The effects of prosody on idiom comprehension

Effects of prosody on idiom processing

1

Running Head: PROSODIC EFFECTS ON IDIOM COMPREHENSION

“That took a load OFF my mind!” – The Effects of Prosody on Idiom Comprehension

Eva Smolka University of Konstanz, Germany

Corresponding Author: Eva Smolka Department of Linguistics University of Konstanz Universitätsstraße 10 78457 Konstanz Germany Email: [email protected] Tel: +49 7531 88-4834 Fax: +49 7531 88-4898


2

Abstract This study tested whether contrastive accent affects the comprehension of idioms. In Experiment 1, participants heard literal and idiomatic sentences in neutral (broad focus), contrastive (narrow focus on the preposition), or monotonous (flat pitch) intonation and made speeded acceptability judgments whether the sentences sounded natural. Only sentences in neutral prosody were acceptable, while contrastive accents were just as unacceptable as monotonous recordings, both for literal and idiomatic sentences. Experiment 2 applied crossmodal associative priming. Participants heard literal and idiomatic sentence primes in neutral and contrastive sentence prosody and made lexical decisions about visual targets that were strongly associated with the figurative meaning. Idiomatic sentences primed their meaning associates relative to literal sentences, but only in neutral prosody. These findings suggest an inhibitory effect of contrastive accents on idiom comprehension and the activation of the figurative meaning. The paper discusses the relevance of prosodic information in present models of idiom processing.

Keywords: idioms, idiom processing, idiomatic meaning, figurative meaning, lexical access, mental lexicon, contrast, intonation, prosody, accent


3

“That took a load OFF my mind!” – The Effects of Prosody on Idiom Comprehension Figurative language is challenging for linguistic and psycholinguistic theory, since the figurative meaning is not derived compositionally from the meaning of its parts and their syntactic assembly. For instance, the figurative meaning of the idiom “She hit the road” cannot be derived by composing the meaning of the individual words (she, hit, the, road) and their syntactic combination (‘an agent hitting some object’) as would be the case in She hit the guitar, despite the parallel syntactic structure. Idioms are considered semantically fixed as the figurative meaning does not allow the replacement of any of the word constituents (e.g., *he kicked the pail; *he pushed the bucket). Furthermore, idioms are considered syntactically fixed, because there are restrictions regarding the syntactic transformations that idiomatic sentences may undergo (e.g., *the road was hit by her; *she hit the long road; *she strongly hit the road). Hence, linguistic and psycholinguistic research are puzzled by the question of how idiomatic meaning is processed and stored in lexical memory (Cacciari & Glucksberg, 1994; Gibbs Jr, 2002; Gibbs, 1994; Swinney & Cutler, 1979; for a review see Titone & Connine, 1999). Seminal studies found that idioms are processed faster than literal sentences and argued for a ‘non-compositional’ representation in which the whole idiomatic meaning of a fixed phrase is stored as a distinct entry in the mental lexicon, similar to the representation of a complex word such as Wachstumsbeschleunigungsgesetz (growth acceleration act) (Bobrow & Bell, 1973; Gibbs, 1980; Swinney & Cutler, 1979). In contrast, ‘compositional’ approaches hold that the meaning of an idiomatic phrase is assembled on the fly from the individual parts (Cacciari & Tabossi, 1988; Caillies & Butcher, 2007; Cutting & Bock, 1997; Gibbs & Nayak, 1989; Holsinger & Kaiser, 2013; Rabanus, Smolka, Streb, & Rösler, 2008; Smolka, Rabanus, & Rösler, 2007; Sprenger, Levelt, & Kempen, 2006). All compositional approaches have in common that the constituents are first processed individually, similar to


4

literal sentences, until the figurative meaning is activated, either by the joint activation of the idiomatic constituents (Rabanus, et al., 2008; Smolka, et al., 2007) or when the ‘idiom key’ – or some kind of ‘unitary entry’ that refers to the idiomatic concept – is reached and activated (for perception see Cacciari & Tabossi, 1988; for production see Cutting & Bock, 1997; Sprenger, et al., 2006). These latter ‘hybrid’ approaches assume that idioms are both compositional (i.e. they are composed of the single word lemmas of the constituents) and unitary (in that each idiom possesses its own lexical entry that stores the whole meaning of the idiom as a distinct entry). For example, Cutting and Bock (1997) concluded that idioms possess a distinct ‘lexical concept node’ that is activated by the idiomatic concept, even though idioms are syntactically analyzed similar to literal sentences. Another hybrid approach (cf. Sprenger, et al., 2006) assumes that the idiom is represented both at the concept level and at the lemma level in form of a so-called ‘superlemma’. This superlemma (e.g. “hit the road”) represents the information on the single lemmas (corresponding to the constituents, e.g. hit, the, and road), and their syntactic functions (e.g. subject, direct object), syntactic categories (e.g. noun phrase, prepositional phrase), and parts of speech (e.g. noun, verb). Even though many perception and production studies have investigated the semantic nature of idioms (Cacciari & Tabossi, 1988; Caillies & Butcher, 2007; Gibbs Jr, 2002; Gibbs, 1992; Gibbs, Nayak, & Cutting, 1989; Rabanus, et al., 2008; Smolka, et al., 2007; Titone & Connine, 1999) and their syntactic behavior (Connine, Blasko, Brandt, & Kaplan, 1992; Cutting & Bock, 1997; Gibbs & Gonzales, 1985; Gibbs & Nayak, 1989), there are hardly any studies on the role of prosody in the representation and processing of idioms. I am aware of only one, rather theoretical paper that deals with prosodic constraints in idiomatic sentences (Ashby, 2006). Ashby found that prosodic manipulations may result in humorous effects if the accent/focus distinction is put not on the word that is accented in neutral, broad focus context, but on the non-compositional part of an idiom, as in “It was raining CATS and


5

dogs”, (p. 1594). The underlying mechanism, in his view, is not a lexical specification of accent position, but the fact that non-default accent positions introduce focus distinctions in the idiomatic part. To date, however we know very little about how idioms are actually produced in context and whether a different accent placement indeed leads to the destruction of the figurative meaning of an idiom. Acoustic-prosodic studies so far have mostly concentrated on the prosodic disambiguation of word strings that have both a literal and a figurative reading (also called ditropic sentences). Van Lancker, Canter, and Terbeek (1981), for instance, showed that literal and idiomatic sentences differ in duration and pitch: sentences with idiomatic meaning are shorter (caused by a shorter duration of lexical items), have fewer pauses, and are produced with fewer number of pitch contours compared to sentences with literal meaning. Bélanger, Baum, and Titone (2009) investigated the prosodic disambiguation of idioms in different groups of participants. In the healthy control group, ( the group that is of interest for the present paper), nouns in the idiomatic reading were realized longer than in the literal meaning, but there was no robust difference in the fundamental frequency (f0), the acoustic correlate of speech pitch (operationalized as mean-f0 and f0-range of the phrase, as well as f0-peaks in content words). Although these two studies disagree on whether idiomatic sentences have longer or shorter constituents, it is conceivable that idiomatic phrases are more restricted in prosodic patterns than literal sentences1. Prosodic realization is typically adapted to the discourse context, at least in intonation languages such as English or German (e.g., Vallduví & Engdahl, 1996). For example, words or constituents that are new or contrastive with respect to the prior discourse are realized with a rising-falling nuclear pitch accent in these languages (Baumann, Grice, & Steindamm, 2006; Kügler & Gollrad, 2015 for German), see Examples (1) and (2).


6

Q1: Who drank tea?

(1) The PRIME minister drank tea.

Q2: What did the prime minister drink?

(2) The prime minister drank TEA.

Upon hearing utterances containing such contrastively accented words, Dutch and English listeners have been shown to activate salient contrastive alternatives (Bishop, 2012; Braun & Tagliapietra, 2010; Husband & Ferreira, 2012). Braun and Tagliapietra (2010), for instance, used a cross-modal associative priming paradigm (e.g., Swinney, Onifer, Prather, & Hirshkowitz, 1979) and studied the processing of Dutch sentences in which prime words had either a contrastive or non-contrastive accent. Visual targets that were contrastively related to sentence-final prime words (e.g., pelican following Henk photographed a flamingo) showed priming relative to an unrelated control sentence, though only when the prime word (flamingo) carried a contrastive accent. By contrast, a cross-modal priming study by Norris, Cutler, McQueen, and Butterfield (2006) presented auditory sentences with contrastive accents on either the prime or another word and found priming for semantically related targets irrespective of the location of the contrastive accent. They concluded that a contrastive accent anywhere in the utterance leads listeners to deeper semantic processing, also resulting in associative priming effects for words that preceded the contrastive accent. To date, there are no empirical data on the processing of idiomatic sentences with contrastive accents, and, not surprisingly present models of idiom processing remain silent about prosody. In the next section, I have therefore interpreted how these models may deal


7

with non-default accent position. If idioms are conceived as complex words, parts of the prosodic form of idioms are expected to be stored together with the idiom representation of its lexical constituents and syntax. Hence, the position of the nuclear pitch accent is either specified in the idiomatic representation or can be generated by rule. Take the idiomatic phrase “Das Mädchen ist nah am Wasser gebaut” (literal translation (L): The girl is built close to the water; figurative translation (F): “The girl is a cry-baby”). Given standard assumptions of focus-to-accent theories (Gussenhoven, 1983; Selkirk, 1984), the nuclear accent is realized on the internal argument of the verb, here the word Wasser (water). A stylized contour of such a neutral (broad focus) production of this idiom is shown in (3). Now consider example (4). Here, the idiom is realized with a contrastive nuclear pitch accent on the preposition, which sounds funny at best or may even destroy the figurative meaning, if the accent on the preposition establishes the alternative far from the water. Hence, such idiomatic sentences with contrastive accents, when presented in isolation, are expected to be unacceptable and may even make the figurative meaning inaccessible.

(3) Das Mädchen ist nah am W: The girl

is close to-the

Wasser gebaut water built

F: The girl is a cry-baby

(4) Das Mädchen ist NAH am

Wasser gebaut

On the other hand, if the literal meaning of the individual words is accessed before or in parallel with the figurative meaning, and the figurative concept is activated by the collocation of the idiomatic constituents, contrastive accent will not hinder the processing of


8

the idiomatic meaning. For instance, we have previously observed in German that the literal meaning of words was activated together with the idiomatic meaning of the whole phrase (Rabanus, Smolka, Streb, & Rösler, 2008; Smolka, Rabanus, & Rösler, 2007). In a sentencepriming experiment, participants read an idiomatic sentence, such as Sie hat ihm gründlich den Kopf gewaschen (L: She thoroughly washed his head; F: “She gave him a piece of her mind”) and made lexical decisions about words associated with the idiomatic meaning, such as Standpauke (telling-off), about words associated with the literal meaning of the verb, such as Kleidung (clothes), and matched un-associated nouns. Results indicated that the literal meaning of the (sentence-final) verb was activated in parallel with the idiomatic meaning of the sentence. Given that (previously collected) sentence completion tasks assessed that the idiomatic meaning was processed before the sentence-final word, these findings indicate that the idiomatic constituents simultaneously activate the literal meanings. Since the position of pitch accents is specified at the phrase-level, incorrect placement of pitch accents may not be detrimental to the processing of idioms. When we assume that the literal meaning of single word constituents is accessed during figurative sentence processing, and that phrase-level prosody is a post-lexical process, we may expect idiom representations – similar to lexical representations –not to store information on the location of pitch accents. This paper examined the effects of contrastive intonation on the processing of idiomatic sentences. The structure of the paper is as follows: Experiment 1 tested how neutral and contrastive intonation contours affect the perceived acceptability of idiomatic sentences as compared to literal sentences in isolation. If idiomatic sentences do not allow much prosodic variation (Ashby, 2006), contrastive intonation contours will be more acceptable in literal than in idiomatic sentences. By contrast, idiomatic and literal sentences with a neutral intonation contour should yield similar acceptability ratings. Experiment 2 was a cross-modal associative priming experiment, in which participants heard idiomatic and literal sentences


9

spoken in a contrastive and neutral intonation contour and made lexical decisions to visual targets that were presented at the onset of the ultimate word (the participle) and were semantically related to the idiomatic sentence. The literal sentences served as unrelated controls for the idiomatic sentences. If contrastive intonation contours do not alter the comprehension of idiomatic sentences, idiomatic sentences with a contrastive intonation contour should yield priming effects similar to those with a neutral intonation contour. If, however, a contrastive intonation contour interrupts the access of the figurative meaning of an idiom, idiomatic sentences with contrastive stress will not yield priming. Note that previous cross-modal priming experiments in English have shown priming effects for visual targets presented at the offset of the idiom, but not at the offset of the penultimate word (Titone & Libben, 2014 for English, see also Caillies & Butcher, 2007 for a manipulation of SOA between idiom off-set and target on-set). However, different from these studies in English, the sentence materials in German hold very high sentence completion rates for the participle, i.e. the ultimate word, indicating that the idiomatic meaning is already available at the onset of the ultimate word. Experiment 1 Experiment 1 tested whether contrastive intonation contours are perceived as acceptable, and to what degree. To this end, an experiment was conducted with speeded acceptability judgments to literal or idiomatic sentences in either (a) neutral (broad focus), (b) contrastive (narrow focus on the preposition), or (c) monotonous sentence prosody. Participants heard the sentences and judged as fast as possible whether the sentence sounded ‘natural’ or not. Monotonous prosody and the neutral prosody conditions served as baseline and top-line, respectively, irrespective of whether the utterances are literal or idiomatic: Monotonous prosody should yield mostly unacceptable responses, while neutral sentences should be perfectly acceptable. Further, contrastive intonation contours should be more


10

acceptable for literal sentences than for idiomatic ones. Note that contrastive prosody is applied in cases of metalinguistic corrections, cases in which even accents on usually unaccented syllables are attested (e.g., I didn't say "proVERB", I said "proFORM"). However, because the current experiment does not present a context for metalinguistic correction, it is not clear whether speakers will treat the utterances with contrastive accents as acceptable at all. Method Participants Twenty-four students at the University of Konstanz participated in the experiment in return for either course credit or payment. They all were native German speakers, had normal or corrected to normal eyesight and no recorded hearing issues. They had not taken part in other experiments involving idioms and were unaware of the purpose of the experiment. Materials Experimental sentences. Sixty-six German sentences were selected from a previously collected sentence pool (for a detailed description see Smolka, et al., 2007). Thirty-three were literal sentences like Sie hat ihre Hände am Kachelofen gewärmt (L: She warmed her hands at the stove), and 33 were idiomatic, such as “Sie hat immer nach den Sternen gegriffen” (F: “She always reached for the stars”). All 66 sentences comprised 7 words, starting with a subject noun followed by an auxiliary and ending in an infinite verb (i.e., a verb participle). Most importantly, all sentences contained a prepositional phrase such as am Kachelofen (at the stove) or nach den Sternen (for the stars). Literal and idiomatic sentences were paired based on sentence completion rate (i.e., cloze probability), that is how often the same verb participle was chosen to complete the last word of the sentence by participants (for a full description of the procedure see Smolka, Rabanus, & Rösler, 2007). The average completion rate was 89% (SD = 10.7, range = 65.4 –


11

100) for idiomatic sentences and 88% (SD = 9.96, range = 64 – 100) for literal sentences. The Appendix lists all sentence materials; speech samples for each of the intonation contours can be downloaded with the supplementary materials. Each idiomatic and literal experimental sentence was recorded in three intonation conditions, see Figure 1: (a) neutral, that is, with a non-contrastive broad focus intonation, typically realized with a high-pitched accent (H*) on the noun of the prepositional phrase, (b) contrastive, that is, with a contrastive accent (L+H*) on the preposition, “Sie hat immer NACH den Sternen gegriffen” and (c) monotonous, that is, with no f0-movement across the utterance. Filler sentences. To prevent strategic effects, a total of 168 filler sentences was inserted. Like the experimental sentences, all were composed of 7 words. Thirty of the filler sentences had the direct object in the preverbal field, all others were similar to the experimental sentences. Of the filler sentences, 30 were recorded with a contrastive intonation contour: 13 with a contrastive accent on the subject pronoun, 7 with a contrastive accent on the object noun, and 10 with a contrastive accent on the sentence-final participle. Furthermore, 15 of the filler sentences were recorded with an intonation phrase break at varying positions in the phrase (after the subject, the verb, etc.). The remaining filler sentences had a neutral, broad focus intonation. Overall, the large number of filler sentences reduced the proportion of (a) idiomatic sentences to 14.1%, (b) sentences with contrastive sentence prosody to 22.2%, and (c) sentences without pitch movement to 9.4% of the whole material set.


12

Figure 1 about here Apparatus and Recordings The complete sentence material was recorded by a female native speaker of Standard German. The recording took place in a sound attenuated cabin at the Phonetics and Phonology Lab of the University of Konstanz. The productions were checked online by two trained phoneticians as the recording took place; productions with inappropriate prosody or hesitations were recorded again at the end of the session. All audio files were scaled in intensity using the PRAAT software package (Boersma & Weenink, 2009). Auditory stimuli were played via AKG GHS 1 headphones and visual stimuli were presented on a 21.5” Dell U2211H flat panel HD widescreen monitor. The presentation of the trials and response collection was conducted with the Presentation software by neurobehavioral systems (https://www. neurobs.com/). Responses were recorded via a push-button box with two buttons. Procedure Critical sentences in the three different prosody versions (neutral/ contrastive/ monotonous) were rotated across three lists as per a Latin square design. Participants received only one experimental list and thus heard each critical sentence only once in one of the three intonation conditions. In total, each list comprised 234 sentences (66 experimental and 168 filler sentences). Each list was further divided into three blocks with 78 items per block, each block containing the same number of stimuli in each intonation condition. The order of trials in a block was pseudo-randomized, so that two idiomatic sentences were separated by a minimum of four other trials. Participants were tested individually in a quiet room and were randomly assigned to one of the lists (eight participants for each list). Each trial started with the presentation of a fixation cross in the center of the screen (in white Arial 24 pt characters on a black


13

background), accompanied by the auditory presentation of a sentence. At the sentence offset, the fixation cross was replaced by a question mark (also in white Arial 24 pt characters on a black background). Participants were told that the experiment was testing machine-generated language and that they would hear computer-generated sentences. Their task was to judge whether the sentences they just heard “sounded natural” as fast as possible, that is whether it had a natural sentence prosody. Right-handers indicated “yes” responses by a right button press and “no” responses with a left button press (and vice versa for left-handers). Button presses and reaction times were recorded relative to the onset of the question mark. The question mark remained on screen until a participant's response (i.e. there was no timeout). The experiment lasted approximately 30 minutes. Results Acceptability Judgments Figure 2 depicts the percentage of “acceptable” responses, split by intonation condition (monotonous, contrastive, neutral) and sentence type (idiomatic, literal). ‘Acceptable’ responses were analyzed using a logistic mixed effects regression model (Baayen, Davidson, & Bates, 2008; Jaeger, 2008) with intonation condition and sentence type as fixed factors, crossed random intercepts for participants and items and by participant and by-items slopes. The results showed a significant effect of intonation condition: Responses to utterances in the neutral condition were judged as significantly more acceptable than utterances with a contrastive (ß = 8.1, SE = 1.1, t = 7.9, p < 0.0001) or monotonous intonation (ß = 7.8, SE = 0.9, t = 8.7, p < 0.0001). Responses to the monotonous and the contrastive condition did not differ from each other (p > 0.6). Furthermore, there was a significant main effect of sentence type. Literal sentences (95.4%) were judged as slightly more acceptable


14

than idiomatic sentences (87.1%, ß = 1.6, SE = 0.6, t = 2.7, p < 0.01). There was no interaction between sentence type and intonation condition (p > 0.3). Figure 2 about here Reaction Times Reaction times are hard to compare across the different intonation conditions due to the very different number of acceptable responses across conditions. To examine whether idiomatic sentences are harder to process than literal sentences, only the reaction times for acceptable responses in the neutral prosody conditions were analyzed (n = 482). Reaction times longer than 1100 ms were removed (n = 8, 1.6% of the total data points) to distribute the data more conventionally. The remaining 474 data points were analyzed using linear mixed effects regression models with sentence type as fixed factor, reaction time to the preceding trial (previous RT) as control predictor, crossed random intercepts for participants and items, and by-participants and by-items slopes. All p-values were calculated on the basis of Satterthwaite approximation by using the lmerTest package (Kuznetsova, Per Bruun, & Rune Haubo Bojesen, 2013). The results showed a significant main effect of the control predictor previous RT (ß = 0.13, SE = 0.02, t = 6.4, p < 0.0001). However, there was no effect of sentence type (ß = 20.2, SE = 12.6, t = 1.6 p > 0.12), indicating that responses to idiomatic sentences (314 ms) were as fast as responses to literal sentences (287 ms). Discussion As expected, sentences presented in the neutral intonation condition were judged as highly acceptable (95.4% for literal and 87.1% for idiomatic sentences), and sentences presented in the monotonous intonation condition were judged as completely unacceptable (3.7% for literal and 1.1% for idiomatic sentences). However, contrary to the expectations, sentences in contrastive prosody were also judged to be unacceptable, and this held for both literal (8.3%) and idiomatic (4.5%) sentences. That is to say, there was a substantial


15

difference in judgments (of more than 80%) between the neutral and the contrastive intonation condition, and contrastive utterances were judged just as unacceptable as the monotonous sentences (monotonous = contrastive < neutral) in both literal and idiomatic sentences. It seems that listeners penalized the fact that contrastive accents on the preposition are rarely encountered without an appropriate discourse context, and this was true irrespective of the utterance being interpreted literally or figuratively. Overall, idiomatic sentences were judged as slightly less acceptable than literal sentences in all three intonation conditions. It is difficult to imagine the source of this effect, especially as both literal and idiomatic sentences had extremely high sentence-completion rates of nearly 90%, which suggests that the chosen idioms are well known to native German speakers. It is possible that the effect of sentence type found here results from the task of conscious judgments. When the same sentence types were applied as primes in a sentence priming paradigm, responses latencies to unrelated targets were similar following literal (661 ms) or idiomatic (654 ms) sentences (cf. Smolka et al., 2007). I will return to this issue in the General Discussion. The acceptability judgments clearly showed that sentences with a contrastive accent on the preposition are unacceptable. This finding may lend empirical support to the theoretical claim by Ashby (2006) who argued that focus distinctions in the idiomatic part are difficult to reconcile with the figurative meaning. However, the fact that listeners reacted in the same way to literal sentences with contrastive accents weakens this explanation. Experiment 2 will investigate whether the figurative meaning is still available when idioms are presented with a contrastive intonation contour. Experiment 2 Experiment 2 tested the online processing of idiomatic sentences with a contrastive and neutral intonation condition. In a cross-modal priming experiment, participants heard an


16

idiomatic or literal prime sentence (matched for the number of words and sentencecompletion rates) and made lexical decisions about visual targets that were associated with the figurative meaning of the idiom. To make idiomatic and literal sentences comparable in predictability, only idiomatic and literal sentences with very high sentence completion rates (on average 90%) were used. Such high completion rates indicate that the last word of the sentence (i.e. the participle) is highly predictable, which in turn indicates that the meaning of the whole sentence is recognized and can be predicted before the ultimate word. In this experiment, the associates (i.e. the visual targets) to the idiomatic meaning were thus presented at the onset of the ultimate word (the participle), when the sentence meaning was already predictable. Hence, figurative priming was tested at the onset of the ultimate word (the participle). The crucial manipulation was that the intonational realization of the auditory prime was either neutral or contrastive. Priming for the idiomatic sentences with neutral and contrastive intonation contours was measured relative to the unrelated literal sentences with neutral and contrastive intonation, respectively. Table 1 provides examples of the experimental set up. Table 1 about here Method Participants Forty native speakers of German took part in the experiment in return for either course credit or payment. They had normal or corrected to normal eyesight and no recorded hearing issues. They did not participate in Experiment 1. Materials Web-based association test. A web-based association experiment was conducted to find the word that best describes the figurative meaning of an idiomatic sentence. The experiment included the 33 idioms of Experiment 1 together with 26 additional idioms with


17

the same syntactic structure (all taken from the sentence pool described in Smolka et al., 2007). Thirty-five native German speakers volunteered in the association experiment. They saw one idiomatic sentence at a time and typed one word that best described the idiomatic meaning (no constraints were given regarding the part-of-speech of the associated word). For each idiom, the association strength of the most frequently named associate was calculated, dividing the number of participants that named that particular associate by the total number of participants. If one associate was mentioned for more than one idiom, one of the idioms was removed from the set. Also, idioms where the most frequently named associate was phonologically or semantically related to one of the words in the idiom were removed. The remaining set of 48 idioms was divided into two groups with comparable association strengths (mean = 39%, SD = 14.9, range = 24 - 80) vs. mean = 20.3%, SD = 13.8, range = 6 – 73, respectively) and sentence completion rates (mean = 89%, SD = 13.5, range = 52 – 100 vs. mean = 91%, SD = 11.3, range = 65 - 100, respectively). One set was used as an experimental set to be paired with existing words as visual targets, the other was used as a filler set to be paired with non-words. Experimental sentences. Twenty-four of the 48 idioms were used as experimental items and paired with their most frequent associate. Furthermore, each of these 24 experimental idioms was paired with a literal sentence that had the same sentence-completion rate (and hence predictability), but whose words were unrelated to the idiomatic associate. These literal sentences served as an unrelated control condition. Table 1 gives an overview of the experimental manipulations. Each experimental sentence (idiomatic and literal) was recorded twice, once with a contrastive accent on the preposition and once with a broad focus intonation, similarly to Experiment 1. Filler sentences. Two hundred and twelve filler sentences were selected. Twenty-four of these were idiomatic and were paired with non-words that were derived from their


18

associates by exchanging one or two letters while preserving the phonotactic constraints of German. Half of these idiomatic filler sentences were recorded with a contrastive and half with a non-contrastive intonation contour. The remaining 188 filler sentences were divided into four groups: 144 were paired with non-word targets (72 recorded with a contrastive and 72 with a non-contrastive intonation), 144 with word targets (72 recorded with a contrastive and 72 with a non-contrastive intonation). Apparatus and Recordings The apparatus and procedure of the recordings were the same as in Experiment 1. Procedure The four sentence primes (idiomatic neutral/ idiomatic contrastive/ literal neutral/ literal contrastive) of a single target were randomly allocated to four lists in such a way that each participant received each visual target only once in one of the four conditions. Each list comprised 236 trials in total, with 24 experimental sentences (12 idiomatic, 12 literal), half of the sentences were in contrastive and half in neutral intonation. The items were presented in three blocks, each with a matched number of idiomatic and literal sentences, intonation contours and target type (word/non-word). The order of the trials was pseudo-randomized, with (a) at least two literal sentences in between two idiomatic sentences, (b) maximum three word or nonword responses in a row, and (c) maximum three trials of the same intonation contour in a row. Participants were randomly assigned to one of the four lists (10 participants each). Each trial started with the presentation of a white fixation cross in the middle of a black screen for 500 ms. The auditory presentation of the sentences started with the presentation of the fixation cross. At the onset of the ultimate word (i.e. the participle) of the prime sentence, the visual target appeared in the center of the screen in white Arial 24 pt characters on a black background. The target remained on the screen until the participant


19

responded, so that there was no timeout. Lexical decisions were recorded relative to the appearance of the target. Feedback on the correctness of a response (FALSCH, incorrect; RICHTIG, correct, in red Arial 24 pt) was given for 500 ms on each of the practice trials and on incorrect responses only during the experimental session. Inter-trial interval was 1,000 ms. The experiment lasted for approximately 30 minutes. Results Participants responded incorrectly in 17 experimental trials. Errors were distributed evenly across conditions (4 to 5 in each condition). After visual inspection of the density distribution of the reaction times (RTs), trials with reaction times larger than 1,100 ms were removed to make the data more normally distributed (removing 38 data points, 4% of the data). The mean RTs of these data, split by intonation condition and sentence type are shown in Figure 3. Figure 3 about here Linear mixed effects analysis were performed on the RT data by using R (R Core Team, 2012) and lme4 (e.g., Bates, 2005; Bates, Maechler, & Bolker, 2012; Baayen, Davidson, & Bates, 2008). The fixed-effects factors were intonation condition and sentence type. The random effects were intercepts for participants and items, as well as by-participants and by-items slopes for the fixed factors (cf. Barr, Levy, Scheepers, & Tily, 2013; Cunnings, 2012). To remove autocorrelational structure from the residual errors (Baayen & Milin, 2010), the response latency and the correctness at the preceding trial (previous RT and previous correct, respectively) were included as control predictors, as well as several control predictors referring to the target, such as the number of letters, the experimental block, partof-speech, and the log-transformed lemma frequency from dlexDB (Heister, et al., 2011). The control predictors number of letters, previous RT and log-transformed lemma frequency were centered (cf. Winter, 2013), that is the average of these variables is taken as reference in the


20

intercept of the model. The best model fit was obtained by comparing the Akaike Information Criterion (AIC) statistics between models (cf. Sakamoto, Ishiguro, & Kitagawa, 1986). The best model is shown in Table 2. The effects of the control predictors were as expected. Lexical decision times to the visual target were faster when (a) the preceding trial was fast, (b) the target was of higher lexical frequency, (c) the target held fewer letters, and (d) occurred later in the experiment (after the first block). With respect to the experimental predictors, there was a significant interaction between sentence type and intonation condition (p = 0.025). As Figure 3 illustrates, there was a priming effect of 33 ms for idiomatic over literal sentences, but only when the intonation contour was neutral and not when it was contrastive. Discussion The results showed a priming effect for idioms only when they were produced with a non-contrastive (neutral) intonation contour. That is, there was no priming when the idiom had a contrastive accent on the preposition. This inhibitory effect of contrastive prosody on idiom processing is discussed in more detail in the General Discussion. The finding of priming for the figurative meaning in idioms with a neutral intonation contour corresponds to previous findings. For example, a recent cross-modal priming experiment on English idioms (Titone & Libben, 2014) and on German idioms (Smolka, Rabanus, & Rösler, 2006), and intra-modal (visual-visual) priming experiments on French idioms (Caillies & Butcher, 2007), and intra-modal priming experiments on German idioms (Rabanus et al., 2008; Smolka et al., 2007) found facilitatory effects for figurative meanings. In the study by Titone and Libben (2014), though, the associate of the figurative meaning yielded priming only when it was presented at the offset of the idiom, and not when it was presented at the onset of the ultimate word, as was the case in our study. Regarding French idioms (Caillies & Butcher, 2007), the associate of the figurative meaning yielded priming as


21

a function of SOA and compositionality of the idiom. The figurative meaning of decomposable idioms was available earlier on (immediately at the offset of the idiom in Experiment 2 and at SOA 350 in Experiment 1) than that of non-decomposable idioms (SOA 500 in both experiments). The differences between these and our study are most likely caused by the different structure of the sentence materials: All German sentences were in the perfect tense, which places the verb—the past participle that bears the lexical meaning of the analytic expression of the German perfect tense—at the very end of the sentence. A cloze probability of roughly 90% (as assessed in a sentence-completion task) ensured that the verb in sentencefinal position was highly predictable. That is, the idiomatic meaning (and hence also the nouns that typically complement the idiomatic meaning) must have been accessed before the verb at sentence-final position. This strongly differs from the structure of the English or French idioms in which the critical nouns that complement the idiomatic meaning were typically the final word of the sentences (e.g., “shut your trap”, “skate on thin ice”; “Aude fut surprise d’appendre que son voisin avait cassé sa pipe”, F: “Aude was surprised to hear that her neighbor had kicked the bucket”). Furthermore, final word predictability in Titone and Libben (2014) was lower and very variable, ranging from 0% (“shut your trap”) to 100% (“skate on thin ice”), while it was rather high (on average 90%) in the German materials used in this study. Therefore, it is not surprising to find an earlier priming effect than in the English study. It will require further research in the future to determine how long the inhibitory effect of contrastive intonation on the figurative meaning lasts. General Discussion This study investigated the effects of contrastive prosody on idiom processing. To this end, an off-line experiment with acceptability judgments and an on-line experiment with cross-modal associative priming were conducted. Both experiments showed that contrastive accent on the preposition hinders idiom processing.


22

Experiment 1 showed that German listeners accepted only utterances with a neutral intonation contour (i.e., with a non-contrastive accent in the default position). Unexpectedly, Experiment 1 further showed that contrastive accent on the preposition is not only problematic for idiomatic sentences but also for literal sentences. Without an appropriate context, listeners seem to be incapable of metalinguistic correction, that is, of accommodating a context that renders these contrastive utterances as acceptable. Interestingly, sentences with a contrastive prosody were judged as unacceptable as utterances with a monotonous (flat) intonation contour, for which it is hard to imagine an acceptable context. It is possible that the implicit time pressure resulting from the use of a button-box as well as the explicit instruction to decide as quickly as possible prevented listeners from accommodation. It is also possible that the instruction to judge the ‘natural’ soundness of the sentences directed listeners to accept only the most expected, that is, the most frequently occurring ‘neutral’ intonation. Experiment 2 is—to my knowledge—the first experiment that examines the effects of prosody during the online processing of idioms. It showed that idiomatic sentences with a contrastive prosody are not only unacceptable but also make it impossible for listeners to access the idiomatic meaning. From a prosodic processing perspective, contrastive accents have been shown to make alternatives to the contrastively accented word accessible (Bishop, 2012; Braun & Tagliapietra, 2010; Weber, Braun, & Crocker, 2006; Husband & Ferreira, 2012; Ito & Speer, 2008; Weber, Grice, & Crocker, 2006). For the experimental sentences used in this study, the alternative that is made accessible refers to a different preposition. In other words, it is possible that participants encounter the contrastive accent on the preposition and search for an alternative, such as auf (up) versus unter (down). This search for literal alternatives may be detrimental for accessing the figurative meaning of the sentence. This mechanism may be one explanation for the lack of a priming effect in idioms with contrastive accents. Note that


23

Norris, et al. (2006) argued that a contrastive accent anywhere in the utterance leads to a deeper semantic processing of the sentences. Their claim is based on associative priming effects that were independent of the location of the contrastive accent in literal prime sentences (i.e., they found associative priming irrespective of whether the contrastive accent was located on the word that was associated with the visual target or on another word). If their assumption holds that contrastive accents lead to deeper semantic processing, then the inhibitory effect of contrastive accents in idiomatic sentences suggests that deeper semantic processing implies more literal processing in the case of idioms. Indeed, the inhibitory effect of contrastive sentence prosody fits well with the previous findings of Smolka et al. (2007), indicating that the literal meaning of the idiomatic constituents is activated together with the figurative meaning of the idiomatic sentence. If contrastive accent leads to deeper semantic processing and hence to more literal processing in the case of idioms, it is possible that the contrastive stress on the preposition focusses on the literal interpretation of the preposition and a search for possible (contrastive) alternatives. This search for alternatives hinders the processing of the idiomatic constituents in a figurative way so that the access and activation of the idiomatic constituents does not spread to activating the idiomatic concept. Nevertheless, it is also conceivable that stress patterns are explicitly integrated in hybrid models of idiom processing. For example, the ‘superlemma’ in the model of Sprenger et al. (2006) could entail—besides the lexical and syntactic properties—also the prosodic characteristics of the idiom in form of its most typical stress pattern (e.g. “hit the ROAD”). This would require the assumption of some kind of ‘prosodic fixedness’, similar to the assumptions of semantic and syntactic fixedness of idioms in these models. To summarize, the present study showed that contrastive accent on the preposition hinders idiom processing. It remains for future experiments to explore how long the


24

inhibitory effect of contrastive accentuation lasts. Previous priming studies on the processing of utterances with an unfamiliar intonation contour (Braun, Dainora, & Ernestus, 2011) suggested that listeners do not recover quickly from uncommon contours. It further remains to be explored whether the contrastive accent is as detrimental to the figurative meaning when it is placed on the noun and not on the preposition, given that a contrastive accent on the noun is prosodically more like a neutral, broad focus condition (the accent location is the same and there are less categorical differences between broad and narrow focus accents on the focus exponent).


25

References Ashby, M. (2006). Prosody and idioms in English. Journal of Pragmatics, 38(10), 15801597. Baayen, H. R., Davidson, D. J., & Bates, D. M. (2008). Mixed-effects modeling with crossed random effects for subjects and items. Journal of Memory and Language, 59(4), 390412. Baayen, R. H., & Milin, P. (2010). Analyzing reaction times. International Journal of Psychological Research, 3(2), 12-28. Barr, D. J., Levy, R., Scheepers, C., & Tily, H. J. (2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 68(3), 255-278. Bates, D. (2005). Fitting linear mixed models in R. R news, 5(1), 27-30. Bates, D., Maechler, M., & Bolker, B. (2012). lme4: Linear mixed-effect models using S4 classes. R package version 0.999999-0. Baumann, S., Grice, M., & Steindamm, S. (2006). Prosodic marking of focus domains categorical or gradient? Paper presented at the 3rd International Conference on Speech Prosody, Dresden, Germany. Bélanger, N., Baum, S. R., & Titone, D. (2009). Use of prosodic cues in the production of idiomatic and literal sentences by individuals with right- and left-hemisphere damage. Brain and Language, 110(1), 38-42. Bishop, J. (2012). Focus, prosody, and individual differences in “autistic” traits: Evidence from cross-modal semantic priming. UCLA Working Papers in Phonetics, 111, 1-26. Bobrow, S. A., & Bell, S. M. (1973). On catching on to idiomatic expressions. Memory & Cognition, 1, 343-346. Boersma, P., & Weenink, D. (2009). Praat: Doing phonetics by computer. Amsterdam.


26

Braun, B., Dainora, A., & Ernestus, M. (2011). An unfamiliar intonation contour slows down on-line speech comprehension. Language and Cognitive Processes, 26, 350-375. Braun, B., & Tagliapietra, L. (2010). The role of contrastive intonation contours in the retrieval of contextual alternatives. Language and Cognitive Processes, 25, 10241043. Cacciari, C., & Glucksberg, S. (1994). Understanding figurative language. In M. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 447-477). San Diego, CA, US: Academic Press. Cacciari, C., & Tabossi, P. (1988). The comprehension of idioms. Journal of Memory and Language, 27(6), 668-683. Caillies, S., & Butcher, K. (2007). Processing of Idiomatic Expressions: Evidence for a New Hybrid View. Metaphor and Symbol, 22(1), 79-108. Connine, C. M., Blasko, D., Brandt, R., & Kaplan, J. (1992). Idiomatic processing: Syntactic frozenness and subjective familiarity. Psychological Research, 54(3), 225-232. Cunnings, I. (2012). An overview of mixed-effects statistical models for second language researchers. Second Language Research, 28(3), 369-382. Cutting, J., & Bock, K. (1997). That's the way the cookie bounces: Syntactic and semantic components of experimentally elicited idiom blends. Memory & Cognition, 25(1), 5771. Gibbs Jr, R. W. (2002). A new look at literal meaning in understanding what is said and implicated. Journal of Pragmatics, 34(4), 457-486. Gibbs, R. W. (1980). Spilling the beans on understanding and memory for idioms in conversation. Memory and Cognition, 8(2), 149-156. Gibbs, R. W. (1992). What do idioms really mean? Journal of Memory and Language, 31(4), 485-506.


27

Gibbs, R. W. (1994). Figurative thought and figurative language. In M. Gernsbacher (Ed.), Handbook of psycholinguistics (pp. 411-446). San Diego, CA, US: Academic Press. Gibbs, R. W., & Gonzales, G. P. (1985). Syntactic frozenness in processing and remembering idioms. Cognition, 20(3), 243-259. Gibbs, R. W., & Nayak, N. P. (1989). Psycholinguistic studies on the syntactic behavior of idioms. Cognitive Psychology, 21(1), 100-138. Gibbs, R. W., Nayak, N. P., & Cutting, C. (1989). How to kick the bucket and not decompose: Analyzability and idiom processing. Journal of Memory and Language, 28(5), 576-593. Gussenhoven, C. (1983). Focus, Mode and the Nucleus. Journal of Linguistics, 19(2), 377417. Heister, J., Würzner, K.-M., Bubenzer, J., Pohl, E., Hanneforth, T., Geyken, A., et al. (2011). dlexDB - eine lexikalische Datenbank für die psychologische und linguistische Forschung [dlexDB - a lexical database for psychological and linguistic research]. Psychologische Rundschau, 62, 10-20. Holsinger, E., & Kaiser, E. (2013). Processing (non)compositional expressions: mistakes and recovery. Journal of Experimental Psychology: Learning Memory, & Cognition, 39(3), 866-878. Husband, M. E., & Ferreira, F. (2012). Generating contrastive alternatives: Activation and suppression mechanisms. Paper presented at the 25th Annual CUNY Conference on Human Sentence Processing, New York. Jaeger, T. F. (2008). Categorical data analysis: Away from ANOVAs (transformation or not) and towards logit mixed models. Journal of Memory and Language, 59(4), 434-446. Kügler, F., & Gollrad, A. (2015). Production and perception of contrast: The case of the fallrise contour in German. Frontiers in Psychology, 6.


28

Kuznetsova, A., Per Bruun, B., & Rune Haubo Bojesen, C. (2013). Lmertest: Tests for random and fixed effects for linear mixed effect models (lmer objects of lme4 package). R package version. Norris, D., Cutler, A., McQueen, J. M., & Butterfield, S. (2006). Phonological and conceptual activation in speech comprehension. Cognitive Psychology, 53, 146 - 193. Rabanus, S., Smolka, E., Streb, J., & Rösler, F. (2008). Die mentale Verarbeitung von Verben in idiomatischen Konstruktionen [The mental processing of verbs in idiomatic constructions]. Zeitschrift für Germanistische Linguistik, 36(1), 27-47. R Core Team (2013). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org/. Selkirk, E. O. (1984). Phonology and Syntax. The Relation between Sound and Structure. Cambridge, MA: MIT Press. Smolka, E., Rabanus, S., & Rösler, F. (2006, October). The processing of verbs in German idiomatic and literal phrases. Talk at the Fifth International Conference on the Mental Lexicon, Montreal, Canada. Smolka, E., Rabanus, S., & Rösler, F. (2007). Processing verbs in German idioms: Evidence against the Configuration Hypothesis. Metaphor and Symbol, 22(3), 213-231. Sprenger, S. A., Levelt, W. J. M., & Kempen, G. (2006). Lexical access during the production of idiomatic phrases. Journal of Memory and Language, 54, 161-184. Swinney, D., & Cutler, A. (1979). The access and processing of idiomatic expressions. Journal of Verbal Learning and Verbal Behavior, 18(5), 523-534. Swinney, D., Onifer, W., Prather, P., & Hirshkowitz, M. (1979). Semantic facilitation across sensory modalities in the processing of individual words and sentences. Memory and Cognition, 7, 159 – 165.


29

Titone, D., & Libben, M. (2014). Time-dependent effects of decomposability, familiarity and literal plausibility on idiom meaning activation. The Mental Lexicon, 9(3), 273-496. Titone, D. A., & Connine, C. M. (1999). On the compositional and noncompositional nature of idiomatic expressions. Journal of Pragmatics, 31(12), 1655-1674. Vallduví, E., & Engdahl, E. (1996). The linguistic realisation of information packaging. Linguistics, 34, 459-519. Van Lancker, D., Canter, G. J., & Terbeek, D. (1981). Disambiguation of ditropic sentences: acoustic and phonetic cues. Journal of Speech and Hearing Research, 24(3), 330-335. Weber, A., Braun, B., & Crocker, M. W. (2006). Finding referents in time: Eye-tracking evidence for the role of contrastive accents. Language and Speech, 49(3), 367-392. Winter, B. (2013). Linear models and linear mixed effects models in R with linguistic applications. arXiv:1308.5499. [http://arxiv.org/pdf/1308.5499.pdf]

Effects of prosody on idiom processing Appendix Idiomatic Sentence Ihm ist ein Stein vom Herzen gefallen. W: Him is a stone off-the heart fallen.

Associate

%

Erleichterung (N)

80.00

relief

F: It took a load off his mind. Er ist ihr in den Hintern gekrochen. W: He is her in the bottom crept.

Schleimer (N)

71.43

yes-man

F: He bootlicked her. Sie hat sich an einen Strohhalm geklammert.

Hoffnung (N)

W: She has herself on a straw clasped.

hope

54.55

F: She was clutching at straws. Es hat ihm sehr am Herzen gelegen. W: It has him much at-the heart lain.

wichtig (A)

57.14

important

F: It was very dear to him. Er ist dabei über seinen Schatten gesprungen. W: He is in-doing-so over his shadow jumped.

Überwindung (N)

42.86

will power

F: He swallowed the bitter pill. Es ist mir über den Kopf gewachsen. W: It is me over the head grown.

Überforderung (N)

42.86

excessive demand

F: It was too much for me. Er hat deutlich über die Stränge geschlagen. W: He has definitely over the cords beaten.

übertreiben (V)

41.18

exaggerate

F: He clearly went way too far. Sie hat sich auf die Socken gemacht. W: She has herself on the socks made.

losgehen (V) leave

42.86

30

Effects of prosody on idiom processing F: She got a move on. Das Mädchen ist nah am Wasser gebaut. W: The girl is close to-the water built.

weinen (V)

47.06

cry

F: The girl is a cry-baby. Sie ist ihm ständig ins Wort gefallen. W: She is him constantly in-the word fallen.

unterbrechen (V)

47.06

interrupt

F: She interrupted him constantly. Er hat alles auf seine Kappe genommen. W: He has everything on his cap taken.

Schuld (N)

40.00

guilt

F: He took the responsibility for everything. Sie ist ihm in den Rücken gefallen. W: She is him in the back fallen.

Verrat (N)

29.41

betrayal

F: She stabbed him in the back. Sie hat damit genau ins Schwarze getroffen. W: She has with-it exactly in-the black hit.

richtig (A)

28.57

right

F: She hit home with it. Er hat einiges auf dem Kasten gehabt. W: He has quite- a-lot on the case had.

Intelligenz (N)

28.57

intelligence

F: He has plenty of grey matter. Sie hat ihn auf den Arm genommen. W: She has him on the arm taken.

veräppeln (V)

31.43

pull someone’s leg

F: She pulled his leg. Er hat alle über einen Kamm geschoren.

gleichmachen (V)

W: He has all over a comb sheared.

equalize

F: He tarred them all with the same brush.

25.71

31

Effects of prosody on idiom processing Er hat sich aus dem Staub gemacht. W: He has himself out-of the dust made.

abhauen (V)

25.71

run away

F: He absconded. Er hat uns wieder übers Ohr gehauen. W: He has us again over-the ear hit.

Betrug (N)

29.41

fraud

F: He ripped us off again. Er ist wieder aus der Haut gefahren. W: He is again out-of the skin driven.

Wut (N)

28.57

rage

F: He lost his temper again. Ihre Schönheit hat ihm ins Auge gestochen. W: Her beauty has him in-the eye stabbed.

auffallend (A)

28.57

striking

F: Her beauty caught his eye. Sie hat immer nach den Sternen gegriffen. W: She has always to the stars reached.

Traum (N)

23.53

dream

F: She always reached for the stars. Er ist schon bald vom Fleisch gefallen. W: He is already soon from-the flesh fallen.

mager (A)

27.27

skinny

F: He was on the verge of being skin and bones. Ihr ist Geld durch die Finger geronnen. W: Her is money through the fingers flown.

verschwenderisch (A)

31.25

extravagant

F: She spends money like water. Er hat sie auf die Folter gespannt. W: He has her on the torture stretched. F: He kept her on tenterhooks.

warten (V) wait

25.71

32


33

Note. Idiomatic prime sentences, their semantic associates and percent association strength (%); W = word-by-word translation, F = figurative meaning; semantic associates were nouns (N), verbs (V), or adjectives (A).


34

Acknowledgements This study was supported by a Young Scholar Fund of the University of Konstanz, as well as by the Volkswagen Foundation, Grant FP 561/11, awarded to Eva Smolka. I thank Bettina Braun for fruitful discussions on earlier versions of this paper; and I thank Sarah Baumann, Susann Tschernitz and Samuel Schweizer for conducting the experiments.


35

Footnotes Footnote 1: Note that differences in the variability of intonational tune in idiomatic and literal sentences need not necessarily be observable as differences in mean pitch over the utterance.


36

Table 1 Examples of Idiomatic and Literal Control Sentences and Visual Targets in Experiment 2 Auditory Prime Sentence Intonation Neutral

Literal Control

Idiomatic

Sie hat herzlich über seine Witze gelacht.

Ihr ist ein Stein vom Herzen gefallen.

(W) She has cheerfully about his jokes laughed.

(W) To her is a stone from the heart fallen.

(L) She laughed cheerfully about his jokes.

(L) A stone has fallen off her heart.

Visual Target

(F) It took a load off her mind. Erleichterung Contrastive

Sie hat herzlich ÜBER seine Witze gelacht.

Ihr ist ein Stein VOM Herzen gefallen. relief

(W) She has cheerfully ABOUT his jokes laughed.

(W) Her is a stone FROM the heart fallen.

(L) She laughed cheerfully ABOUT his jokes.

(L) A stone has fallen OFF her heart. (F) It took a load OFF her mind.

Note. Capitals indicate contrastive accent locations. Translations of a sentence into English are given word-by-word (W), literally (L), and figuratively (F).


37

Table 2 Fixed Effects of the Predictors in the Linear Mixed-Effect Model for Response Latencies in Experiment 2 Estimate Intercept (literal sentence, neutral

Std. Error

t-value

p-value

670.9

intonation, block 1) Previous RT (centered)

0.09

0.02

4.4