Lexical Disambiguation in Verb Learning: Evidence from the Conjoined-Subject Intransitive Frame in English and Mandarin Chinese

Arunachalam, Sudha; Syrett, Kristen; Chen, YongXiang

doi:10.3389/fpsyg.2016.00138

ORIGINAL RESEARCH article

Front. Psychol., 16 February 2016

Sec. Psychology of Language

Volume 7 - 2016 | https://doi.org/10.3389/fpsyg.2016.00138

Lexical Disambiguation in Verb Learning: Evidence from the Conjoined-Subject Intransitive Frame in English and Mandarin Chinese

$\r\nSudha Arunachalam*$ Sudha Arunachalam¹^*

Kristen Syrett²

YongXiang Chen^3,4

¹Department of Speech, Language, and Hearing Sciences, Boston University, Boston, MA, USA
²Department of Linguistics, Rutgers University-New Brunswick, New Brunswick, NJ, USA
³School of Education Science, Shanxi University, Taiyuan, China
⁴Key Laboratory of Behavioral Science, Institute of Psychology, Chinese Academy of Sciences, Beijing, China

When presented with a novel verb in a transitive frame (X is Ving Y), young children typically select a causative event referent, rather than one in which agents engage in parallel, non-causative synchronous events. However, when presented with a conjoined-subject intransitive frame (X and Y are Ving), participants (even adults, as we show) are at chance. Although in some instances, children older than three can obtain above-chance-level performance, these experiments still appear to rely upon a within-experiment contrast with the transitive frame. This leads us to ask whether children can achieve success with the intransitive frame without such a contrast among constructions, and map a novel verb appearing in such a frame onto a non-causative meaning. Building on recent evidence that adverbial modifiers can support word learning for adjectives and for verbs (when both nominal and verbal candidate interpretations are considered) by directing children to a particular construal of a scene, we test the hypothesis that a semantically informative modifier, together, will provide children with additional lexical information that allows them to narrow down verb meaning and identify a non-causative interpretation for a novel verb appearing in the conjoined-subject intransitive frame. We find that for English-speaking children and adults it does, but only when together directly modifies the verb phrase, suggesting that participants appeal to compositionality and not just the brute addition of another word, even one that is semantically meaningful, to arrive at the intended interpretation. Children acquiring Mandarin Chinese, in contrast, do not succeed with the translation-equivalent of together (although adult speakers do), but they do with dōu (roughly, the distributive quantifier “each”). Our results point to a valuable source of information young children learning verbs: modifiers with familiar semantics.

Introduction

Language learners the world over face the task of mapping unfamiliar words to meaning. Verbs are particularly difficult, given children's ontological expectation that new words might label objects (rather than events or properties) (e.g., Golinkoff et al., 1996; Waxman and Booth, 2001), the difficulty of identifying verb meanings simply by observing the world (e.g., Gleitman, 1990), and the relational foundations upon which verb meanings are built (e.g., Gentner, 1982). Correspondingly, experimental studies have shown that children more successfully identify novel noun meanings than novel verb meanings (e.g., Imai et al., 2008).

To overcome the challenges of acquiring verbs, children benefit from the principled relationship between the syntactic environments in which a verb appears and its semantic representation. Informative syntactic information includes the number and position of nouns (or NPs or DPs) appearing with the verb (e.g., Naigles, 1990; Fisher, 2002; Lidz et al., 2003; Gertner et al., 2006), and the range of syntactic frames in which it appears (e.g., Naigles, 1996; Gillette et al., 1999; Lee and Naigles, 2008). The syntactic bootstrapping hypothesis posits that observations of the syntactic environment in which a verb appears paired with knowledge of the syntax-semantics mapping can guide the child's verb learning process, and indeed syntactic bootstrapping abilities are well-established in young children (e.g., Landau and Gleitman, 1985; Naigles, 1990; Fisher et al., 1994; Fisher, 2002).

Much of the evidence documenting children's syntactic bootstrapping abilities with verbs have involved simple transitive frames, for which children map the NP/DP preceding the verb to an agentive semantic role and the NP/DP following the verb to a patient role, therefore interpreting the transitive frame as denoting a causative event. For example, in a now-classic study, Naigles (1990) presented 2-year-old English learners with a novel verb in a transitive frame (e.g., The duck is gorping the bunny) accompanied by a visual scene in which, simultaneously, a duck performed an action on a bunny (e.g., pushing into a bending position), and the duck and bunny each performed an action in parallel (e.g., circling their arms). At test, the events were teased apart: in one test scene, the duck performed the causative action on the bunny; in the other, both performed the synchronous (arm circling) action. Children were asked to “find gorping.” Naigles found that children who had heard the transitive frame preferred to look at the causative event over the scene with the non-causative, synchronous events. This finding has been replicated in numerous studies with children as young as 1.5 years (e.g., Naigles and Kako, 1993; Yuan and Fisher, 2009; Noble et al., 2011; Yuan et al., 2012; Arunachalam et al., 2013).

While children's understanding of the mapping between transitive frames and causative events is robust, the same cannot be said of all links between syntactic structure and semantic representations. Given that with the transitive frame, learners assign the NP/DP in the subject position an agentive role (presumably demonstrating their awareness that the preverbal argument is a subject/agent and the post-verbal argument is an object/patient), one might predict that with the conjoined-subject intransitive (e.g., The duck and the bunny are gorping) they would assign both of the NP/DP referents in subject position to agentive roles, and thus prefer the non-causative, synchronous events scene. However, children in these tasks do not consistently choose the synchronous scene (Naigles, 1990; Naigles and Kako, 1993; Hirsh-Pasek and Golinkoff, 1996; Arunachalam and Waxman, 2010; Noble et al., 2011).

This decalage between the transitive and intransitive frames also surfaces in the acquisition of Mandarin Chinese, interestingly so given that Mandarin (a Sino-Tibetan language) differs from English in relevant ways. For example, Mandarin verbs lack inflectional morphology indicating tense and aspect, and Mandarin allows a verb's arguments to be omitted when their referents are readily retrievable (e.g., Huang, 1984; Saito, 2007). Given the possibility of argument drop, both a transitive frame with its object dropped and an intransitive frame are realized as a subject NP or DP followed by a verb, thus rendering them indistinguishable on the surface. Mandarin learners might thus be predicted to make less use of morphosyntactic information in verb learning, as the surface properties of the sentence provide fewer cues. On the other hand, in languages that allow OV word order—possible but marked in Mandarin (Sun and Givón, 1985)—learners should have a weaker bias to map subjects to agents, and may thus find it easier to establish mappings between intransitive and non-causative referents (Chang et al., 2006). However, recent work finds that just like their English-acquiring counterparts, Mandarin-acquiring 2- and 3-year-olds take the transitive frame as a cue to causativity for both familiar verbs (Lee and Naigles, 2008) and novel verbs (Jiang and Haryu, 2014), and they are at chance in choosing between causative and synchronous scenes as referents for novel verbs in the conjoined-subject intransitive (Jiang and Haryu, 2014). Thus, at least with these two frames, Mandarin learners perform just as English learners do (see Matsuo et al., 2012 for similar results from learners of Japanese, which also permits argument drop).

This pattern raises two questions: why is the conjoined-subject intransitive frame challenging for young learners—robustly so across typologically different languages—and is there other linguistic information that might help to direct their attention to the intended interpretation? With respect to the first question, we suggest that the conjoined-subject intransitive frame is underinformative, providing insufficient semantic information for children to achieve the intended mapping given the experimental context. With respect to the second, we show that the addition of a single lexical item that highlights the semantic information that is underdetermined can support mapping to the intended, synchronous events referent.

Although we will argue that the frame was simply underinformative in the experimental contexts presented, another possibility is that children have not mastered the conjoined-subject intransitive structure (Noble et al., 2011; Gertner and Fisher, 2012), and that they instead use a strategy that does not require adult-like mapping of syntactic structure to semantic representation. For example, prior to age two, children may simply track the number and order of the nouns they hear, such that when presented with the intransitive frame, they map the first NP/DP to the Agent role and the second to the Patient role (Chang et al., 2006; Yuan and Fisher, 2009; Gertner and Fisher, 2012; Yuan et al., 2012; Messenger et al., 2014). However, at least by age two, children do not consistently select the causative scene as an option for both the transitive and intransitive frames, as would be predicted if they counted the nouns or consistently assigned the first NP/DP a causative agentive role in both cases. Instead, they typically perform at chance with the conjoined-subject intransitive, while consistently choosing the causative scene for the transitive frame (Arunachalam and Waxman, 2010; Noble et al., 2011). Thus, children's performance in these studies, at least for children over age two, is not likely due to systematic misinterpretation of the intransitive as a frame that signals a causative event.

Here, we adopt another perspective on the conjoined-subject intransitive, and in lieu of attributing children's performance to a lack of syntactic competence, we hypothesize that this frame provides insufficient semantic information for learners to select between the test scenes (Arunachalam and Syrett, 2014). The intransitive frame may be underinformative in at least two ways. First, it is compatible with two different construals, arising from the fact that a DP denoting a plurality occupies the subject position (e.g., Scha, 1981; Gillon, 1987; Landman, 1989a,b; Link, 1991; Verkuyl and van der Does, 1991). Under one interpretation of the conjoined subject, the collective interpretation, the property is predicated of the entire group (or sum) of participants, and there is one single, atomic event to which the verb refers, in which each member of the plurality is a participant. Under the second, distributive interpretation, the property is predicated of each event participant individually, and there are thus multiple sub-events within the larger event. Given a causative scene involving two event participants, and a non-causative scene in which two participants engage in identical, synchronous events, each scene might be a candidate for the verb in the conjoined-subject intransitive frame, but each with a different interpretation. The causative scene might be a better choice under a collective interpretation, in which both event participants, though engaged in different individual activities, may have a more general goal predicated of both. By contrast, the synchronous scene is a better choice under a distributive interpretation, because both event participants are engaged in the same lower-level action (e.g., waving), and each is therefore a participant in his/her own sub-event.

Second, a verb in a conjoined-subject intransitive frame is compatible with a range of meanings at multiple taxonomic levels. For example, it might mean something general like “playing,” “moving,” or “behaving nicely,” or it might denote a more specific action such as “waving”—which is typically the level of the intended referent in syntactic bootstrapping studies. Note that while children acquiring new nouns tend to map them at the “basic” category level (e.g., “dog”), rather than the subordinate (e.g., “beagle”), or superordinate (e.g., “animal”) level (Waxman and Markow, 1998), there is not a clear basic level for verbal event categories as there is for object categories (e.g., Huttenlocher and Lui, 1979; Maratsos and Deák, 1995). It is therefore not implausible that children might posit more general or superordinate meanings for novel verbs (Of course, this second kind of ambiguity is present for any frame, not just the conjoined-subject intransitive. A transitive frame such as, “The boy lorped the ball” in the context of a kicking event could be construed at multiple taxonomic levels, such as “punted,” “kicked,” or simply “touched.” Nevertheless, the primacy of causative events and the transitive-causative mapping in early childhood are well-established (e.g., Slobin, 1985; Naigles and Kako, 1993; Lidz et al., 2003; Bunger and Lidz, 2004), and so this problem is likely to be more acute for the intransitive frame).

If our claim is correct that children's difficulty with the conjoined-subject intransitive is due to the frame's underinformativity, we would predict that adults, too, would perform at chance in a task pitting causative and non-causative interpretations against each other with the intransitive frame; after all, they cannot be said to lack the requisite syntactic knowledge. For Mandarin, too, if a putative difficulty of using syntactic bootstrapping in Mandarin is not to blame for children's performance, Mandarin-speaking adults should perform at chance. We also predict that in the face of this indeterminancy, supplementing the frame with additional complementary semantic information that highlights the intended non-causative construal should increase preference for the synchronous events interpretation, for both children and adults, in both English and Mandarin. After all, even young children integrate semantic information with syntactic structure to converge on the most likely interpretation of utterances (e.g., Fisher et al., 1994). Thus, we suggest that one way in which learners might overcome the underinformativity of the intransitive frame is by attending to semantic information supplied by a co-occurring informative lexical item, which highlights the non-causative interpretation supported by the intransitive frame. Importantly, we claim that the lexical item alone is not doing all the heavy lifting: it is in virtue of its composition with the frame that it plays an important role in verb learning.

We hypothesize that adverbial modifiers may be particularly helpful in this regard, because they often carry important information about how an event unfolds, or how participants are engaged in the event. Previous research indicates that children can recruit the semantics of adverbial modifiers in adjective learning (Syrett and Lidz, 2010) and verb learning (Syrett et al., 2014). Relevant to the current goals, Syrett et al. (2014) found that the addition of the manner-of-motion adverb slowly to a transitive frame containing a novel verb (e.g., Let's see a boy and a balloon. He's gonna pilk it slowly) (a frame where children had otherwise performed at chance, because of the pronominal arguments) helped children map the verb to a meaning depicted by an appropriate motion event. Crucially, other modifiers of comparable or greater lexical frequency, or that that provided less specific information about manner of motion (right now, nicely) did not perform the same function (A similar pattern surfaced in Syrett and Lidz (2010) with adjectives). Syrett et al. (2014) concluded that the lexical semantics of slowly led children to focus their attention on the motion depicted in the event. Thus, the mere addition of another word is not enough; it must be semantically rich, sufficiently familiar to children, and directly informative about the context at hand.

Given these findings, we sought to identify an adverb that would highlight a distributive interpretation of the verb in the intransitive frame (one where each agent in subject position had the property predicated of her) and lead participants to select the scene depicting synchronous, non-causative events. We began by targeting together, which, under one interpretation, indicates that two events share spatial proximity and temporal contiguity (Lasersohn, 1998). Syrett and Musolino (2013, 2015) have found that 4-year-olds accept sentences with a plural subject and a VP modified by together (e.g., Two boys pushed a car together) in both collective and distributive contexts, indicating that young English-speaking children are aware of this meaning of together. If at age four, children allow for a distributive construal of sentences with together, their lexical entry for this modifier should support such a construal prior to age four, and specifically at the age range in the current study. Because plural subjects and conjoined subjects both denote a plurality, we predicted that conjoined-subject intransitive frames with together would support a distributive reading of the verb and direct participants to the scene in which two event participants engage in spatiotemporally coordinated (synchronous) actions. We further predicted that together would only support this interpretation if it served as a modifier within the verb phrase; simply hearing together alone, in a separate utterance, should not suffice.

Although together can highlight either collective or distributive interpretations, depending on the interpretations available in the context, notice that given this choice of contexts and with the addition of together to the intransitive frame, while the synchronous scene is an ideal instantiation of the distributive interpretation, the causative scene is not an ideal instantiation of a collective interpretation, in that it is not obviously the case that two agents are actively involved in a collective event—if there is a larger collaborative goal, it is certainly not salient. Thus, we predict that for both adults and children, together will highlight the distributive interpretation, leading them to view the scene with the synchronous events as the better referent.

Interestingly, our choice of adverb presents the opportunity to witness a cross-linguistic difference between English and Mandarin. To preview our results, we found that although together boosts performance of English-speaking children and adults, its translation-equivalent does not do so for Mandarin-speaking children—though it does for Mandarin-speaking adults. We suspected that children were simply not familiar enough with this word and its semantics to be able to benefit from it. We therefore targeted another lexical item for Mandarin learners: dōu is a modifier quantifier roughly equivalent to English all, every, or each (Cheng, 1995; Lin, 1998; Xiang, 2008) (although there are semantic reasons, which we will not discuss here, to think that it is not entirely equivalent to those universal quantifiers). We chose dōu because of independent evidence that by age 3–4 years, Mandarin preschoolers appreciate its universal quantificational import when it appears with a plural DP, and recognize that it can not only distribute a property over members of a set of individuals, but that it can be a quantificational adverb, distributing over events (Zhou and Crain, 2011), and by age four, they produce this term appropriately in their own utterances (Hsieh, 2008). We hypothesized that since preschool-age Mandarin speakers are aware of the meaning of dōu, just as preschool-age English speakers are aware of the meaning of together, and both terms invite the learner to look for a scene in which the predicate applies uniformly across event participants, both should affect interpretation of the intransitive frame and direct choices to the non-causative, synchronous scene. We found that, indeed, when presented with novel verbs in conjoined-subject intransitive frames with dōu, Mandarin-acquiring children also preferred the synchronous scene. Thus, it is a matter of finding the right modifier, given the language at hand.

Thus, in the current study, our goals were three-fold. First, we sought to see whether adults, like children, are at chance in choosing between causative and non-causative referents for novel verbs in the conjoined-subject intransitive, to see if child performance is more likely due to lack of syntactic competence or to the frame's lack of semantic informativity. We also aimed to replicate the previous findings with children in the same task. Second, we asked whether complementary semantic information in the form of an additional lexical item (a modifier) could help children and adults zero in on the intended (non-causative) interpretation of the verb. Finally, we asked how these aspects of verb learning play a role in English and in Mandarin, thereby highlighting both a commonality in the acquisition process across two very different languages, but also differences in the lexical information that matters.

To achieve these objectives, we used a design developed by Yuan and Fisher (2009), in which, unlike Naigles's (1990) classic paradigm, participants first hear a novel verb in (transitive or intransitive) sentences in the context of a conversation between two actors in the absence of any visual information about events. They then view two scenes side by side at test, and are asked to, “Find Ving.” To succeed, participants must have gleaned the relevant syntactic properties from the dialogues and applied that knowledge to identify which test scene best matched their syntactic/semantic representation; even before 2 years of age, children successfully map novel transitive verbs to causative events on the basis of these dialogues (e.g., Yuan et al., 2012; Arunachalam et al., 2013; Messenger et al., 2014).

Using this “dialogue” paradigm, we focused on participants' performance with the conjoined-subject intransitive, with and without an additional modifier, in English and Mandarin. For both English (Experiment 1) and Mandarin (Experiment 2), we first present the bare intransitive, no-modifier condition (Experiments 1a/2a). We then introduce a semantically informative modifier intended to direct participants' attention to the synchronous scene (Experiments 1b/2b-c). For English, we further manipulate in Experiment 1b whether the adverb modifies the VP directly or appears in a separate sentence, to determine whether the modifier enables success on its own or whether it does so only by virtue of its compositional relationship to the verb.

Experiment 1

In Experiment 1, we begin with English-speaking adults and children. In Experiment 1a we test the hypothesis that the intransitive frame itself provides insufficient information for choosing between causative and synchronous event referents for a novel verb given our experimental design, which was modeled after many previous tasks. In Experiment 1b, we ask whether adding a relevant adverbial modifier, together, increases preference for the synchronous events scene, and in addition, whether it does so regardless of whether it is semantically composed with the predicate or independent of the VP.

Experiment 1a

Methods

Participants

Participants were 15 native English-speaking adults, all students at Rutgers University—New Brunswick, and 20 children (10 males, 10 females). Two additional children were excluded from analysis for failure to provide clear pointing responses during training (see below). We targeted a relatively wide age range, from 2:4 to 3:11 (mean 3:0), because as reviewed above, previous syntactic bootstrapping work with children at about 2:3 has shown success with the transitive frame and chance performance with the conjoined-subject intransitive, and Noble et al. (2011) found that older 2-year-olds continued to perform at chance in their task, whereas 3-year-olds succeeded. We therefore included ages above and below this approximate point. Adults and older children (>3:2) were recruited from the Central NJ area, and younger children from the Boston, MA area. Populations were comparable in terms of SES. Parents signed an informed consent form approved by either Boston University or Rutgers University on behalf of their children.

Materials

The experiment consisted of four trials, each introducing a different novel verb. In each trial, participants were first exposed to a Familiarization Phase, during which they heard conversations between two actors viewed on the screen, which incorporated a novel verb eight times, always appearing in a conjoined-subject intransitive frame (see Table 1).

TABLE 1

Table 1. Familiarization Phase viewed by English participants in Experiments 1a, 1b (“VP Modifier” condition) and Mandarin participants in Experiments 2a, 2b.

The Test Phase presented two scenes side-by-side on a white background. One was a prototypical causative scene with an agent acting on a patient, and the other a non-causative scene with actors engaged in synchronous events. The left-right position of the causative and synchronous scenes was counterbalanced across trials. Participants were directed to make a selection between the scenes (e.g., Find biffing!) by pointing (children) or selection on a response sheet (adults) (see Table 2).

TABLE 2

Table 2. Example of a test phase viewed by English participants in Experiments 1a, 1b (“VP Modifier” condition) and Mandarin participants in Experiments 2a, 2b.

The novel verbs were monosyllabic words faithful to the phonotactic patterns of English, and that have been used in other novel verb learning studies: moop (/mup/), biff (/bIf/), lorp (/l rc p/), fez (/fεz/). Each involved different test scenes; two trials depicted people, while two other trials involved a person and an inanimate object (e.g., a ball) see Figure 1. These test scenes were modeled closely after prior work (Arunachalam and Waxman, 2010) to maximize the interpretability of the results. We included both animate and inanimate object trials both because Arunachalam and Waxman (2010) did and because Pozzan et al. (2015) found a difference between animate and inanimate trial types. However, we found no effects of animacy in our task; thus we only report analyses with trial type collapsed.

FIGURE 1

Figure 1. Still images from the test scenes from one trial, depicting an animate agent (the girl) and inanimate patient (the ball). Mandarin-speaking participants viewed videos of Chinese actors performing these same actions.

The visual stimuli in the Familiarization and Test phases were recorded in a quiet laboratory setting. The auditory stimuli in the Test phase were recorded by a female native speaker of American English in a recording booth, using a child-directed register. Utterances were edited, controlling for intensity, using Praat (Boersma and Weenink, 2005).

Procedure

Participants watched the video on a computer monitor. Children indicated their choice of scene during the Test phase by pointing. Adults were given a pamphlet, with a page for each trial, and were asked to circle “LEFT” or “RIGHT.” They were not allowed to revisit pages for completed trials.

Before the experimental session proper, children participated in two training trials, designed to encourage pointing. In each, children saw two dynamic scenes side-by-side. The first depicted two familiar characters, and the second depicted two activities. The experimenter asked, e.g., “Can you find dancing?” No novel words were used during training. Those who did not point unambiguously or pointed incorrectly on both training trials were replaced in the design. On experimental trials, no feedback was provided, though child participants received general encouragement during the task.

Coding and Analysis

Pointing responses were recorded in real time (for children), or reviewed later (for adults) by an experimenter naïve to study hypotheses. We coded which scene participants pointed to: 1 for a point to the (correct) synchronous scene, and 0 for a point to the causative scene. All participants pointed on at least three trials. Because there are two visual scenes, chance performance is 50%. We fitted the data—separately for adults and children—to a mixed-effects logistic regression model with a logit link (binomial family) with Subject and Item as random effects. We centered age around 0 by subtracting the mean age from each participant's age. Analyses were conducted using the glmer() function in R (v. 2.14.2) (Bates and Bolker, 2012; R Development Core Team, 2012). To test significance, we used the z-test and p-values output by glmer().

Results

Adults. Adults chose the synchronous scene 50% of the time. See Table 3 for means and standard deviations for all experiments. The intercept parameter estimate for the mixed-effects logistic regression model is 0.027, indicating that performance is not different from chance (probability of choosing the synchronous actions scene is 0.51). Parameter estimates for all models for all experiments are in Table 4.

TABLE 3

Table 3. Summary of participant responses (proportion of selection of Synchronous scene) across all experiments.

TABLE 4

Table 4. Parameter estimates from all models.

Children. Like the adult participants, and like children in many other previous studies, children performed at chance. We fitted the data to a model as with the adults, but given the relatively large age range we tested, we also included Age in months as a random effect. The intercept parameter estimate for the model is 0.089 (p = 0.84), indicating that children's performance did not differ from chance; Age did not contribute significantly (p = 0.87).

Discussion

This experiment replicated previous verb learning studies targeting the intransitive frame by showing that (in the absence of another other within-experiment information that could favor any particular response) children do not prefer the synchronous scene as a referent for a novel verb in a conjoined-subject intransitive frame. However, in this experiment, we also added adults, who also were at chance. The results thus support our hypothesis that chance-level performance with the conjoined-subject intransitive does not stem from lack of syntactic knowledge, and that the frame itself (at the very least in this experiment and others after which it was modeled) is underinformative—a finding consistent with one pilot study with adults (Sheline et al., 2013) and a possibility entertained by Arunachalam and Waxman (2010) and Noble et al. (2011). These results lay the foundation for Experiment 1b, in which we ask whether addition of the modifier together increases preference for the synchronous scene.

Experiment 1b

Experiment 1b included two conditions. Both introduced novel verbs in intransitive frames accompanied by together, but differing in its position. In the “VP Modifier” condition together appeared in the VP (e.g., The boy and the girl biffed together). In the “Predicative” condition together appeared in a separate sentence in predicative position following a copula (e.g., The boy and the girl biffed. They were together). This manipulation allowed us to test whether the mere presence of the modifier would support verb learning, or whether the modifier must compose with the verb to bring performance above chance level. The stimuli for the VP Modifier condition are included in Table 1. See Table 5 for the Predicative condition.

TABLE 5

Table 5. Dialogue presented to English participants during the Familiarization Phase of Experiment 1b (“Predicative” condition).