Page 25 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Appendix A 1. Questi animali giocano in piscina. E poi la zebra bagna l’ippopotamo. (These animals are playing in the swimming pool. Then the zebra is soaking the hippo.) 2. Questi animali hanno caldo. E poi il coccodrillo spruzza la giraffa. (These animals feel hot. Then the crocodile is splashing the giraffe.) 3. Questi animali fanno amicizia alla festa. E poi l’orsacchiotto lecca la marmotta.
r Fo
(These animals met each other at the party. Then the little bear is licking the groundhog.) 4. Questi uccelli litigano per il cibo. E poi il falco graffia l’aquila.
Pe
(These animals are arguing for food. Then the hawk is scratching the eagle.)
er
5. Questi animali fanno conoscenza alla fattoria. E poi l’asino annusa la mucca.
Re
(These animals met each other in the farm. Then the donkey is sniffing the cow.) 6. Questi animali fanno il bagno al polo. E poi il pinguino asciuga la foca.
vi
(These animals are bathing in the pole. Then the penguin is drying the seal.)
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
7. Questi animali hanno litigato a scuola. E poi la lumaca sgrida il passerotto. (These animals argued at school. Then the snail is scolding the sparrow.) 8. Questi pesci nuotano nel mare. E poi lo squalo insegue la balena. (These fishes are swimming in the sea. Then the shark is chasing the whale.) 9. Questi animali si fanno gli auguri a Natale. E poi la gallina abbraccia il gufo. 25
http://mc.manuscriptcentral.com/fla
First Language
Page 26 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
(These animals are cheering for Christmas. Then the chicken is hugging the howl.) 10. Questi animali si fanno i dispetti. E poi la capra tira il porcellino. (These animals are playing tricks on each other. Then the goat is pulling the little pig.) 11. Questi insetti non stanno mai fermi al parco. E poi la mosca acchiappa il bruco. (These insects never rest in the park. Then the fly is catching the caterpillar.) 12. Questi animali litigano. E poi la gallina punge il pavone.
r Fo
(These animals are arguing. Then the hen is pricking the peacock.) 13. Questi animali giocano col fango. E poi la scimmia sporca il pappagallo.
Pe
(These animals are playing with mud. Then the monkey is dirtying the parrot.)
er
14. Questi animali giocano a moscacieca. E poi la rana cattura il criceto.
Re
(These animals are playing blind man's bluff. Then the frog captures the hamster.) 15. Questi animali si allenano per la corsa. E poi l'oca rincorre il gallo.
vi
(These animals are training for the run. Then the goose is chasing the cock.)
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
16. Questi animali recitano delle poesie alla festa. E poi il ghiro applaude la talpa. (These animals are reciting poems at the party. Then the dormouse is praising the mole.) 17. Questi animali si conoscono alla festa. E poi la tartaruga fotografa il topolino. (These animals know each other at the party. Then the turtle is taking a picture of the little mouse.) 26
http://mc.manuscriptcentral.com/fla
Page 27 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
18. Questi animali si incontrano alla festa. E poi la cicala abbraccia il grillo. (These animals meet each other at the party. Then the cicada is hugging the cricket.)
r Fo er
Pe ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
27
http://mc.manuscriptcentral.com/fla
First Language
Page 28 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Appendix B Experimental items were normed with respect to prosodic salience of the referents and perceptual salience of the visual elements in the pictures. First, we conducted a norming study in three adult participants (M = 24;5 years) that did not participate to the experiment, had to indicate which referent in a sentence (e.g., squalo and balena) they perceived as more prosodically emphasized. The 18 experimental sentences were presented in the neutral-stress, in the subjectstress, or in the object-stress condition. The object referent was indicated as the more prosodically
r Fo
emphasised referent significantly more often in the object-stress condition (78%) in comparison with the neutral-stress (35%) and with the subject-stress (22%) condition. We performed a mixed effects logit model with prosodic emphasis as fixed factor and participants and items as random
Pe
factors. Interestingly, the perceived emphasis on the subject referent in the subject-stress condition was stronger than in the neutral-stress, though not significantly (p= .09).
er
Second, experimental pictures were selected with respect to a perceptual salience bias for
Re
characters. In a control experiment, eight children (one male; M = 4;7), that did not participate to the current experiment, saw a total of forty pictures depicting two animal characters (e.g., squalo or
vi
balena), appearing in random order on a computer screen. Children were asked to name the
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
character they could identify first. Then we compared proportions of mentions for each character towards the other one. In our experimental set, we included only those pictures where both characters were indicated by children as equally perceptually accessible. All words employed in the experiment were drawn from the Lessico Elementare (Elementary Lexicon) (Marconi et al., 1993). We used the total frequency usage of each word and included only words with 2 digits of frequency, ranging from 11.4 (hippo) to 96.16 (cow).
28
http://mc.manuscriptcentral.com/fla
Page 29 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
References Avesani, C., & Vayra, M. (2000). Costruzioni marcate e non marcate in Italiano. Il ruolo dell’intonazione. In D. Locchi (Ed.) Atti delle X giornate di Studio del GFS (pp. 1-14). Napoli: Il Torcoliere. Bard, E., Robertson, D., & Sorace, A. (1996). Magnitude estimation of linguistic acceptability. Language, 72, 32-68. Bassano, D., Korecky-Kröll, K., Maillochon, I., van Dijk, M., Laaha, S., Van Geert, P. &
r Fo
Dressler, W. U. (2013). Prosody and animacy in the development of noun determiner use: a crosslinguistic approach. First Language, 33(5), 476 –503.
Pe
Blom, E., & Baayen, R.H. (2012). The impact of verb form, sentence position, home language and L2 proficiency on subject-verb agreement in child L2 Dutch. Applied Psycholinguistics, 1, 1-35.
er
Re
Bock, J.K., & Warren, R.K. (1985). Conceptual accessibility and syntactic structure in sentence formulation. Cognition, 21, 47-67.
vi
Brennan, S.E. (1995). Centering attention in discourse. Language and Cognitive Processes,
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
10, 137-167.
Chen, A.J. (2010). Is there really an asymmetry in the acquisition of the focus-toaccentuation mapping? Lingua, 120, 1926-1939. Chen, A.J. (2011). Tuning information packaging: intonational realization of topic and focus in child Dutch. Journal of Child Language, 38(5), 1055-1083.
29
http://mc.manuscriptcentral.com/fla
First Language
Page 30 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Cherubini, P., Rusconi, P., Russo, S. and Crippa, F. (2013). Missing the dog that failed to bark in the nighttime. On the overestimation of occurrences over non-occurrences in hypothesis testing. Psychological Research, 77, 348–370. Chomsky, N., & Halle, M. (1968). The sound pattern of English. Cambridge, USA: MIT Press. Cinque, G. (1993). A null theory of phrase and compound stress. Linguistic Inquiry, 24, 239-298.
r Fo
Colozzo, P., & Whitely, C. (2014). Keeping track of characters: factors affecting referential adequacy in children’s narratives. First Language, 34(2), 155 –177.
Pe
Costa, J., & Szendroi, K. (2006). Acquisition of focus marking in European Portuguese – evidence for a unified approach to focus. In Torrens, V., Escobar, L. (Eds.). The acquisition of
er
syntax in Romance languages ( pp.319-329). Amsterdam: John Benjamins.
Re
Cutler, A., & Swinney, D. A. (1987). Prosody and the development of comprehension. Journal of Child Language, 14, 145–167.
vi
De Ruiter, L.E. (2009). The prosodic marking of topical referents in the German "Vorfeld" by children and adults. Linguistic Review, 26, 329-354.
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Garvey, C., & Caramazza, A. (1974). Implicit causality in verbs. Linguistic Inquiry, 5, 459– 464. Gelman, A., & Hill, J. (2006). Data analysis using regression and multilevel/hierarchical models. Cambridge, UK: Cambridge University Press.
30
http://mc.manuscriptcentral.com/fla
Page 31 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Gennari, S., Gualmini, A., Crain, S., Meroni, L., & Maciukaite, S. (2001). How adults and children manage stress in ambiguous contexts. Proceedings of 1st Workshop on Cognitive models of semantic processing, University of Edinburgh. Gillis, R., & Nilsen, E.S., (2014). The role of cognitive flexibility in children's ability to detect communicative ambiguity. First Language, 34, 58 – 71. Gleitman, L.R. (1990).The structural source of verb meaning. Language Acquisition, 1, 355.
r Fo
Grassmann, S., & Tomasello, M. (2010). Prosodic stress on a word directs 24-month-olds' attention to a contextually new referent. Journal of Pragmatics, 42, 3098-3105.
Pe
Grosz, B.J., & Sidner, C.L. (1986). Attention, intentions, and the structure of discourse. Computational Linguistics, 12, 175-204.
er
Grosz, B.J., Joshi, A.K., & Weinstein, S. (1983). Providing a unified account of definite noun phrases in discourse. Proceedings of the 21st Annual Meeting of the Association for
Re
Computational Linguistics, Cambridge, Massachusetts.
Gualmini, A. (2004). Some knowledge children don't lack. Linguistics, 42, 957-982.
ew
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Gualmini, A., Maciukaite, S., & Crain, S. (2002). Children's insensitivity to contrastive stress in sentences with ONLY. Proceedings of the 26th Annual Penn Linguistics Colloquium, University of Pennsylvania. Guasti, M. (2002). Language acquisition. The growth of grammar. Cambridge, Massachusetts: MIT Press.
31
http://mc.manuscriptcentral.com/fla
First Language
Page 32 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Höhle, B., Berger, F., Müller, A., Schmitz, M., & Weissenborn, J. (2009). Focus particles in children’s language: Production and comprehension of auch (also) in German learners from 12 months to 4 years. Language Acquisition, 16, 36–66. Hornby, P. A., & Hass, W. A. (1970). Use of contrastive stress by preschool children. Journal of Speech and Hearing Research, 13(2), 395-399. Ito, K., & Speer, S. R. (2008). Anticipatory effects of intonation: eye movements during instructed visual search. Journal of Memory and Language, 58, 541-573.
r Fo
Jaeger, T. F. (2008). Categorical data analysis: away from ANOVAs (transformation or not) and towards logit mixed models. Journal of Memory and Language, 59, 434–446. Kaup, B., Lüdtke, J., & Zwaan, R.A. (2006). Processing negated sentences with
Pe
contradictory predicates: Is a door that is not open mentally closed? Journal of Pragmatics, 38,
er
1033 – 1050.
Kaup, B., Yaxley, R. H., Madden, C. J., Zwaan, R. A., & Lüdtke, J. (2007). Experiential
Re
simulations of negated text information. Quarterly Journal of Experimental Psychology, 60, 976– 990.
ew
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Keller, F., & Alexopoulou, T. (2001). Phonology competes with syntax: experimental evidence for the interaction of word order and accent placement in the realization of Information Structure. Cognition, 79, 301-372. Levelt, W.J.M. (1989). Speaking: from intention to articulation. Cambridge, Massachusetts: The MIT Press.
32
http://mc.manuscriptcentral.com/fla
Page 33 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Mäkinen, L., Loukusa, S., Nieminen, L., Leinonen, E., & Kunnari, S. (2014). The development of narrative productivity, syntactic complexity, event content, and referential cohesion in four to eight-year-old Finnish children. First Language, 34, 24–42. Marconi, L., Ott, M., Pesenti, E., Ratti, D., & Tavella, M. (1993). Lessico elementare. Dati statistici sull’ italiano scritto e letto dai bambini delle elementari [Elementary lexicon: Statistical data on Italian as written and read by elementary school children]. Bologna: Zanichelli. McDaniel, D., & Maxfield, T. (1992). Principle B and contrastive stress. Language
r Fo
Acquisition, 2, 337-358.
McKoon, G., Ratcliff, R., Ward, G., & Sproat, R. (1993). Syntactic prominence effects on
Pe
discourse processing. Journal of Memory and Language, 32, 593-607. Mckoon, G., Ratcliff, R., Ward, G., & Sproat, R. (1993). Syntactic prominence effects on
er
discourse processes. Journal of Memory and Language, 32(5), 593-607.
Re
Morris, B. J. (2003). Opposites attract: the role of predicate dimensionality in preschool children’s processing of negations. Journal of Child Language, 30, 419–440.
vi
Nieuwenhuis, S., Forstmann, B.U., & Wagenmakers, E.J. (2011). Erroneous analyses of
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
interactions in neuroscience: a problem of significance. Nature Neuroscience, 14, 1105-1107. Orsolini, M., Rossi, F., & Pontecorvo, C. (1996). Re-introduction of referents in Italian children’s narratives. Journal of Child Language, 23, 465–486. Pickering, M. J., & Majid, A. (2007). What are implicit causality and consequentiality? Language and Cognitive Processes, 22, 780-788.
33
http://mc.manuscriptcentral.com/fla
First Language
Page 34 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Prat-Sala, M., & Branigan, H. (2000). Discourse constraints on syntactic processing in language production : a cross-linguistic study in English and Spanish. Journal of Memory and Language, 42, 168-182. Sekerina, I. A., & Trueswell, J.C. (2012). Interactive processing of contrastive expressions by Russian children. First Language, 32, 63-87. Serratrice, L. (2007). Referential cohesion in the narratives of bilingual English-Italian children and monolingual peers. Journal of Pragmatics, 39( 6), 1058-1087.
r Fo
Tannenbaum, P. H., & Williams, F. (1968). Generation of active and passive sentences as a function of subject and object focus. Journal of Verbal Learning and Verbal Behavior, 7, 246–250.
Pe
Tian, Y., Breheny, R. & Ferguson, H.J. (2010). Rapid communication why we simulate negated information: a dynamic pragmatic account. Quarterly Journal of Experimental Psychology, 63 (12), 2305 – 2312.
er Re
Trueswell, J. C., Sekerina, I., Hill, N. M., & Logrip, M.L. (1999). The kindergarten-path effect: studying on-line sentence processing in young children. Cognition, 73, 89–134.
vi
Vallduví, E. (1992). The information component. New York: Garland.
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Wells, B., Peppé, S., & Goulandris, A. (2004). Intonation development from five to thirteen. Journal of Child Language, 31, 749-778.
Whitely, C., & Colozzo, P. (2013). Who’s who? Memory updating and character reference in children’s narratives. Journal of Speech, Language, and Hearing Research, 56, 1625–1636. Zhou, P., Su, Y., Crain, S., Gao, L. Q., & Zhan, L. K. (2012). Children’s use of phonological information in ambiguity resolution: A view from Mandarin Chinese. Journal of Child Language, 39, 687-730. 34
http://mc.manuscriptcentral.com/fla
Page 35 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
r Fo er
Pe ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
35
http://mc.manuscriptcentral.com/fla
First Language
Page 36 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Tables Table 1 Proportions of object next mentions (out of all the continuations that involved as subject either the object or the subject referent of the previous utterance) and Standard Deviations for each condition and age group.
r Fo 4-year-olds
Age-group 5-year-olds
Pe
6-year-olds
Adults
M
SD
M
SD
M
SD
M
SD
Neutral-stress
.44
.50
.37
.48
.20
.41
.59
.50
Subject-stress
.42
.49
.30
.46
.33
.47
.51
.51
Object-stress
.32
.46
.60
.49
.76
.43
er .53
Re .50
ew
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
36
http://mc.manuscriptcentral.com/fla
Page 37 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Table 2 Summary of the fixed effects in the mixed logit models for the cross-age comparison (N = 739, log-likelihood = -470.9), 4-year-olds (N =199, log-likelihood = -124.6), 5-year-olds (N =213, log-likelihood = -129.0), 6-year-olds (N = 193, log-likelihood = -114.6), and adults (N = 134, loglikelihood = -83.55). In all the subsequent models, proportions of object next mentions were compared across all conditions by selecting as reference category (i.e., the category against which all other categories are compared): i) neutral-stress, ii) subject-stress condition. In all the models,
r Fo
the sign of the coefficient for the fixed factor of sentence stress assumes a positive value, denoting that the odds for an object next mention become larger when the sentence stress is on the object. Cross-age comparison
Predictor
Pe
B
SE
Wald Z
p
.27
-1.19
.23
er
Reference category: Age = 4-year-olds
Re
Reference category: Prosodic Emphasis = Subject
vi
(Intercept)
-.32
ew
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Prosodic Emphasis = Object
-.45
.36
-1.25
.20
Prosodic Emphasis = Neutral
.06
.35
.17
.86
Age = 5-year-olds
-.52
.39
-1.33
.18
Age = 6-year-olds
-.39
.39
-1.02
.31
Age = Adults
.38
.42
.90
.36
37
http://mc.manuscriptcentral.com/fla
First Language
Page 38 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Prosodic Emphasis = Object: Age = 5-year-olds
1.43
.52
2.80
.004
Prosodic Emphasis = Neutral: Age = 5-year-olds
.22
.50
.45
.65
Prosodic Emphasis = Object: Age = 6-year-olds
1.61
.52
3.09
.002
Prosodic Emphasis = Neutral: Age = 6-year-olds
-.82
.54
-1.52
.12
Prosodic Emphasis = Object: Age = adults
1.58
.59
2.69
.007
Prosodic Emphasis = Neutral : Age = adults
.25
.55
.46
.64
Prosodic Emphasis = Object: Age = 5-year-olds
1.21
.50
2.40
.01
Prosodic Emphasis = Object: Age = 6-year-olds
2.43
.55
4.40
.001
1.33
.59
2.22
.02
r Fo
Reference category: Prosodic Emphasis = Neutral
er
Pe
Prosodic Emphasis = Object: Age = adults
Re
Note. Random effects for subject and item had SD of .37 and .17 respectively. ________________________________________________________________________________ B
SE
ew
Predictor
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Wald Z
p
4-year-olds
Cohen’s d
Reference category: Neutral-stress (Intercept)
-.45
.38
-1.17
.24
Prosodic Emphasis = Subject
-.26
.41
-.63
.53
.04
Prosodic Emphasis = Object
-.33
.41
-.80
.42
.25 38
http://mc.manuscriptcentral.com/fla
Page 39 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Reference category: Subject-stress Prosodic Emphasis = Object
-.07
.42
-.17
.85
.21
Note. Random effects for subjects and items had SD of 1.14 and .28 respectively. 5-year-olds Reference category: Neutral-stress (Intercept)
-.62
.36
-1.70
.08
Prosodic Emphasis = Subject
-.37
.40
-.91
.36
.14
Prosodic Emphasis = Object
.79
.38
2.05
.04
.32
.41
2.83
.004
.47
r Fo
Pe
Reference category: Subject-stress Prosodic Emphasis = Object
er 1.16
Re
Note. Random effects for subjects and items had SD of 1.19 and .003 respectively. 6-year-olds Reference category: Neutral-stress (Intercept)
-1.55
Prosodic Emphasis = Subject Prosodic Emphasis = Object
ew
vi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
.37
-4.10
.001
.78
.42
1.85
.06
.29
2.00
.45
4.38
.001
.88
1.22
.41
2.98
.002
.56
Reference category: Subject-stress Prosodic Emphasis = Object
39
http://mc.manuscriptcentral.com/fla
First Language
Page 40 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Note. Random effects for subjects and items had SD of .34 and .48 respectively. Adults Reference category: Neutral-stress (Intercept)
.41
.42
.98
.32
Prosodic Emphasis = Subject
-.32
.46
-.69
.48
.16
Prosodic Emphasis = Object
.86
.49
1.75
.07
.36
1.17
.48
2.42
.01
.53
r Fo
Reference category: Subject-stress Prosodic Emphasis = Object
Pe
Note. Random effects for subject and item had SD of .65 and .64 respectively.
er ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
40
http://mc.manuscriptcentral.com/fla
Page 41 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Figures Figure 1. Example of experimental picture.
r Fo er
Pe Note: This illustration was presented with the preamble (in Italian) ‘These fishes are swimming in the sea’.
ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
41
http://mc.manuscriptcentral.com/fla
First Language
Page 42 of 43
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Figure 2. Example of experimental picture.
r Fo
Note: This picture was presented in association with the set of experimental sentences presented in 1-3.
er
Pe ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
42
http://mc.manuscriptcentral.com/fla
Page 43 of 43
First Language
Running head: PROSODIC CUES AND SENTENCE CONTINUATION
Figure 3. Proportions of object next mentions in neutral-stress, subject-stress and object-stress conditions in all age groups. Error bars refer to the Standard Error of the Mean.
r Fo er
Pe ew
vi
Re
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
43
http://mc.manuscriptcentral.com/fla
