On the parity of structural persistence in language production and comprehension
Introduction
Fundamental to the explanation of how humans communicate is an understanding of the mental processes that support language comprehension and production. A crucial requirement of successful communication is that speakers and listeners can access similar information about words and how words combine to express an idea. Thus, English listeners understand English speakers because they make use of shared knowledge about English words and syntactic rules. The same listeners experience speech in unknown languages as meaningless streams of sound. The simple difference is that knowing the speaker’s language allows the listener to develop an idea that is similar enough to the speaker’s that communication occurs. How this happens remains a mystery. How can an idea become sound, and sound become a vestige of the same idea? In the current study, we tested a hypothesis about what the syntactic systems of language production and comprehension do to make this feat possible.
Given how little we know about the relationship between language comprehension and language production, the simplest workable alternatives are obvious: Speakers and listeners call on similar information in similar ways, or they call on similar information in different ways. The information must be similar or communication would founder, but how the information is used could be the same or different in the two modalities. A strong hypothesis is that speakers and listeners know similar things and use their knowledge in similar ways (Bresnan and Kaplan, 1984, Kempen et al., 2012, Pickering and Garrod, 2013, Sag and Wasow, 2011). Yet it is undeniable that listeners can understand words and sentences that they do not and perhaps cannot produce (Clark & Malt, 1984), that comprehension and production begin and end with different information, and that the peripheral sensory and motor apparatus for sensation and action are necessarily distinct. Even the creation of computational models in which comprehension and production call on the same information in the same ways is far from straightforward (Fodor, Bever, & Garrett, 1974). This argues for differences between listening and speaking.
Speculation aside, the debate is an empirical one with proponents and compelling evidence on both sides. Support for separable processing systems across comprehension and production comes from several areas of study, including the emergence of comprehension before production in language acquistion (Benedict, 1979, Gertner et al., 2006, Tomasello, 2000), the neuropsychological impairments that yield double dissociations between modalities in aphasia (Caramazza, 1997, Goodglass and Kaplan, 1972, Hillis and Caramazza, 1995, Linebarger et al., 1983), and the sheer difficulty of performance, with production seeming much harder (for instance, in driving; Lee and Watter, 2014, Recarte and Nunes, 2003). Nevertheless, there are counterarguments resting on evidence that is more consistent with substantial similarity across production and comprehension. In language acquisition, fine motor control may account for timing differences (Bonvillian et al., 1983, Petitto and Marentette, 1991). In neuropsychological impairments, deficits in general cognitive resources might obscure underlying uniformity (Caplan, 1996, Caplan and Waters, 1995, Caplan et al., 2007). Apparent differences in difficulty could stem from people’s typical failure to create as much representational detail after listening as before speaking, even though such detail is achievable (Bock et al., 2007, Kempen et al., 2012).
Particularly compelling observations about the relationship between comprehension and production come from situations where the two modalities continuously interact, like self-monitoring and conversation. Self-monitoring of one’s own speaking and listening could depend on tight coupling between comprehension and production (Garrett, 1980, Levelt, 1983, Levelt, 1989, Townsend and Bever, 2001), just as seamless interaction between comprehension and production is a necessity for coherent conversation. In both of these instances, episodes of comprehension may have an immediate impact on upcoming production, and vice versa. Garrod and Pickering (2004) described the mutuality between comprehension and production as a progressive process of alignment between conversation partners. As conceived, alignment means that speakers and listeners develop the same linguistic representations for many kinds of referring expressions at many levels (Brennan and Clark, 1996, Watson et al., 2004), including syntactic structure (Branigan, Pickering, & Cleland, 2000).
The linkage between language comprehension and production is a focus of current research on structural priming and persistence. Structural priming (incidental experience with a syntactic structure) and structural persistence (incidental adaptation to the same structure) have consequences for both speakers and listeners. (Note our use of the term priming to refer to experience with a structure and persistence to refer to structural consequences of that experience.) Whether the consequences or the mechanisms of priming are the same is a matter of debate. In the next two sections we consider the implications for this debate of existing findings about structural persistence.
What is structural persistence? Descriptively, structural persistence is the product of a structure-specific influence of an experienced syntactic pattern on later episodes of comprehension and production. It can arise even when lexical, semantic, and thematic information differ between a priming exposure and subsequent encounters with, or uses of, the same structure. For example, speakers who say The 747 was landing by the control tower are later on more likely to say The mailman is being chased by a dog than they would otherwise be, using a passive structure in the ensuing sentence even when its voice, topic, and just about everything else changes (Bock, 1986, Bock, 1989, Bock and Loebell, 1990).
Persistence of structure in language production has been observed for several kinds of structures in different languages (Bock, 1986, Bock and Loebell, 1990, Cleland and Pickering, 2003, Corley and Scheepers, 2002, Hartsuiker and Kolk, 1998, Hartsuiker and Westenberg, 2000, Konopka and Bock, 2009, Pickering and Branigan, 1998, Scheepers, 2003), in young children (Huttenlocher et al., 2004, Savage et al., 2003, Shimpi et al., 2007), in spontaneous speech (Gries, 2005), and in bilinguals, across their languages (Hartsuiker et al., 2004, Loebell and Bock, 2003, Shin and Christianson, 2009). Most important for present purposes is that persistence in production arises regardless of whether priming occurs in an episode of language production or language comprehension (Branigan et al., 2000, Lombardi and Potter, 1992, Potter and Lombardi, 1998), with the same strength and duration (Bock, Dell, Garnsey, Kramer, & Kubose, 2007). That is, persistence in language production is a cross-modality phenomenon.
The occurrence of structural persistence between prime and target structures, without other shared information, is one of its theoretically most provocative features. What seems to persist is an abstract syntactic process or representation. Yet when information overlap is present, when specific words recur in specific structures, there is an increase in the magnitude of persistence (Cleland and Pickering, 2003, Pickering and Branigan, 1998). The increase has been demonstrated chiefly (but not exclusively) with the repetition of verbs, which play a prominent part in the syntax of a sentence. This lexically dependent effect, often called the lexical boost, implies the interaction of a specific word effect with a general syntactic effect.
The lexical boost suggests that individual words can strongly influence syntactic processes. Curiously, though, the impact of lexical repetition on persistence has been found to differ in the priming of language production by earlier episodes of production compared to the priming of comprehension by earlier episodes of comprehension. In production, abstract and lexically boosted persistence are both well attested; in comprehension, lexically boosted persistence alone is commonly observed.
Research on structural priming in comprehension suggests that, as in production, the effect of understanding a particular structure persists in a way that influences subsequent comprehension of the same structure (Arai et al., 2007, Carminati et al., 2008, Thothathiri and Snedeker, 2008a, Thothathiri and Snedeker, 2008b, Tooley et al., 2009, Traxler, 2008, Traxler and Tooley, 2008). For example, after reading a sentence with a reduced relative clause (e.g. The man watched by the woman was tall and handsome), subsequent reduced-relative sentences are read faster (Traxler & Tooley, 2008) and with less disruption (reflected in smaller P600s in measures of event-related potentials; Ledoux et al., 2007, Tooley et al., 2009).
Similar results have been found in auditory comprehension with different forms of dative sentences (prepositional-object and double-object structures like The boy gave a flower to the girl and The boy gave the girl a flower, respectively; Arai et al., 2007, Carminati et al., 2008). When listeners performed a task in which scenes corresponding to such sentences were displayed while the sentences were presented, participants showed different patterns of anticipatory eye movements to individual objects depending on which structure was primed. For instance, after double-object priming, listeners who heard “The boy gave” tended to look at the picture of the girl sooner than the picture of the flower. This suggests that the double-object form persisted in a way that biased later parsing of dative sentences toward post-verbal indirect objects (e.g. the girl) rather than direct objects (e.g. the flower after The boy gave…).
The distinctive feature of structural persistence in language comprehension is that the reliability of the effects often hinges on the prime and target sentences having the same verb. So, to see persistence after the reduced-relative prime The man watched by the woman was tall and handsome, the target should include the verb watch (e.g. The mouse watched by the cat hid under the table). Likewise, for datives like the pair above, repetition of the verb give is typically needed for persistence to be observable. With different prime and target verbs, the evidence for persistence in comprehension is scarce (though not completely absent; see Pickering et al., 2013, Thothathiri and Snedeker, 2008a, Thothathiri and Snedeker, 2008b, Traxler, 2008; see also Scheepers & Crocker, 2004 for evidence of constituent order priming in comprehension). This casts doubt on the reliability of abstract persistence of structure on its own.
The centrality of lexical support to structural persistence in comprehension points toward a crucial contrast with language production. In production, abstract persistence not only occurs in the absence of lexical support, but can also last for a long time (Bock and Griffin, 2000, Kaschak, 2007, Kaschak et al., 2011, Luka and Choi, 2012). Yet the increased effect of structural priming that stems from lexical repetition does not last, instead declining quickly (Hartsuiker et al., 2008, Konopka and Bock, 2005). This means that persistence in production cannot depend on lexical support. For comprehension, lexically boosted persistence has been shown to survive across two intervening sentences (Pickering et al., 2013; Tooley, Swaab, Boudewyn, Zernstein, & Traxler, 2014), though its longer-term impact is unknown.
The contrast between production and comprehension in the abstractness of persistence hints at fundamental differences in processing. In language production, speakers can start out with an abstract, rudimentary syntactic structure to which words are bound as the structure unfolds (Bock, Ferreira, M. Goldrick, & The Oxford handbook of language production. Oxford, 2014). In comprehension, because listeners encounter words one at a time, the structural information used for parsing can come from specific-word information in the mental lexicon (e.g., Boland and Boehm-Jernigan, 1998, Garnsey et al., 1997, Trueswell et al., 1994). This is consistent with lexicalist models of language comprehension (MacDonald, Pearlmutter, & Seidenberg, 1994). In essence, language production may build on abstract structure while language comprehension builds on words.
This intuitively attractive and theoretically profound claim is consistent with most existing findings about structural persistence in language production and comprehension. Yet despite its appeal, the claim remains tenuous. Its support comes from contrasts between production and comprehension priming in distinct experiments where the initial priming experiences, and just about everything else, differ. There are differences in prime presentation techniques, in tested structures, and in the depth of processing required by the tasks. In comprehension, structural processing is assessed in complex structures and measured as it proceeds, calling on reading times, modulations in the EEG signal, or anticipatory eye movements to visual referents. In production, structural formulation is most often assessed with simple sentence structures and measured from the outcomes of picture description and sentence completion. The upshot is that existing research on structural persistence offers precious little evidence for fundamental differences in the syntactic components of speaking and understanding. What evidence there is comes from experiments that lack an essential element: A manipulation of modality.
A more compelling test of the hypothesis that different structural processes are at work requires a more persuasive equation of the priming that precedes language production and comprehension. In the present experiment, we assessed structural persistence using the same priming procedure, the same sentences and sentence structures, and the same participants, at the same time.
In order to create a priming environment that was the same across modalities, we used a modified version of a task introduced by Potter and Lombardi (1990). The task splits prime exposure into three parts. First, participants are shown a word-by-word, rapid serial visual presentation (RSVP) of a prime sentence. Then they complete a short distractor task, which serves to disrupt verbatim memory of the sentence while keeping the message intact. Third and finally, participants try to repeat the priming sentence aloud. This prime trial is followed by an identical three-part trial in which a target sentence is presented in RSVP. The structures of the RSVP prime and target sentences are manipulated to have either matching or mismatching syntactic structures, drawn from structural paraphrases such as the dative alternation. Persistence of structure is indicated when participants are more likely to produce the target sentence using syntax that matches, rather than mismatches, the syntax of the prime sentence. This task successfully elicits structural persistence after priming from comprehension or production (Konopka and Bock, 2009, Lombardi and Potter, 1992, Potter and Lombardi, 1998).
In our adaptation, the comprehension and production priming procedures were equivalent up to the point when target sentences were produced (spoken aloud) or comprehended (read silently). Production and comprehension trials were interleaved pseudorandomly, to ensure comparability in expectations and preparations for speaking or reading. The conditions thus equated the effort and depth of prime processing for both modalities, in addition to equating the priming and target sentences themselves.
The hypothesis that structural processes are more lexically driven and less abstract in comprehension than in production would get support from a difference in the effects of lexical repetition on structural persistence. Specifically, if persistence in comprehension depends on lexical overlap between prime and target sentences, while persistence in production does not, the existence of a real disparity in the mechanisms of structural processing becomes more likely. Furthermore, if the magnitude of persistence is demonstrably larger in one modality than the other, the probability of modality differences in the stability or the demands of structural processing increases. If, however, abstract structural persistence emerges in both modalities, at similar magnitudes, there is a stronger case to be made for parity in the structural processes of language comprehension and production.
Section snippets
Participants
The participants were 286 students and community members from the University of Illinois who were compensated with either course credit or $10. All participants were native speakers of English with normal or corrected-to-normal vision. Of the 286 participants, 30 were excluded from analyses due to (a) learning a different language prior to learning English (n = 1), (b) computer error (n = 2), or (c) failing to produce enough scorable responses (n = 27). The 256 remaining participants were included in
Results
Fig. 3 shows the broad patterns of persistence in comprehension and production. The effects in comprehension are given as the difference between primed and unprimed reading times, and in production as the proportions of primed targets produced. The height of the bar for each modality represents total persistence. As the figure shows, structural persistence occurred in both modalities, whether the prime and target verbs were the same or different. The details for calculating and statistically
Discussion
This experiment constituted a large-scale controlled comparison between sentence comprehension and production in how structural priming plays out. In both modalities, structural priming yielded structural persistence. The persisting structures were abstract, in the sense that persistence emerged regardless of modality and, more important for present purposes, regardless of whether the verbs in the prime and target sentences were the same. Verb repetition modestly increased the amount of
Acknowledgements
We thank the members of the Language Production Laboratory at the University of Illinois for their advice, Natalia Wolosowicz and Javier Ospina for their contributions to data collection, and the National Institutes of Health (T32HD055272) and National Science Foundation (BCS0843866) for their support.
References (97)
- et al.
Priming ditransitive structures in comprehension
Cognitive Psychology
(2007) Syntactic persistence in language production
Cognitive Psychology
(1986)- et al.
Framing sentences
Cognition
(1990) - et al.
Lexical constraints and prepositional phrase attachment
Journal or Memory and Language
(1998) - et al.
Syntactic co-ordination in dialogue
Cognition
(2000) - et al.
Little houses and casas pequeñas: Message formulation and syntactic form in unscripted speech with speakers of English and Spanish
Cognition
(2008) - et al.
Watching the eyes when talking about size: An investigation of message formulation and utterance planning
Journal of Memory and Language
(2006) - et al.
A study of syntactic processing in aphasia I: Behavioral (psycholinguistic) aspects
Brain and Language
(2007) - et al.
Thematic roles assigned along the garden path linger
Cognitive Psychology
(2001) - et al.
The use of lexical and syntactic information in language production: Evidence from the priming of noun phrase structure
Journal of Memory and Language
(2003)
Conflicting cues and competition between notional and grammatical factors in producing number and gender agreement: Evidence from Hebrew
Journal of Memory and Language
The contributions of verb bias and plausibility to the comprehension of temporarily ambiguous sentences
Journal of Memory and Language
Why is conversation so easy?
Trends in Cognitive Sciences
Syntactic priming persists while the lexical boost decays: Evidence from written and spoken dialogue
Journal of Memory and Language
Word order priming in written and spoken sentence production
Cognition
Alignment as a consequence of expectation adaptation: Syntactic priming is affected by the prime’s prediction error given both prior and recent experience
Cognition
Lexical or syntactic control of sentence formulation? Structural generalizations from idiom production
Cognitive Psychology
Monitoring and self-repair in speech
Cognition
Sensitivity to grammatical structure in so-called agrammatic aphasics
Cognition
The regeneration of syntax in short term memory
Journal of Memory and Language
Dynamic grammar in adults: Incidental learning of natural syntactic structures extends over 48 h
Journal of Memory and Language
Subject–verb agreement processes in comprehension
Journal of Memory and Language
Agreement processes in sentence comprehension
Journal of Memory and Language
The representation of verbs: Evidence from syntactic priming in language production
Journal of Memory and Language
Regeneration in the short-term recall of sentences
Journal of Memory and Language
Syntactic priming in immediate recall of sentences
Journal of Memory and Language
Lexical and phonological effects on syntactic processing: Evidence from syntactic priming
Journal of Memory and Language
Syntactic priming of relative clause attachments: Persistence of structural configuration in sentence production
Cognition
The representation of lexical and syntactic information in bilinguals: Evidence from syntactic priming
Journal of Memory and Language
Syntactic priming and the lexical boost effect during sentence production and sentence comprehension: An fMRI study
Brain and Language
Syntactic processing in Korean–English bilingual production: Evidence from cross-linguistic structural priming
Cognition
On the interpretation of the number attraction effect: Response time evidence
Journal of Memory and Language
Response time distributional evidence for distinct varieties of number attraction
Cognition
Give and take: Syntactic priming during spoken language comprehension
Cognition
Syntactic priming during language comprehension in three and four year old children
Journal of Memory and Language
Do young children have adult syntactic competence?
Cognition
Semantic influences on parsing: Use of thematic role information in syntactic ambiguity resolution
Journal of Memory and Language
Agreement attraction in comprehension: Representations and processes
Journal of Memory and Language
Early lexical development: Comprehension and production
Journal of Child Language
Toward a cognitive psychology of syntax: Information processing contributions to sentence formulation
Psychological Review
An effect of the accessibility of word forms on sentence structures
Journal of Memory and Language
Closed-class immanence in sentence production
Cognition
Structure in language: Creating form in talk
American Psychologist
Syntactically speaking
Persistent structural priming from language comprehension to language production
Cognition
Car talk, car listen
The persistence of structural priming: Transient activation or implicit learning?
Journal of Experimental Psychology: General
Broken agreement
Cognitive Psychology
Cited by (62)
Do particle verbs share a representation with their root verbs? Evidence from structural priming
2021, Journal of Memory and LanguageCitation Excerpt :As this is an abstract learning procedure based on the statistical regularities of linguistic processing including production and comprehension, our theory predicts an independent but connected lemma representation for particle verbs and their root verbs that is used in both top-down (production) and bottom-up processing (comprehension). This cross-modalities hypothesis is also consistent with evidence from previous studies that found structural priming effects within or between comprehension and production (Bock, 1986b; Bock, Dell, Chang, & Onishi, 2007; Branigan et al., 2005; Thothathiri & Snedeker, 2008; Tooley & Bock, 2014). Thus, it will be interesting to investigate whether the dual lemma model holds for comprehension.
Monolingual and bilingual logical representations of quantificational scope: Evidence from priming in language comprehension
2021, Journal of Memory and LanguageLingering misinterpretation in native and nonnative sentence processing: Evidence from structural priming
2021, Applied Psycholinguistics
- 1
Present address: Texas State University, 601 University Drive, San Marcos, TX 78666, United States. Tel.: +1 512 245 4384.