No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Phrasal Ordering Constraints in Sentence Production: Phrase Length and Verb Disposition in Heavy-NP Shift
Abstract
Heavy-NP shift is the tendency for speakers to place long or "heavy" noun phrase direct objects at the end of a sentence rather than in the canonical post-verbal position. Three experiments using several task variations confirmed that length of the noun phrase influenced the ordering of the noun phrase and prepositional phrase during production. We also found that heavy-NP shift was strongly constrained by the "shifting disposition" of individual verbs. Verbs that do not require their complements (e.g., sentential complements) to appear in an adjacent position yielded more shifting during production than verbs that more frequently appear adjacent to their complements. Analyses of decision/preparation times suggested that shifted and unshifted structures competed for selection. These findings point to the simultaneous activation of lexically derived syntactic representations and ordering options in sentence planning. A multiple constraints framework provides a means of reconciling the e...
... They could, in particular, initially activate multiple sentence plans that all start with an unmarked noun phrase and only later decide on the specific plan (cf. Myachykov, Scheepers, Garrod, Thompson, & Fedorova, 2013;Stallings, MacDonald, & O'Seaghdha, 1998). Momma (2021) demonstrates that the elements of a dependency are planned together. ...
... As long as speakers have not yet committed to a sentence structure, syntactic alternatives could be activated and prepared in parallel (Dell & O'Seaghdha, 1994;Myachykov et al., 2013). While this provides more flexibility in planning, e.g., to accommodate attention or accessibility fluctuations (Ferreira, 1996;Velde & Meyer, 2014;Wagner, Jescheniak, & Schriefers, 2010), it likely also requires speakers to simultaneously maintain multiple utterance plans or to handle potential competitor plans (Hwang & Kaiser, 2014a;Myachykov et al., 2013;Stallings et al., 1998). ...
... This could be achieved through the increased activation of the cortical networks that are implicated in syntactic information processing (Davidson & Indefrey, 2007;Vassileiou, Meyer, Beese, & Friederici, 2018), resulting in alpha-band desynchronization. This interpretation is also corroborated by the reaction time results from the current study: Basque intransitive sentences were initiated slower than ergative-marked transitive sentences, consistent with findings from the literature implicating higher processing loads for the planning of sentences under more flexible conditions (Hwang & Kaiser, 2014a;Myachykov et al., 2013;Stallings et al., 1998). ...
Languages differ in how they mark the dependencies between verbs and arguments, e.g., by case. An eye tracking and EEG picture description study examined the influence of case marking on the time course of sentence planning in Basque and Swiss German. While German assigns an unmarked (nominative) case to subjects, Basque specifically marks agent arguments through ergative case. Fixations to agents and event-related synchronization (ERS) in the theta and alpha frequency bands, as well as desynchronization (ERD) in the alpha and beta bands revealed multiple effects of case marking on the time course of early sentence planning. Speakers decided on case marking under planning early when preparing sentences with ergative-marked agents in Basque, whereas sentences with unmarked agents allowed delaying structural commitment across languages. These findings support hierarchically incremental accounts of sentence planning and highlight how cross-linguistic differences shape the neural dynamics underpinning language use.
... For the language task, we modified an experimental paradigm designed to evaluate the rates of production of a sentence type called Heavy Noun Phrase Shift (HNPS; Stallings et al., 1998). HNPS is the tendency for speakers to produce long or "heavy" noun phrase (NP) direct objects at the end of a verb phrase rather than immediately following the verb (Stallings et al., 1998;Wasow & Arnold, 2003). ...
... For the language task, we modified an experimental paradigm designed to evaluate the rates of production of a sentence type called Heavy Noun Phrase Shift (HNPS; Stallings et al., 1998). HNPS is the tendency for speakers to produce long or "heavy" noun phrase (NP) direct objects at the end of a verb phrase rather than immediately following the verb (Stallings et al., 1998;Wasow & Arnold, 2003). For example, in utterances that consist of a subjectverb phrase, a direct object NP, and a PP, English speakers tend to prefer to place the object NP before the PP (e.g., I explained the plans to Michael) even though the alternative (e.g., I explained to Michael the plans) is also acceptable in English. ...
... However, with very long object NPs, speakers become more likely to produce the "shifted" form with the PP-first (as in I explained to Michael the extravagant plans for the new amusement park near the interstate). This pattern has been attested in both spontaneous speech and written corpora (Hawkins, 1994;Wasow & Arnold, 2003) and in speech elicitation experiments in the laboratory (Stallings et al., 1998). ...
Hysteresis in motor planning and syntactic priming in language planning refer to the influence of prior production history on current production behaviour. Computational efficiency accounts of action hysteresis and theoretical accounts of syntactic priming both argue that reusing an existing plan is less costly than generating a novel plan. Despite these similarities across motor and language frameworks, research on planning in these domains has largely been conducted independently. The current study adapted an existing language paradigm to mirror the incremental nature of a manual motor task in order to investigate the presence of parallel hysteresis effects across domains. We observed asymmetries in production choice for both the motor and language tasks that resulted from the influence of prior history. Further, these hysteresis effects were more exaggerated for subordinate production forms implicating an inverse preference effect that spanned domain. Consistent with computational efficiency accounts, across both tasks participants exhibited reaction time savings on trials in which they reused a recent production choice. Together, these findings lend support to the broader notion that there are common production biases that span both motor and language domains.
... crouches/plants trees") [7,8]. At the same time, this intermittent compatibility with multiple sentence plans might also require that 2 possible options need to be kept distinct while speakers construct a syntactic plan [9,10]. ...
... Speakers were thus able to primarily fixate on the agent as the action initiator [63] and as the referent of the first noun phrase in the sentence. In this way, speakers delayed the completion of relational processing [46] and held available all (aligned) structures that were compatible with a sentence starting with an unmarked agent noun phrase [10]. This scenario yields a natural interpretation of the increased theta ERS in the aligned condition as reflecting the simultaneous and noncommittal engagement of multiple compatible structures. ...
... This can be achieved by increased active neural information processing [29,75,76], e.g., in the form of increased engagement of cortical networks that are involved in processing syntactic information [40,42]. Converging evidence for our interpretation of the alpha ERD effect comes from reaction time findings that suggest higher processing loads when speakers need to separate alternative plans during sentence planning [9,10]. The need to keep distinct overlapping unmarked structures might also have contributed to the increased fixations to unmarked agents (Fig 2A) as longer fixation durations can indicate increased processing [49,77]. ...
Planning to speak is a challenge for the brain, and the challenge varies between and within languages. Yet, little is known about how neural processes react to these variable challenges beyond the planning of individual words. Here, we examine how fundamental differences in syntax shape the time course of sentence planning. Most languages treat alike (i.e., align with each other) the 2 uses of a word like “gardener” in “the gardener crouched” and in “the gardener planted trees.” A minority keeps these formally distinct by adding special marking in 1 case, and some languages display both aligned and nonaligned expressions. Exploiting such a contrast in Hindi, we used electroencephalography (EEG) and eye tracking to suggest that this difference is associated with distinct patterns of neural processing and gaze behavior during early planning stages, preceding phonological word form preparation. Planning sentences with aligned expressions induces larger synchronization in the theta frequency band, suggesting higher working memory engagement, and more visual attention to agents than planning nonaligned sentences, suggesting delayed commitment to the relational details of the event. Furthermore, plain, unmarked expressions are associated with larger desynchronization in the alpha band than expressions with special markers, suggesting more engagement in information processing to keep overlapping structures distinct during planning. Our findings contrast with the observation that the form of aligned expressions is simpler, and they suggest that the global preference for alignment is driven not by its neurophysiological effect on sentence planning but by other sources, possibly by aspects of production flexibility and fluency or by sentence comprehension. This challenges current theories on how production and comprehension may affect the evolution and distribution of syntactic variants in the world’s languages.
... A number of corpus studies looking at the dative alternation and heavy NP shift in English (Wasow and Arnold, 2003;Wasow, 1997a,b) have shown that the likelihood of a verb appearing in the double object structure or the prepositional object structure is indicative of the orderings observed. In a series of production studies, Stallings et al. (1998) demonstrated that speakers tend to use heavy NP shift with a verb like explain in comparison to a verb such as release. This is because explain more regularly takes a sentential complement that is not typically an NP, which prefers to be adjacent to the verb, leading the NP to be shifted. ...
... I consider the direct object NP to be more argument-like (Merlo and Ferrer, 2006) in comparison to the PP. By the Principle of Argument Closer (see Section 2.2.2) (Culicover and Jackendoff, 2005;Staub and Clifton, 2006;Stallings et al., 1998), the structure of (10) will be more preferred to that of (11). ...
... Since the focus was on how specific verbs affect ordering preference extremity, I estimated the effect of verb bias (Wasow and Arnold, 2003;Stallings et al., 1998) as the probability of a sentence being realized as the prepositional object structure predicted from just the random effect of the verb (eliminating the contributions from the fixed effects). Specifically, let V be the random effect intercept of a particular verb derived from the regression model. ...
Why are languages the way they are? In this dissertation, I take up this question with
a focus on crosslinguistic constituent orderings. Specifically, borrowing insights from
language processing and language evolution, I ask what abstract constraints as well as
idiosyncratic biases govern language users’ choice among grammatical alternatives of the
same syntactic constructions across genres and languages. Adopting a data-driven ap-
proach, I explore three directions in particular. First, from Chapter 3 to Chapter 6,
taking advantage of large-scale multilingual corpora, I investigate and quantify the roles
of numerous factors that are motivated by long-standing linguistic theories as well as previ-
ous empirical findings in word order preferences. I show that while the effect of individual
factors depends on the ordering structures of different languages, generally the predictive
power and direction of these constraints are more dependent on whether the orderings
are in the preverbal or the postverbal domains. In addition, besides these abstract con-
straints that yield probabilistic typological tendencies, in Chapter 7 I ask why language
users have idiosyncratic ordering preferences and how regularization of this idiosyncrasy
arises diachronically, using Bayesian iterated learning models that simulate the process
of language change. Lastly, I adopt the theoretical framework of dependency syntax to
develop a dependency treebank for Hupa, an endangered Dene language of northwestern
California, as a way to formalize and model the syntax of indigenous languages.
... As discussed above, Wasow (1997) shows that optionally transitive verbs are more likely to occur in HNPS constructions than those which are obligatorily transitive. In a related finding, Stallings et al. (1998) show that verbs which can take either NP or S complements are more acceptable in HNPS constructions as compared to verbs which only take NP complements, arguing that the frequent occurrence of V-PP/Adv-S word order for these verbs facilitates the processing of HNPS (see also Staub et al. 2006). ...
... Nevertheless, end-weight, now typically formalized as relative constituent length, remains a central factor for any understanding of HNPS (Wasow and Arnold 2003). Note, however, that the evidence given in support of the research cited above is largely drawn from corpora or speech production (possibly after a planning period, such as Stallings et al. 1998), prompting the comprehension-oriented study described in the following section. ...
... (2) a. Alex indicated to Curtis the problems. All verbs in the test items subcategorize for both NP and S, which is known to increase HNPS acceptability (Stallings et al. 1998). We also limit all potential interveners to two-word PPs, which should again increase overall acceptability (as compared to longer interveners) (Wasow 1997). ...
This squib tests theories of heavy NP shift that link constituent order to parsing. Our results indicate that increasing the weight of an object NP cannot make a heavy NP shift construction more acceptable than a comparison sentence in the canonical order, contra parsing-based theories. In addition, we examine acceptability as it relates to verb disposition. We find no significant differences between different subcategorizations, contrary to the findings of corpus studies. These results reveal a disconnect between production and comprehension; we further conclude that parser sensitivity to constituent structure is unlikely to affect speaker production of heavy NP shift.
... These same contingencies also appear to shape sentence planning. For example, in language production, the frequencies with which verbs appear in alternative syntactic contexts has consequences for sentence production choices of sentences containing those verbs (Arnold, Wasow, Asudeh & Alrenga, 2004;Bernolet & Hartsuiker, 2010;Stallings et al., 1998) as do the distributional pairings between noun animacy and sentence structure (Bresnan & Ford, 2010;Reali & Christiansen, 2007;Gennari & MacDonald, 2009). Like comprehenders, language producers implicitly learn statistical patterns of their linguistic environment, and this information affects production choices and accuracy (Boyd & Goldberg, 2011;Chang, 2009;Dell, Reed, Adams & Meyer, 2000;Warker & Dell, 2006). ...
... This type of lexico-syntactic learning also underlies a variety of verb-bias effects (F. Ferreira, 1994;Stallings et al., 1998;in comprehension: Garnsey et al., 1997;Trueswell, Tanenhaus & Kello, 1993) and generally findings that certain words are more likely to occur in some contexts than other words. ...
... In sum, speakers seem to learn fine-grained patterns regarding the use of different words and structures to describe different events and event participants, so that subtle aspects of event adversity, noun animacy, and even aspects of verb type (F. Ferreira, 1994), individual verbs (Stallings et al., 1998), and word co-occurrences (Wasow, 1997) affect the tendency to produce particular lexico-syntactic combinations to describe different events. However, there also appears to be generalization from one neighborhood to another, where neighborhood here can be seen both as structure (main clause and relative clause) and event-structure, where intransitive adverse events may generalize to transitive ones in Japanese. ...
Cross-linguistic studies allow for analyses that would be impossible in a single language. To better understand the factors that underlie sentence production, we investigated production choices in main and relative clause production tasks in three languages: English, Japanese and Korean. The effects of both non-linguistic attributes (such as conceptual animacy) and language specific properties (such as word order) were investigated. Japanese and Korean are structurally similar to each other but different from English, which allowed for an investigation of the production consequences of non-linguistic attributes in different typological or word order contexts (when Japanese and Korean speakers make similar production choices that are unlike those of English speakers), as well as production choices that differ despite typological similarity (when Japanese and Korean speakers make different choices). Speakers of all three languages produced more passive utterances when describing animate entities, but the overall rate of passives varied by task and language. Further, the sets of items that were most likely to elicit passives varied by language, with Japanese and Korean speakers more likely to produce passives when patients were adversely affected by the depicted event. These results suggest a number of factors that contribute to language production choices across three languages, and how general cognitive constraints on sentence production may interact with the structure of a specific language.
... De variabelen die deel uitmaken van de eerste drie groepen, die ook in mindere of meerdere mate in het empirische onderzoek van Jansen (1978van Jansen ( , 1979 en Braecke (1990) aan bod komen, worden in deze studie op verschillende manieren geoperationaliseerd. De laatste groep grammaticale en semantische classificatie van het (hoofd)werkwoord is nieuw, maar komt wel meermaals aan bod in andere onderzoeken naar syntactische variatiefenomenen in het Nederlands en het Engels (o.a., Stallings, MacDonald & O'Seaghdha 1998;Bresnan et al. 2005;Bresnan & Ford 2010, Grondelaers et al. 2009, Hoeksema 2014. De operationalisatie van het werkwoord kan ook implicaties hebben voor de neerlandistiek en meer bepaald voor de verklaringen in de ANS. ...
... Dat is uiteraard een assumptie -er zijn tot op heden geen multifactoriële studies over de plaatsing van VzC's in gesproken Nederlands -maar wel een beredeneerde assumptie: veel van de variabelen die in het vorige hoofdstuk significant bleken te zijn, zijn ook significant voor andere types van woordvolgordevariatie, zowel in het Nederlands als in het Engels (bv. syntactische complexiteit in de Engelse Heavy NP shift (Stallings, MacDonald & O'Seaghdha 1998) of in Particle Placement (Hawkins 1994;Gries 2002a); discursieve toegankelijkheid in de aan-vs. afwezigheid van er (Grondelaers en Speelman 2007); definietheid bij het Nederlandse woordvolgordefenomeen scrambling (Van Bergen & de Swart 2010); semantische klasse en de werkwoorden in de Engelse Dative Alternation (Bresnan et al. 2005 )), zowel in geschreven als in gesproken taal (bv. ...
... afwezigheid van er (Grondelaers en Speelman 2007); definietheid bij het Nederlandse woordvolgordefenomeen scrambling (Van Bergen & de Swart 2010); semantische klasse en de werkwoorden in de Engelse Dative Alternation (Bresnan et al. 2005 )), zowel in geschreven als in gesproken taal (bv. de Heavy NP shift in gesproken taal (Stallings, MacDonald & O'Seaghdha 1998) en in geschreven taal ). Onder die assumptie is het interessant vast te stellen dat het aantal korte(re) VzC's in Braeckes onderzoek (naar gesproken Nederlands) relatief groter is dan het aantal in onze dataset: in Braeckes onderzoek telt 94,4% (of 2676/ 2834) van de VzC's 2-8 lettergrepen tegenover 63,3% (of 1092/ 1725) in eigen onderzoek en 63,3% (of 1794/ 2834) telt slechts 2-4 lettergrepen versus 27,9% in eigen onderzoek (481/ 1725). ...
... The verb influences the choice of realization in dative alternation (Bresnan et al. 2007, Gries 2005, Gries & Stefanowitsch 2004, Wasow & Arnold 2003 and can also influence phenomena like heavy NP shift (Stallings, MacDonald, & O'Seaghdha 1998, Staub, Clifton, & Frazier 2006 and passivization (Manning 2003). Dative alternation is influenced by the verb as certain verbs have a bias towards the choice of the realized dative (Bresnan et al. 2007, Wasow & Arnold 2003. ...
... Thus if rhythmic feet in words are counted as part of dependency length calculations (in a revised definition), this factor is directly related to dependency length minimization. Further, in the case of heavy NP shift, both comprehension (Staub et al. 2006, van Schijndel et al. 2014) and production (Stallings et al. 1998) studies have shown that the properties of individual verbs (e.g., transitivity) can influence the shifting of NPs. This has a direct effect on dependency length calculations, and thus this factor does interact with the minimal dependency length preference. ...
... Typically, both behavioural experiments (Arnold et al. 2000, Stallings et al. 1998, Staub et al. 2006) and corpus studies (Bresnan et al. 2007, Szmrecsanyi 2004, Wasow 2002 related to syntactic choice focus on a single or a very limited set of constructions. In contrast, our study is conceived as an investigation involving multiple construction types (other cross-construction studies include Reitter, Keller, & Moore 2011, Reitter & Moore 2014). ...
We investigate the extent to which syntactic choice in written English is influenced by processing considerations as predicted by Gibson's (2000) Dependency Locality Theory (DLT) and Surprisal Theory (Hale, 2001; Levy, 2008). A long line of previous work attests that languages display a tendency for shorter dependencies, and in a previous corpus study, Temperley (2007) provided evidence that this tendency exerts a strong influence on constituent ordering choices. However, Temperley's study included no frequency-based controls, and subsequent work on sentence comprehension with broad-coverage eye-tracking corpora found weak or negative effects of DLT-based measures when frequency effects were statistically controlled for (Demberg & Keller, 2008; van Schijndel, Nguyen, & Schuler 2013; van Schijndel & Schuler, 2013), calling into question the actual impact of dependency locality on syntactic choice phenomena. Going beyond Temperley's work, we show that DLT integration costs are indeed a significant predictor of syntactic choice in written English even in the presence of competing frequency-based and cognitively motivated control factors, including n-gram probability and PCFG surprisal as well as embedding depth (Wu, Bachrach, Cardenas, & Schuler, 2010; Yngve, 1960). Our study also shows that the predictions of dependency length and surprisal are only moderately correlated, a finding which mirrors Dember & Keller's (2008) results for sentence comprehension. Further, we demonstrate that the efficacy of dependency length in predicting the corpus choice increases with increasing head-dependent distances. At the same time, we find that the tendency towards dependency locality is not always observed, and with pre-verbal adjuncts in particular, non-locality cases are found more often than not. In contrast, surprisal is effective in these cases, and the embedding depth measures further increase prediction accuracy. We discuss the implications of our findings for theories of language comprehension and production, and conclude with a discussion of questions our work raises for future research.
... For example, word order has been shown to play an important role in processing ease (c.f. Stallings, MacDonald, O'Seaghdha, 1998;Hawkins, 1994Hawkins, , 2004, such that sentence (1) is much easier to process than sentence (2): ...
... Listeners' sensitivity to the complexity of the direct object NP demonstrates that the grammaticalization of structures as defined by the complexity hierarchy of Keenan and Comrie (1977) affects processing. These results are also consistent with findings showing that processing is facilitated by shifting long or complex structures to the end of a sentence (Stallings et al, 1998). ...
... Consistent with Hawkins' predictions, results of several production experiments have shown that constituent length affects word order preference. For example,Stallings, MacDonald, and O'Seaghdha (1998) showed that speakers generally preferred word orders in which shorter constituents preceded longer ones, thus minimizing the syntactic domain. They demonstrated that the length of a direct object noun phrase (NP) affects word order choice, with participants strongly preferring 'Mary explained [to Sam] PP [the recently published theory of genetic inheritance] NP ' over 'Mary explained [the recently published theory of genetic inheritance] NP [to Sam] PP '. ...
Recent theories have proposed that processing difficulty affects both individuals’ choice of grammatical structures and the distribution of these structures across languages of the world (Hawkins, 2004). Researchers have proposed that performance constraints, such as efficiency, integration, and storage costs, drive languages to choose word orders that minimize processing demands for individual speakers (Hawkins, 1994; Gibson, 2000). This study investigates how three performance factors, adjacency, dependency, and complexity, affect reading times for sentences with verb-particle constructions. Results from a self-paced reading task indicate that reading times increase with each performance factor, such that shifted sentences, more dependent verb-particle constructions, and more complex noun phrases are more difficult. More importantly, I explore the relative weightings and interactions of the performance factors. The results support the notion that processing ease affects grammaticalization, such that those structures which are more easily processed by individuals (subject relatives and adjacent dependent constituents) are also more common crosslinguistically (Keenan & Hawkins, 1987).
... First, it should be noted that there are a few diff erent ways in which weight can be measured, including syllables, words, and syntactic nodes. Stallings, MacDonald, and O'Seaghdha ( 1998 ) argue, based on previous production studies, that length in syllables is not relevant for constituent ordering during sentence production, and instead favor word-based measures. Wasow ( 1997 ) considers whether word-based measures are equivalent to measures of structural embedding. ...
... ph r a s e l e n g t h a n d d i s c o u r s e fa c to r s i n c o m b i nat i o n Current research on constituent order alternations shows that various factors in addition to grammatical weight infl uence speakers' choice of ordering. Such factors turn out to be somewhat diff erent across languages and constructions, but some examples include information status (Arnold et al., 2000 ;Bresnan et al., 2007 ), animacy (Bresnan et al., 2007 ;Rosenbach, 2005 ), iconicity of sequence (Diessel, 2008 ), verb bias (Stallings et al., 1998 ), and lexical semantic dependency (Lohse et al., 2004 ;Wiechman & Lohmann, 2013 ). Most relevant here are previous results for RCE, to which we now turn. ...
... Like Arnold et al. ( 2000 ), Stallings et al. ( 1998 ) show that speakers' choice of structure in elicited production depends on some of the same factors shown to aff ect corpus frequencies. Their study, on which the methods for our Experiment 2 are based, examined the eff ects of grammatical weight and verb bias on speakers' choice of Heavy NP Shift ( Todd delivered to Al the package ) vs. canonical order ( Todd delivered the package to Al ). ...
In one type of Relative Clause Extraposition (RCE) in English, a subject-modifying relative clause occurs in a displaced position following the matrix VP, as in: Some options were considered that allow for more flexibility. Although RCE incurs a discontinuous dependency and is relatively infrequent in discourse, previous corpus and acceptability judgment studies have shown that speakers prefer RCE over adjacent ordering when the RC is long in relation to the VP, the subject NP is indefinite, and the main verb is passive/presentative (Francis, 2010; Francis & Michaelis, 2014; Walker, 2013). The current study is the first to relate these conditional preferences to online measures of production. For a spoken production task that required speakers to construct sentences based on visual cues, results showed that the same factors that modulate choice of structure—VP length, RC length, and definiteness of the subject NP—also modulate voice initiation time. That is, when the sentential context warrants a particular structure, that structure becomes easier to produce. Following the approach of MacDonald (2013), we explain these findings in terms of two production biases, one of which favors early placement of shorter, more accessible phrases and the other of which promotes rapid retrieval from memory of the most frequently-used sub-types of a construction.
... Such assumptions about the incremental nature of language production have wide empirical support with evidence ranging from natural speech errors, to elicited productions in pairedassociation and picture naming tasks, etc. (Roelofs, 1998;Brown-Schmidt and Konopka, 2014;Roelofs, 2013Roelofs, , 1997Ferreira and Swets, 2002;Smedt, 1994;Schriefers et al., 1990;Levelt et al., 1991Levelt et al., , 1999, although see Stallings et al. (1998) for some complications. ...
... The effect of phrase length on constituent ordering, so-called "heavy-NP shift", is wellstudied (Kimball, 1973;Stallings et al., 1998;Arnold et al., 2004, etc.). Under IG, this results because objects consisting of more words simply have more pieces to combine, and thus should take more time to build within the production system. ...
This dissertation investigates the wide-ranging implications of a simple fact: language unfolds over time. Whether as cognitive symbols in our minds, or as their physical realization in the world, if linguistic computations are not made over transient and shifting information as it occurs, they cannot be made at all. This dissertation explores the interaction between the computations, mechanisms, and representations of language acquisition and language processing—with a central theme being the unique study of the temporal restrictions inherent to information processing that I term the immediacy of linguistic computation. This program motivates the study of intermediate representations recruited during online processing and acquisition rather than simply an Input/Output mapping. While ultimately extracted from linguistic input, such intermediate representations may differ significantly from the underlying distributional signal. I demonstrate that, due to the immediacy of linguistic computation, such intermediate representations are necessary, discoverable, and offer an explanatory connection between competence (linguistic representation) and performance (psycholinguistic behavior). The dissertation is comprised of four case studies. First, I present experimental evidence from a perceptual learning paradigm that the intermediate representation of speech consists of probabilistic activation over discrete linguistic categories but includes no direct information about the original acoustic-phonetic signal. Second, I present a computational model of word learning grounded in category formation. Instead of retaining experiential statistics over words and all their potential meanings, my model constructs hypotheses for word meanings as they occur. Uses of the same word are evaluated (and revised) with respect to the learner's intermediate representation rather than to their complete distribution of experience. In the third case study, I probe predictions about the time-course, content, and structure of these intermediate representations of meaning via a new eye-tracking paradigm. Finally, the fourth case study uses large-scale corpus data to explore syntactic choices during language production. I demonstrate how a mechanistic account of production can give rise to highly "efficient" outcomes even without explicit optimization. Taken together these case studies represent a rich analysis of the immediacy of linguistic computation and its system-wide impact on the mental representations and cognitive algorithms of language.
... To our knowledge, one study that directly addresses the question of whether L2 learners can fully attain native-like production strategies is Dennison (2008), focusing on phrasal ordering preference. In verb-medial languages such as English, short phrases tend to be uttered before long or heavy phrases (Wasow, 1997;Clark and Wasow, 1998;Stallings et al., 1998;Arnold et al., 2000). For instance, compared to a bare noun ("home"), a complex noun phrase (NP) modified by an RC is most likely to be shifted to the end, as in I invited home a friend that I missed very much, rather than the other way around (I invited a friend that I missed very much home). ...
... Furthermore, deferring constituents that are long and complex would buy speakers more time for structural formulation. Since DCLs are morpho-syntactically less weighty than RCs in terms of phrasal length and structural complexity, uttering the DCL first conforms to the accessibility principle or the short-before-long preference widely attested in SVO languages like English (Wasow, 1997;Clark and Wasow, 1998;Stallings et al., 1998;Arnold et al., 2000). In short, uttering DCLs first can relieve Chinese speakers of the pressure to plan the complex SR. ...
English as a verb-medial language has a short-before-long preference, whereas Korean and Japanese as verb-final languages show a long-before-short preference. In second language (L2) research, little is known regarding how L1 processing strategies affect the ultimate attainment of target structures. Existing work has shown that native speakers of Chinese strongly prefer to utter demonstrative-classifier (DCL) phrases first in subject-extracted relatives (DCL-SR-N) and DCLs second in object-extracted relatives (OR-DCL-N). But it remains unknown whether L2 learners with typologically different language backgrounds are able to acquire native-like strategies, and how they deviate from native speakers or even among themselves. Using a phrase-assembly task, we investigated advanced L2-Chinese learners whose L1s were English, Korean, and Japanese, because English lacks individual classifiers and has postnominal relative clause (RC), whereas Korean and Japanese have individual classifiers and prenominal RCs. Results showed that the English and Korean groups deviated from the native controls’ asymmetric pattern, but the Japanese group approximated native-like performance. Furthermore, compared to the English group, the Korean and Japanese groups favored the DCL-second configuration in SRs and ORs. No differences were found between the Korean and Japanese groups. Overall, our findings suggest that L1 processing strategies play an overarching role in L2 acquisition of asymmetric positioning of DCLs in Chinese RCs.
... Previous studies have shown that speakers' syntactic choice is influenced by a subcategorization preference that is inherent to the verb (Ferreira, 1994; Lexical cueing in phrase encoding 6 Potter, 1992;Stallings, MacDonald, & O'Seaghdha, 1998). For example, Lombardi and Potter (1992) found that the speakers' wordings in sentence recall were susceptible to the intrusion of a lure word. ...
... Specific to the syntactic flexibility in sentence production, speakers prepare several syntactic alternatives in parallel. The final selection among those options relies upon multiple constraints (e.g., Ferreira, 1994;Myachykov, Garrod, & Scheepers, 2012;Prat-Sala & Branigan, 2000;Stallings et al., 1998). In many cases, these constraints might not work in concert. ...
This paper examined the role of lexical processing in phrase structure building in sentence production. We asked whether speakers exploit a lexical cue as a lexical guide (i.e. the cued word occurs earlier in the sentence) and as a retrieval cue (i.e. a cue facilitates the retrieval of a memorised structure). In two experiments, participants recalled Dutch genitive sentences. In some recall trials, they received a lexical cue that repeated one argument of the to-be-recalled sentence. In two further experiments, participants read a genitive sentence and then generated a new one from a visually-presented triplet of arguments. The visual salience of the arguments and lexical overlap were manipulated. In all four experiments, speakers consistently started the phrase with the cued word. There was no evidence of a lexical cueing effect on structure retrieval. The findings suggest that speakers mainly exploit lexical information as a lexical guide when formulating phrase structures.
... Probably the most well-studied NP weight effect is the short-before-long tendency in English word order variation (Arnold et al. 2000;Bresnan et al. 2007;Stallings and MacDonald 2011;Stallings et al. 1998). The example in (1) shows two possible word orders of an English sentence with a verb phrase (VP) that contains two prepositional phrases (PP), one with a short NP (i.e., PP1) and the other a long NP that contains a relative clause (i.e., PP2). ...
... While the MiD principle seems to be more successful in explaining cross-language differences in weight effects, questions remain as to how comprehension-oriented factors can condition word order preferences in online sentence production. Both short-before-long and long-before-short patterns have been observed in online sentence production tasks, in English (Stallings et al. 1998) and Japanese (Yamashita and Chang 2001) respectively. The fact that these production tasks did not involve any role of listener-either real or imaginary-and offered no incentive for listener accommodation undermines a comprehension-oriented account of the results. ...
Background
In the literature of sentence production, both short-before-long and long-before-short word order tendencies have been observed across languages. Specifically, SVO languages such as English show the short-before-long noun phrase (NP) shift, placing heavy NPs near the end of the sentence; on the other hand, verb-final languages such as Japanese and Korean show the long-before-short NP shift, placing heavy NPs earlier in the sentence. In this paper, we examine the effects of NP weight on word order variation in Mandarin Chinese, which not only has a predominantly SVO word order but also allows a grammaticalized SOV construction (i.e., the ba construction).
Methods
We conducted a corpus analysis with two verb-specific datasets extracted from the 10 million-word Academia Sinica Balanced Corpus of Modern Chinese (Version 5.0; Chen et. al., Proceeding of the 11th Pacific Asia Conference on Language, Information and Computation, 1996). Each dataset contained more than 900 sentences, in either SVO or ba construction and possible to be converted to the other word order without changing sentence meaning. Generalized mixed-effects models were built to examine the effect of NP weight on the surface word order (SVO vs. ba), while controlling for other factors that are also known to influence the SVO-ba alternation (e.g., verb complement, animacy, and givenness of the object NP, sentence structure, and structural parallelism in the context). The accuracy of the modeling results was inspected by comparing the word order predictions made by the models with both actual word orders observed in the corpus and naturalness ratings of alternative word orders by native speakers in behavioral experiments.
Results
Our results show a U-shaped NP weight effect on SVO-ba alternation, in that both very short and very long NPs are more likely to be shifted to preverbal positions than NPs with medium weight. These results provide evidence that both conceptual and positional factors are operating in the preverbal domain in Mandarin.
Conclusion
Taken together with previous findings of positional factors operating in the postverbal domain in Mandarin, our results suggest that the relative sensitivity to conceptual and positional factors can vary within a language. We discuss findings in the framework of the sentence production model.
... In this work, we investigate whether RNNs trained on a language modeling objective learn certain syntactic preferences exhibited by humans, especially those involving word order. We draw on a rich literature from quantitative linguistics that has investigated these preferences in corpora and experiments (e.g., McDonald et al., 1993;Stallings et al., 1998;Rosenbach, 2008). ...
... We tested whether RNNs show length-based preferences similar to Example (1). 2 We adapted 40 items from Stallings et al. (1998) which consist of a verb followed by an NP and a temporal PP adjunct, where the order of the NP and the PP and the length of the NP are manipulated. If the networks show human-like word ordering preferences, there should be a penalty for PP-NP order when the NP is short, but this penalty should be smaller or nonexistent when the NP is long. ...
RNN language models have achieved state-of-the-art results on various tasks, but what exactly they are representing about syntax is as yet unclear. Here we investigate whether RNN language models learn humanlike word order preferences in syntactic alternations. We collect language model surprisal scores for controlled sentence stimuli exhibiting major syntactic alternations in English: heavy NP shift, particle shift, the dative alternation, and the genitive alternation. We show that RNN language models reproduce human preferences in these alternations based on NP length, animacy, and definiteness. We collect human acceptability ratings for our stimuli, in the first acceptability judgment experiment directly manipulating the predictors of syntactic alternations. We show that the RNNs' performance is similar to the human acceptability ratings and is not matched by an n-gram baseline model. Our results show that RNNs learn the abstract features of weight, animacy, and definiteness which underlie soft constraints on syntactic alternations.
... Le paradigme expérimental que nous utilisons est inspiré des paradigmes de production contrainte, utilisée dans les travaux psycholinguistiques, notamment la tâche du rappel de phrase, utilisée par Stallings et al. (1998) pourétudier l'ordre relatif entre des compléments verbaux en anglais, ou par Yamashita et Chang (2001) pourétudier l'ordre des constituants dans les phrases transitives et ditransitives en japonais. Dans ces expériences, les participants doivent construire une phraseà partir des syntagmes affichés sur l'écran, qu'ils aurontà formuler (à l'oral) après un laps de temps. ...
... Cette observation est cohérente avec celle faite par Stallings et al. (1998) pour l'anglais. Selon ces auteurs, le domaine postverbal est fortement contraint par la présence du verbe qui a une influence importante sur ce qui suit. ...
Dans le domaine postverbal, l’ordre des compléments du verbe est relativement libre (Paul donnera un livre à chacun / Paul donnera à chacun un livre). Nous cherchons à déterminer quels sont les facteurs qui influencent le choix d’un ordre par rapport à l’autre. Nous nous intéressons en particulier à deux facteurs : le poids grammatical et l’accessibilité discursive. En ce qui concerne le poids, nous soulevons la question de sa définition : cette notion doit-elle être définie en termes de complexité syntaxique ou de longueur ? L’accessibilit é discursive, définie comme le caractère donné ou nouveau des référents, a été mentionnée comme un facteur pertinent dans certains travaux sur le français (Blinkenberg 1928, Berrendonner 1987), mais aucune étude empirique n’a permis de montrer son rôle dans le domaine postverbal. A partir d’une étude expérimentale de complétion de phrases, nous montrons que ces deux facteurs expliquent en partie l’ordre choisi par les locuteurs. Les résultats obtenus permettent d’affirmer que le poids grammatical doit être défini en termes de longueur des constituants et de complexité syntaxique (dans la lignée de Wasow et Arnold 2005, sur l’anglais), et de montrer que l’effet de complexité est plus important que celui de la longueur. De plus, le statut donné ou nouveau des référents a un effet significatif sur le choix des locuteurs, mais cet effet est de petite taille en comparaison à celui du poids grammatical.
... In flexible languages (i.e. with flexible word order and null pronouns) and the postverbal domain (cf. Stallings et al. 1998 ...
... Inspired by (on screen) sentence recall task paradigms (e.g. Stallings et al. 1998, Yamashita & Chang 2001. ...
In this paper, we present a usage-based study to constituent ordering in Persian in line with studies on word order variations that take into account functional factors. The results of our empirical study, combining corpus-based and experimental data, goes against the widespread theoretical view of Persian’s phrase structure. More precisely, our results undermine the backbone argument put forward to support this view which consist of a broadly admitted claim regarding the relative order between objects in a ditransitive sentence. Persian is an SOV language with DOM triggered by definiteness and/or specificity and marked by the enclitic =rā, (1). This enclitic is also used as a topicalizer, placed in the initial position, beyond DO, (2). It should be noted that Persian do not formally mark nouns for definiteness, a single noun without any formal determination, ketāb, can receive two reading (except in the DO position): 1) bare noun, that is, a noun lacking any determination or quantification, nonspecific and under-specified for number, ‘a book, some books’; 2) definite noun ‘the book’. (Existential) indefiniteness is, however, formally marked by the determiner ye(k), ye ketāb, the enclitic =i, ketāb=i, or the combination of both, ye(k) ketāb=i, ‘a book’. Furthermore, disposing only of around 250 simplex verbs, verbal concepts are mainly expressed in Persian by complex predicates (cf. Samvelian 2012), that is the combination of a simplex verb and a non-verbal element, prototypically involving bare nouns (dars xāndan ‘to study’), (2). (1) a. Maryam ketāb xarid Marya book bought-3SG ‘Maryam bought a book/some books.’ b. Maryam ketāb=rā xarid Maryam book=DOM bought-3SG ‘Maryam bought the book.’ (2) tābestān=rā dars mi-xān-am summer=DOM lesson IPFV-read.PRS-3SG ‘Summer, I will study.’ Most theoretical studies, namely in the framework of the generative grammar, postulate a phrase structure for Persian reflecting a structural asymmetry between rā-marked and non-rā-marked DOs (e.g. Brown & E. Karimi 1994, Ghomeshi 1997, Karimi 2003, 2005). Despite their substantial difference, these studies posit two different positions, either base-generated or resulting from a movement, for rā-marked vs. non-rā-marked DOs, roughly represented in (3). Furthermore, they share the same methodological approach, based mainly, if not exclusively, on the authors (categorical) grammaticality judgments. (3) a. [VP DP[+rā] [V’ PP V]] b. [VP [V’ PP [V’ DP[-rā] V]]] Karimi (2005) provides the most thorough argumentation in favor of this analysis. Considering rā-marking a matter of specificity, the author assimilates non-rā-marked DOs, to bare nouns, minimizing their significant differences, and puts forward a number of observations suggesting syntactic and semantic asymmetries between rā-marked and bare DOs to support this analysis. Namely, the semantic incorporation of non-rā-DOs into the verb, similar to the combination involved in CPs, and their lack of (semantic and syntactic) autonomy contrary to rā-marked DOs, and the unmarked word order between the DO and the IO, which is, broadly assumed to follow the schema in (4). (4) a. DO[+rā] IO V b. IO DO[-rā] V Adopting a truly empirical approach, we have studied the relative order between the DO and IO in Persian in line with studies on word order preferences (e.g. Hawkins 1994, Arnold et al 2000, Yamashita & Chang 2001, Wasow 2002, Bresnan et al 2007). Including a preliminary multifactorial corpus study and follow up (off-line) psycholinguistic experiments conducted via web-based questionnaires, our study takes into account functional factors shown to play significant role in word order preferences cross-linguistically such as length, givenness, humanness, verbal lemma, via mixed-effect regression modeling. Most importantly, our results show that the oversimplifying dual classification of DOs based on markedness is flawed: in comparable context, while rā-marked DOs show a strong preference for the DO-IO order and bare DOs for the inverse, the picture is less clear-cut for in-between DOs. In one hand, indefinite/quantified (non-rā-marked DOs) - contrary to what is expected by the theory - show a clear preference, although moderate, for the DO-IO order, grouping with rā-¬marked DOs. In the other, bare DOs carrying adjuncts, show a much less strong preference for the IO-DO order. More generally, extra-syntactic factors such as relative length and humanness show up to play a significant role. Indeed, ordering preferences between the DO and the IO, rather than being dichotomous, are best reflected by a continuum based on the degree of determination of the DO - closely related to its discourse accessibility (cf. Gundel et al. 1993) - as well as other (functional) factors related to the conceptual accessibility, such as relative length (cf. Yamashita & Chang 2001) and humanness. Furthermore, examining all other arguments put forward, we conclude that there is no empirical ground for positing two syntactic positions for the DO. Accordingly, the dual phrase structure analysis for Persian that provides wrong predictions with respect to the unmarked word order is refuted. This study thus highlights the importance of using empirically solid methods in theoretical syntax.
... Cognitive accessibility refers to "the speed and accuracy with which concepts are retrieved from memory so that they are available to consciousness and linguistic processing" (Tachihara & Goldberg, 2020, p. 233). More accessible words tend to occur before less accessible words (Stallings et al., 1998). As a longer antonym contains more syllables or characters than a shorter one, it naturally requires greater effort in planning and preparation in the production process. ...
This study presents a contrastive analysis of antonym order in English and Chinese coordinate structures using a multifactorial corpus method. The analysis yields the following findings. First, antonym ordering in coordinate structures is driven largely by the same ordering constraints across the two languages. Chronology and Positivity are the most prominent semantic motivating constraints, and Morphology is the most important formal motivating constraint. Second, there are some differences in constraints on antonym ordering between English and Chinese. Age and Gender exert a significant effect on Chinese antonym ordering, but show no effect on English antonym ordering. Hierarchical Superiority significantly affects Chinese antonym ordering, but has only a marginal effect on English antonym ordering. Furthermore, this study explores the motivations underlying these similarities and differences and argues that there is a clear correlation between iconicity and antonym order. The exploration reveals that their similarities in semantic constraints may be attributed to the iconicity of closeness and the iconicity of temporal sequence, while their differences may be ascribed to the iconicity of cultural values and norms. Their similarities in formal constraints stem from cognitive accessibility. These findings provide further evidence that antonym order is determined by general cognitive principles.
... Information about the date of writing is of course useful for studying and plotting the diachronic development. The weight of the object, measured in words, is interesting because heavy objects are more likely to undergo rightward extraposition, an independent type of a syntactic transformation which affects our VO/OV diagnostics (Wasow 1997;Stallings, MacDonald, and O'Seaghdha 1998;Thráinsson 2007;Stallings and MacDonald 2011). ...
This article describes PaCQL (Parsed Corpus Query Language), a novel query language for carrying out research on parsed historical corpora, an important task for the digital humanities. PaCQL implements and enhances many of the most important features of earlier software that is designed for computational research in historical syntax and combines such functionality with a search engine which employs a fast in-memory index that cuts down waiting time in many realistic research scenarios. A web interface is provided with an automatically created summary of the main quantitative findings. The primary goal of this project is to contribute to the development of software tools which are designed from the ground up specifically with the needs of the digital humanities in mind.
... Since we are interested in how specific verbs affect ordering preference extremity, we estimated verb bias (Stallings, MacDonald, & O'Seaghdha, 1998;Wasow & Arnold, 2003) for each verb as the probability of a sentence being realized as the prepositional object structure based on just the random effect of the verb (eliminating the contributions from the fixed effects). Specifically, let V be the random effect intercept of a particular verb derived from the regression model, the probability of this verb being realized in the V-NP-PP structure is then calculated as follows. ...
We examine how idiosyncrasies of specific verbs in syntactic constructions affect constituent ordering preferences. Previous work on binomial expressions in English has demonstrated that the polarization of ordering preferences for a given binomial type depends on its overall frequency. The higher the frequency of a binomial type, the stronger and more extreme preference/regularization language users will have for one alternative over the other (e.g. "facts and techniques" > "techniques and facts"; "bread and butter" >>> "butter and bread"). Here using the dative constructions in English as the test case, we show that the same frequency-dependent regular-ization exists in syntactic structures above the word level. The more frequent a dative construction type is, governed by the head verb, the stronger preference there is for one alternation over the other. Further, we present evidence that the regulariza-tion patterns can be accounted for via iterated learning model-ing of language change, suggesting that frequency-dependent regularization emerges via the interactions between language production, language learning and cultural transmission.
... DLM predicts that all words in dependencies will be under a pressure to be close to each other, but it does not make any predictions about which dependencies will be under especially strong pressure. However, empirically, DLM effects in word order preferences and also in online processing difficulty show asymmetries based on the details about particular dependencies (Stallings et al., 1998;Demberg and Keller, 2008). ...
... This boost in activation constitutes the priming effect: when speakers choose a structure to produce, they are assumed to essentially sample from the available structures proportional to their relative activation, so that a more highly activated structure is more likely to be selected again (see also Dubey, Keller & Sturt, 2008). This assumption, too, is broadly shared among models of production (see, e.g., Dell, 1986;Dell & Chang, 2014;Ferreira, 1996;Segaert, Wheeldon & Hagoort, 2016;Stallings, MacDonald & O'Seaghdha, 1998). Activation is assumed to exhibit power law decay over time (an assumption shared with many memory models). ...
Native language (L1) processing draws on implicit expectations. An open question is whether non-native learners of a second language (L2) similarly draw on expectations, and whether these expectations are based on learners’ L1 or L2 knowledge. We approach this question by studying inverse preference effects on lexical encoding. L1 and L2 speakers of Spanish described motion events, while they were either primed to express path, manner, or neither. In line with other work, we find that L1 speakers adapted more strongly after primes that are unexpected in their L1. For L2 speakers, adaptation depended on their L2 proficiency: The least proficient speakers exhibited the inverse preference effect on adaptation based on what was unexpected in their L1; but the more proficient speakers were, the more they exhibited inverse preference effects based on what was unexpected in the L2. We discuss implications for L1 transfer and L2 acquisition.
... The size of the tree on either side of the root reflects whether a sentence is "top" or "tail" heavy, or more balanced. This feature is based on the principle that sentences are easier to process, and thus are judged to be natural and well worded, if the dependencies of the head are roughly evenly distributed on either side (Temperley, 2008), and that heavy noun phrases are hard to process at the beginning of the sentence (Stallings et al., 1998). Table 2: Results of baselines, top two feature-based classifiers and models using subset of features. ...
... Both listeners and speakers are sensitive to probability distributions (Jaeger cites, e.g., Hale 2001, MacDonald 1994Bell et al. 2003;Stallings et al. 1998, i.a., for production). Both speaker and listeners also rapidly adapt to changes in such distributions with relatively little exposure (Clayards et al. 2008, as cited by Jaeger). ...
Language scholars as far back as Behaghel (1909) have noted the tendency to produce short before long. What underlies this principle of end-weight? Several accounts that develop an incremental view of production—speakers produce what they can as soon as they can—may be seen as necessarily motivated by a constrained capacity view of human cognitive faculties. Such resources—for example, working memory capacity—typically vary across the population, while the preference for end-weight has previously been modeled as fixed across the population. In the present work, it is argued that if the principle of end-weight is motivated by limits on such an underlying resource that varies by individual, we should see significant variation by individual in strength of preference for end-weight. Corpus and experimental studies subsequently confirm significant individual variation in strength of such preferences. Next, we see the principle of end-weight at play in several English syntactic-alternation constructions, including dative shift, verb-particle placement, prepositional phrase ordering, and heavy-NP shift, among others. Looking across these constructions, if individual preference for end-weight is motivated by a limited, underlying cognitive resource, such a resource should nevertheless remain constant within each individual across constructions. Having thus found individual variation in the principle of end-weight, a given speaker's strength of preference may be expected to remain relatively steady across constructions—that is, consistency in variation. Testing this proposition presents a methodological challenge. Existing syntactically parsed corpora of natural speech are generally built shallow—many speakers but insufficient contributions per speaker to model individual end-weight preference. To address this, a deep parsed corpus is developed—yielding up to 2.5M words per speaker—as well as novel statistical approaches for exploring correlation of per-speaker preferences across models of separate constructions. Multiple studies with this corpus and complementary experiments indicate a significant measure of consistency in each individual's relative strength of end-weight preference across constructions, as predicted. These results do not prove constrained capacity; other compatible accounts include the possibility that we develop individual production preferences through exposure to prior distributions. If so, though, the results above introduce the provision that in developing our individual production preferences, each speaker synthesizes a single rule—or at least a substantially consistent preference—across different syntactic-alternation constructions. This may be seen as a syntactic neighborhood effect. Finally, though, a pair of prior studies suggest that Australians may show greater strength of preference for end-weight in the dative-shift alternation than Americans, an effect of group that is compatible with exposure to differing prior distributions but not at all predicted by constrained capacity. If preference is variable by individual but consistent across constructions—as the core of the present work indicates—might we expect the same at the varietal level? An additional experiment conducted here with American, British, and Australian groups and across four different constructions finds that the Australian preference for end-weight is significantly stronger across the set of constructions. While constrained capacity may be at play in variation by individual, the cross-varietal results support the role of learning by exposure to prior distribution, with syntactic neighborhood effects appearing cross-construction for both individuals and groups.
... An important point pertaining to the information structure of pre-versus postmodified NPs also merits discussion here: Under some theories of language processing, ease of sentence comprehension is largely determined by preferences during language production, with constructions preferred and generated more frequently during production being easier to comprehend (Hopman & MacDonald, 2018;MacDonald, 2013MacDonald, , 2016; also see Levy, 2008). Because production proceeds in an incremental manner (Bock, 1982;Bock & Warren, 1985;De Smedt & Kempen, 1987;Kempen & Hoenkamp, 1987;Levelt, 1989), with easier bits being produced earlier than harder bits (Ferreira, 1996;Ferreira & Swets, 2002;Stallings, MacDonald, & O'Seaghdha, 1998;Yamashita & Chang, 2001; see also Lau, & Hwang, 2016), it could be the case that postmodified NPs are easier to produce than premodifiers, and therefore more frequently encountered by comprehenders. This might again be due the fact the postmodified NPs conform to the "given-before-new" principle. ...
We examined whether the position of modifiers in English influences how words are encoded and subsequently retrieved from memory. Compared with premodifiers, postmodifiers might confer more perceptual significance to the associated head nouns, are more consistent with the "given-before-new" information structure, and might also be easier to integrate because the head noun is available before the modifications are encountered. In 4 experiments, we investigated whether premodified (the cruel and merciless king), and postmodified (the king who was cruel and merciless) noun phrases (henceforth, NPs) could induce variations in ease of subsequent retrieval. In Experiments 1, 2, and 3, participants used more pronouns (he), as opposed to full descriptions (the king) to refer to postmodified NPs than to unmodified competitors, but pronominal reference to premodified NPs and unmodified competitors did not differ, suggesting that postmodified NPs are more accessible in memory. When the data from all 3 experiments were combined, we also observed significantly more pronominal reference to post- than to premodified NPs, as well as a greater increase in pronominal reference rates between postmodified NPs and unmodified competitors than between premodified NPs and unmodified competitors. In Experiment 4, words following critical pronouns were read faster when the pronouns referred to modified than to unmodified NPs, and also when the pronouns referred to post- rather than premodified NPs. Taken together, our results show enhanced retrieval facilitation for postmodified NPs compared with premodified NPs. These results are the first to demonstrate that the linear position of modifications results in measurable processing cost at a subsequent point. The results have important implications for memorybased theories of language processing, and also for theories assigning a central role for discourse status and information structure during sentence processing.
... ɸɳ > @ ʆ ɸɮ @ɷ > >ʈ ʆ ɸɮ ʈ ɷ ʀ 37 ɷ ʇ>ɳ> ʆ ɸ ʀ ʆ > ɷ ɳ @ɷ > >ʈ ʆ ɸɮ ɷ> ʆ ɸ ʈ @ɸ ɳɷ @ɸ ʄ ɳ> ʄ @ @ ʆ ɸɮ ɷ> < ʀ ɳɷ @ɳɷɳ> » ɳɷ « (Mazurkewich, 1984;Pinker, 1989 . > ʈ < ɳ ʇ ʇ > ɳ @>ɷɳ ɷ> ʄ ʄ ɳ ʆɮ ɸɳ > = ʇ @ɳ » ʆɸʈ ɳ ʀ ɮ 2 « ɳ > = ʆ ɸ ɷ> ɳ ɸ < ɳ ʈ ɮ < ɷʺ @ @ ɸɳ ɷɳ> @ ɷɳ ) Arnold & et al., 2000;Ross, 1967;Stallings & et al., 1998;Wasaw, 1997 .( ɳ ɷ> ɳ ʄ ɳ ɳ @ 3 ) Garrett, 1980;Bock & Levelt, 1994 Bock & Levelt, 1994;Garrett, 1980 ...
... 대표적으로 한국어 사용자들은 문장의 산출이나 이해 과정에서 길이가 긴 문장성분이 짧은 문 장성분에 선행하는 어순('long before short', 이하 'LbS')을 반대의 어순보다 선호하는 것으로 나타났 는데 (남윤주 & 홍우평, 2013, Choi, 2007, Nam, Hong, & Yun, 2016, 이러한 어순 처리의 선호도는 일본어에서도 확인된 바 있으며 따라서 한국어나 일본어 같은 핵어후치 언어에 국한된 현상으로 간주 된다 (Yamashita & Chang, 2001;Hawkins, 1994Hawkins, , 2004. 영어과 같은 핵어선치 언어에서는 길이가 짧은 문장성분이 긴 문장성분 앞에 나타나는 어순('short before long', 이하 'SbL')이 선호되는 것으 로 밝혀졌기 때문이다 (Arnold et al. 2000;Bock & Levelt, 1994, Ferreria, 1996Stallings et al. 1998). (mental grammar)과 별개로 존재하는 언어처리장치(language processor)가 택하는 통사처리 전략의 존재를 시사한다. ...
... A characterization of such biases is illustrated in the tendency for items that are more accessible to appear both earlier in utterances and at more salient syntactic structures than items that are more difficult to retrieve. This is exemplified in the preference for individuals to speak of previously mentioned information before introducing new information in the discourse (Bock and Irwin 1980;Ferreira and Yoshita 2003;Tanaka et al. 2011) or in the tendency to minimize dependency length (Stallings et al. 1998;Yamashita and Chang 2001). ...
Experience-based approaches to language hold that individuals become sensitive to distributed emergent phenomena in their linguistic experience. The purpose of this paper is to bring together experience-based perspectives from the domains of cognitive psychology and linguistics. First, we present an overview of the cognitive processes that underpin experience-based learning, and review the cognitive biases that have been attributed to the emergence of distributional regularities in language. We then discuss the P-chain (Dell, G. S. & F. Chang. 2014. The P-chain: Relating sentence production and its disorders to comprehension and acquisition.
... See also [3,12]). Verbs that appear more frequently in structures where the verb and its complement are non-adjacent are more likely to allow the shifting of complements to different sentence positions [13]. Verb-specific syntactic probabilities can also affect pronunciation-words in an utterance tend to be shortened when verbs appear in highly likely syntactic structures and lengthened when they appear in unlikely structures [14]. ...
How do speakers choose between structural options for expressing a given meaning? Overall preference for some structures over others as well as prior statistical association between specific verbs and sentence structures (“verb bias”) are known to broadly influence language use. However, the effects of prior statistical experience on the planning and execution of utterances and the mechanisms that facilitate structural choice for verbs with different biases have not been fully explored. In this study, we manipulated verb bias for English double-object (DO) and prepositional-object (PO) dative structures: some verbs appeared solely in the DO structure (DO-only), others solely in PO (PO-only) and yet others equally in both (Equi). Structural choices during subsequent free-choice sentence production revealed the expected dispreference for DO overall but critically also a reliable linear trend in DO production that was consistent with verb bias (DO-only > Equi > PO-only). Going beyond the general verb bias effect, three results suggested that Equi verbs, which were associated equally with the two structures, engendered verb-specific competition and required additional resources for choosing the dispreferred DO structure. First, DO production with Equi verbs but not the other verbs correlated with participants’ inhibition ability. Second, utterance duration prior to the choice of a DO structure showed a quadratic trend (DO-only < Equi > PO-only) with the longest durations for Equi verbs. Third, eye movements consistent with reimagining the event also showed a quadratic trend (DO-only < Equi > PO-only) prior to choosing DO, suggesting that participants used such recall particularly for Equi verbs. Together, these analyses of structural choices, utterance durations, eye movements and individual differences in executive functions shed light on the effects of verb bias and verb-specific competition on sentence production and the role of different executive functions in choosing between sentence structures. © 2017 Thothathiri et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
... The argument construction may specify the conditions that the lexical meaning of the verb needs to satisfy in order to be compatible with it (van Trijp 2015: 621-624). To investigate these conditions, verb disposition may be studied, i. e., the preferences of verbs to appear with a particular argument construction (Stallings et al. 1998). In this way, it can be shown that the ditransitive construction is more compatible with verbs whose lexical meaning expresses a "caused reception", like to give, while the prepositional dative construction is more compatible with verbs expressing a "caused motion", like to bring (Goldberg 1992;Gries and Stefanowitsch 2004;Colleman 2009). ...
This article presents the results of a corpus study of the Dutch
psychological verbs ergeren ‘to annoy’, interesseren ‘to interest’, storen ‘to disturb’,
and verbazen ‘to amaze’. These verbs exhibit a syntactic alternation between their
seemingly synonymous transitive and reflexive argument constructions, as in
Elizabeth ergert John vs. John ergert zich aan Elizabeth (both: ‘Elizabeth annoys
John’). On the basis of current theoretical insights, four hypotheses are formulated
predicting the language user’s preferred argument construction. It is argued that the
popular agentivity hypothesis, as proposed in studies by, for instance, Dowty,
Langacker, and Zaenen, should be broken up into the token- and type-level agentivity
hypotheses. Both agentivity hypotheses come with different theoretical entailments,
and make distinct predictions about the quantitative data. These data
confirm the token-level agentivity hypothesis, while not doing the same for the
type-level agentivity hypothesis. Additionally, it is found that stimuli and experiencers
that are heavier in terms of informational weight both prompt the use of the
reflexive construction, and that the individual preferences of the verbs could not be
predicted based on their historical semantic development.
... Hoffmann [76] claimed that such a short-before-long linearization is cognitively motivated by the pressure to reduce the recognition domain, or, the distance between the verb and the head of the last constituent governed by the verb. Nevertheless, some availability-based accounts are held as able to better explain the weight effects than the distance-based accounts [77,78]. Interestingly, the corpus analyses [22] show that, Japanese, which is a SOV language, presents a recurring pattern of long-before-short arrangement: long preverbal phrases tend to occur ahead of short ones. ...
... Adopting a quantitative approach to word order variations (e.g. Wasow, 1997;Stallings et al., 1998;Yamashita & Chang, 2001;Wasow, 2002;Bresnan et al., 2007), these studies investigate the relative order between the two objects in ditransitive constructions in Persian and examine the effect of functional factors such as the relative length and animacy on ordering preferences. More precisely, explore the predictions of the DOM criterion on the relative order between the DO and the IO (cf. ...
The aim of this paper is to tease apart two available views of the VP in Persian. The prevailing view of the Persian VP initially suggested in generative
studies assumes a hierarchical structure with two object positions, mainly motivated by the existence of differential object marking in Persian.
Building on quantitative studies, we revisit this hierarchical view and show that it is not born out by the data. A flat structure view of the VP, on the
contrary, is in line with the data.
... 6 As shown by Sigurjónsdóttir and Nowenstein (2016), discourse status can influence the choice between a Canonical Passive and a truth-conditionally equivalent New Impersonal Passive, which is therefore another case where structural optionality in passive-like constructions is to some extent arbitrated by discourse context. between an active/passive), there is a tendency to place long phrases at the end of clauses (e.g., Wasow 1997, Stallings et al. 1998 The only relevant difference between (18a)/(18b) and (19a)/(19b) is the phonological weight of the agent. 7 Therefore, it appears that phonological weight is an independent type of pressure to prefer the variant with a 'by'-phrase, in addition to the agent expressing new information. ...
We investigate 'by'-phrases in the Icelandic impersonal passive and argue that they are grammatical, contra previous reports in the literature. However, it is only acceptable to use them when there are agent-specific pressures to realize the agent on the linear right, i.e., when the agent expresses new information or when it is phonologically heavy. We develop a formal analysis in the spirit of the Voice-adjunction theory of 'by'-phrases and consider facts from historical syntax.
... Dabei werden Dinge, die salient sind und im Fokus der Aufmerksamkeit stehen, eher zu Beginn im Satz erwähnt (Forrest 1996). Auch kürzere und einfachere Phrasen werden, wenn möglich, früh im Satz platziert (Stallings et al. 1998 Befunde wie die, die im vorangehenden Absatz berichtet wurden, legen die Hypothese nahe, dass Informationen vom Sprecher in der Regel so verpackt werden, dass sie möglichst einfach vom Gegenüber verstanden werden (▶ Abschn. 14.5.1). ...
In diesem Kapitel beschäftigen wir uns mit den Prozessen, die der menschlichen Sprachverarbeitung jenseits der Wortebene zugrunde liegen. Zunächst werden die unterschiedlichen Aspekte der menschlichen Sprachverarbeitung vornehmlich aus der Perspektive der Sprachverstehensforschung behandelt. Hier geht es um die syntaktische, semantische und pragmatische Analyse von Sätzen sowie die Forschung zum Textverstehen. Die Sprachproduktion wird in einem gesonderten Abschnitt behandelt. An Ende des Kapitels werden die neurobiologischen Grundlagen der menschlichen Sprachverarbeitung diskutiert, einige Anwendungsbeispiele aufgeführt und ein Ausblick auf zukünftige Forschung im Bereich der Sprachpsychologie gegeben.
Schlüsselwörter: Ambiguität; Bedeutung; Bedeutungskomposition; Kohärenz; Parsing; Pragmatik; Satzverstehen; Semantik; Sprachproduktion; Sprachwissen; Syntax; Textlinguistik, Textverstehen; Versprecher
... An important idea that has emerged in the sentence production literature is that it is a by-product of speakers' preference for starting their sentences with the information that is most accessible to them (Bock, 1982;Bock & Warren, 1985;Tanaka, Branigan, McLean, & Pickering, 2011), a strategy MacDonald has referred to as "Easy First" (MacDonald, 2013). Experimental and corpus investigations of language production have shown that animate, frequent, imageable, and shorter expressions tend to occur earlier in sentences than expressions that are inanimate, less frequent, abstract, and long (eg, Bock, 1982;Bock & Warren, 1985;Ferreira, 1994;Stallings, MacDonald, & O'Seaghdha, 1998;Tanaka et al., 2011). The preference to order given before new (eg, Bock & Irwin, 1980) is thus simply another example of Easy First. ...
The good-enough language processing approach emphasizes people's tendency to generate superficial and even inaccurate interpretations of sentences. At the same time, a number of researchers have argued that prediction plays a key role in comprehension, allowing people to anticipate features of the input and even specific upcoming words based on sentential constraint. In this chapter, we review evidence from our lab supporting both approaches, even though at least superficially these two perspectives seem incompatible. We then argue that what allows us to link good-enough processing and prediction is the concept of information structure, which states that sentences are organized to convey both given or presupposed information, and new or focused information. Our fundamental proposal is that given or presupposed information is processed in a good-enough manner, while new or focused information is the target of the comprehender's prediction efforts. The result is a theory that brings together three different literatures that have been treated almost entirely independently, and which can be evaluated using a combination of behavioral, computational, and neural methods.
Cet article présente une étude de l’alternance actif/passif en français parlé adoptant une approche quantitative, multifactorielle et probabiliste. Elle s’appuie sur trois corpus oraux du projet Orféo (Debaisieux & Benzitoun, 2020). Un échantillon de 400 énoncés (200 actifs, 200 passifs) a été extrait, annoté, puis analysé par régression logistique à effets mixtes. Un modèle prédictif de l’alternance entre actif et passif est alors proposé. Celui-ci permet de mettre en évidence l’effet de contraintes multiples sur le choix de construction : contraintes de longueur croissante des constituants, de codage harmonique des arguments, effets de sémantique verbale et d’amorçage. Les facteurs en jeu en français parlé se retrouvent également en français écrit journalistique (Da Cunha & Abeillé, 2020) et en anglais, suggérant une certaine unité de la construction à travers les langues et les genres textuels. L’étude pointe enfin certaines différences entre français parlé et écrit, montrant que les effets de définitude, pronominalité et longueur sont plus forts à l’oral.
This study applies a corpus-based quantitative approach to the word order typology and linguistic theories about word order in several genetically unrelated language varieties in northwestern Iran, such as Mukri Kurdish, Northeastern Kurdish and Armenian (Indo-European), Jewish Neo-Aramaic (Semitic), and Azeri Turkic (Turkish). Despite the difference in the default position of the direct object, the existing corpora of published and personal field data of narrative free speech demonstrate that these languages share the clause-final position of Targets predominantly (e.g., physical and metaphorical goals, recipients , addressees, and resultant-states) in their word order. Yet, Targets are more flexible in Mukri Kurdish, Northeastern Neo-Aramaic, and Azeri Turkic, whereas they are less flexible in Armenian and Northeastern Kurdish. Among various factors relevant to the placement of Targets, morphosyntactic features such as parts of speech exhibit constraints and clear preferences in the pre-and postverbal placement of Targets.
We investigate the relative impact of two influential theories of language comprehension, viz., Dependency Locality Theory (Gibson, 2000; DLT) and Surprisal Theory (Hale, 2001; Levy, 2008), on preverbal constituent ordering in Hindi, a predominantly SOV language with flexible word order. Prior work in Hindi has shown that word order scrambling is influenced by information structure constraints in discourse. However, the impact of cognitively grounded factors on Hindi constituent ordering is relatively underexplored. We test the hypothesis that dependency length minimization is a significant predictor of syntactic choice, once information status and surprisal measures (estimated from n-gram i.e., trigram and incremental dependency parsing models) have been added to a machine learning model. Towards this end, we setup a framework to generate meaning-equivalent grammatical variants of Hindi sentences by linearizing preverbal constituents of projective dependency trees in the Hindi-Urdu Treebank (HUTB) corpus of written text. Our results indicate that dependency length displays a weak effect in predicting reference sentences (amidst variants) over and above the aforementioned predictors. Overall, trigram surprisal outperforms dependency length and parser surprisal by a huge margin and our analyses indicate that maximizing lexical predictability is the primary driving force behind preverbal constituent ordering choices in Hindi. The success of trigram surprisal notwithstanding, dependency length minimization predicts non-canonical reference sentences having fronted direct objects over variants containing the canonical word order, cases where surprisal estimates fail due to their bias towards frequent structures and word sequences. Locality effects persist over the Given-New preference of subject-object ordering in Hindi. Accessibility and local statistical biases discussed in the sentence processing literature are plausible explanations for the success of trigram surprisal. Further, we conjecture that the presence of case markers is a strong factor potentially overriding the pressure for dependency length minimization in Hindi. Finally, we discuss the implications of our findings for the information locality hypothesis and theories of language production.
Acceptability rating questionnaires are a highly accessible tool for the experimental study of different syntactic phenomena, however, for the study of word order preferences, they are not as efficient as sentence production experiments. In this paper, we present a constrained sentence production task implemented via web-based self-administrated questionnaires. In order to highlight the advantages of such a paradigm, we compare the results of two experiments, an acceptability rating and a sentence completion task, using the same experimental material in order to study constituent ordering preferences in French and Persian. Our results show that acceptability rating data reflect the variation observed in production data only in a reduced manner and, consequently, make it difficult to study factors with small effect size or to identify a canonical word order among different grammatically possible alternatives.
Resumen: Comprender cómo se procesa el lenguaje ha sido interés central de la psicolingüística hace varias décadas ya que entender los procesos involucrados aporta al estudio del lenguaje, pero, también, al estudio y comprensión de la mente. En este artículo se presenta una revisión de los modelos de procesamiento sintáctico y sus aportes a la comprensión acerca del funcionamiento del lenguaje y de la cognición. En primer lugar, se retoman algunas de las discusiones clásicas acerca del procesamiento del lenguaje y los procesos cognitivos en general. En segunda instancia, se realiza una revisión de distintos modelos de procesamiento sintáctico propuestos en las últimas décadas, sus características y supuestos acerca de la facultad del lenguaje. Por último, se comparan las diferentes propuestas en función de los debates clásicos presentados y su posición acerca de la universalidad de ciertos procesos cognitivos.
Palabras-clave: procesos cognitivos; comprensión del lenguaje; procesamiento sintáctico; adjunción.
Abstract: Understanding how language is processed has been a theme of an utmost interest for psycholinguistics for several decades, since fully acknowledging the processes involved contributes not only to the study of language, but also to both the study and understanding of the mind. This article presents a review of syntactic parsing models and their contributions to the comprehension of the functioning of language and cognition. First of all, some of the classical discussions about language processing and cognitive processes in general are analyzed. Secondly, different models of syntactic parsing proposed in the last decades as well as their characteristics and assumptions about the language faculty are reviewed. Finally, the different proposals are compared in terms of the classical debates presented and their position on the universality of certain cognitive processes.
Keywords: cognitive processes; language comprehension; syntactic parsing; attachment.
Heaviness (or phrasal length) has been shown to trigger mirror-image constituent ordering preferences in head-initial and head-final languages (heavy-late vs. heavy-first). These preferences are commonly attributed to a general cognitive pressure for processing efficiency obtained by minimizing the overall head-dependents linear distance-measured as the distance between the verb and the head of its left/right-most complement (Hawkins's Minimizing Domains) or as the sum of the distances between the verb and its complements (Dependency Length Minimization). The alternative language-specific accessibility-based production account, that considers longer constituents to be conceptually more accessible and views heavy-first as a salient-first preference, is dismissed because it implies differential sentence production in SOV and SVO languages. This paper studies the effect of phrasal length in Persian, a flexible SOV language displaying mixed head direction and differential object marking. We investigated the effect of linear distance as well as the effect of conceptual enrichment in two sentence production experiments. Our results provide clear evidence that support DLM while undermining Hawkins's MiD. However, they also show that some length effects cannot be captured by a dependency-distance-minimizing model and the conceptual accessibility hypothesis also needs to be taken into account to explain ordering preferences in Persian. Importantly, our findings indicate that distance minimization has a less strong effect in Persian than previously shown for other SOV languages.
In experiments, previous studies have observed that the choice of word order in Japanese is influenced not only by heaviness but also by animacy. To be more concrete, the heavy constituent tends to precede the light one and the animate referent tends to come before the inanimate one. Yet, the present corpus analysis demonstrates that word order changes are not motivated by animacy. This discrepancy can be accounted for by supposing that animacy has an impact on word order only when the effects of other factors are neutralized. It is possible that the effects of animacy are so weak that they work only in psycholinguistic experiments because other factors are controlled. On the other hand, the effects of animacy are not observed in this corpus study probably because other factors are not controlled in actual examples. Thus, I propose that researchers should utilize both naturalistic and experimental evidence in order to draw reasonable conclusions about the grammatical aspects of language.
Introduction: According to performance theory and efficiency principles, constituent ordering is influenced by processing efficiency, i.e. the ordering of constituents is intended to facilitate processing. Adopting an online reading time paradigm, the research was aimed at studying the role of grammatical weight in likelihood of post-verbal movement and processing loads of pairs of corresponding sentences. Method: The aforesaid paradigm, as an efficient experimental technique to assess processing speed of syntactic structures, was designed by OpenSesame Software. Two sets of 24 sentences in three levels of weight (light, medium and heavy) and two structural levels (unmarked preverbal constituent and post-verbal one) were presented to 40 randomly selected Persian-speaking participants (between the ages 18 and 40) with university education. The first set consisted of sentences containing relative clause in two configurations and the second set of pairs of scrambled (post-verbal) and unscrambled sentences. Sentences were followed by a yes/no question. Data was analyzed using SPSS and ANOVA test. Results: In light level, the mean reading time of sentences with canonical constituent ordering was less than those with post-verbal constituents. However, this trend got reversed by the increase of grammatical weight of the constituent subjected to scrutiny. In other word, the post-verbal movement of constituents resulted in reduction of mean reading time in medium-weight sentences. This decline was markedly more significant in heavy sentences. Conclusion: The ordering of constituents and the likelihood of movement is highly weight-sensitive. As a conclusion, the end-weight principle as an incentive to optimize the processing efficiency was approved in Persian.
Keywords:
Constituent ordering, grammatical weight, language processing, efficiency principles, grammatical movement
The so-called Principle of Inherency in Dutch linguistics offers an explanation for the
obligatory position of a certain number of constituents immediately before the verbal
remainder. The goal of this contribution is to provide an answer to the question
whether this principle has validity in French as well. Through a – necessarily partial -
contrastive description of the complement of direction in contemporary Dutch and
French, we show that this principle is inoperable in French. We hypothesize that the
observed divergence is related to the supra-segmental level of the two languages.
Dutch is characterized by a non-contrastive and mobile sentence accent and thus has a
‘flexible’ pragmatic structure, whereas French, which lacks a sentence accent, has a
word-group accent which triggers a ‘strict’ pragmatic structure. Consequently, the
Principle of Inherence, based on a fixed position, has full latitude to be realized in
Dutch, unlike French.
Three experiments investigated whether speakers use constituent ordering as a mechanism for avoiding ambiguities. In utterances like “Jane showed the letter to Mary to her mother,” alternate orders would avoid the temporary PP-attachment ambiguity (“Jane showed her mother the letter to Mary,” or “Jane showed to her mother the letter to Mary”). A preference judgment experiment confirmed that comprehenders prefer the latter orders for dative utterances when the former order would have contained an ambiguity. Nevertheless, speakers in two on-line production experiments showed no evidence of an ambiguity avoidance strategy. In fact, they were slightly more likely to use the former order when it was ambiguous than when it was not. Speakers’ failure to disambiguate with ordering cannot be explained by the use of other ambiguity mechanisms, like prosody. A prosodic analysis of the responses in Experiment 3 showed that while speakers generally produced prosodic patterns that were consistent with the syntactic structure, these patterns would not strongly disambiguate the PP-attachment ambiguity. We suggest that speakers do not consistently disambiguate local PP-attachment ambiguities of this type, and in particular do not use constituent ordering for this purpose. Instead, constituent ordering is driven by factors like syntactic weight and lexical bias, which may be internal to the production system.
Arne Lohmann , English coordinate constructions: A processing perspective on constituent order. Cambridge: Cambridge University Press, 2014. Pp. 214. ISBN 978-1-107-04088-5. - Volume 20 Issue 2 - Matthias Eitelmann
Variations in postverbal constituent ordering have been attributed to both grammatical complexity (heaviness) and discourse status (newness), although few studies compare the two factors explicitly. Through corpus analysis and experimentation, we demonstrate that both factors simultaneously and independently influence word order in two English constructions. While past investigations of these factors have focused on their effects in language comprehension, we argue that postponing heavy and new constituents facilitates processes of planning and production.
in interactive activation models of language processing, inhibition is a molecular, quasi-neural process by which representational units send inhibitory signals to one another, or regulate their own activation by self-inhibition / in this chapter, we focus on the use of the concept of inhibition in the interactive activation literature, but we will also examine the implications of our analysis for the interpretation of experimental data / we survey the wide range of functions served by inhibitory mechanisms in language processing and examine the question of whether inhibition is an identifiable process in language production and comprehension
This chapter discusses the temporal patterns of rapid movement sequences in speech and typewriting and what these patterns might mean in relation to the advance planning or motor programming of such sequences. The chapter discusses response factors that affect the time to initiate a prespecified rapid movement sequence after a signal when the goal is to complete the sequence as quickly as possible as well as how such factors affect the rate at which movements in the sequence are produced. The response factor of central interest is number of elements in the sequence. The effect of the length of a movement sequence on its latency is based partly on the possibility that it reflects a latency component used for advance planning of the entire sequence: The length effect would then measure the extra time required to prepare extra elements. The idea that changes in reaction time might reflect changes in sequence preparation in this way proposed that simple reaction time increased with the number of elements in a sequence of movements made with one arm. A part of the reaction time includes the time to gain access to stored information concerning the whole sequence: a process akin to loading a program into a motor buffer, with sequences containing more elements requiring larger programs, and larger programs requiring more loading time.
A dynamical systems approach to parsing is proposed in which syntactic hypotheses are associated with attractors in a metric space. These attractors have many of the properties of traditional syntactic categories, while at the same time encoding context-dependent, lexically specific distinctions. Hypotheses motivated by the dynam ical system theory were tested in four reading time experiments examining the interaction of simple lexical frequencies, frequencies that are contingent on an environment defined by syntactic categories, and frequencies contingent on verb argument structure. The experiments documented a variety of contingent frequency effects that cut across traditional linguistic grains, each of which was predicted by the dynamical systems model. These effects were simulated in an implementation of the theory, em ploying a recurrent network trained from a corpus to construct metric representations and an algorithm implementing a gravitational dynamical system to model reading time as time to gravitate to an attractor.
IPF (Incremental Parallel Formulator) is a computer model in which the formulation stage in sentence generation is distributed among a number of parallel processes. Each conceptual fragment which is passed on to the Formulator gives rise to a new process, which attempts to formulate only that fragment and then exits. The task of each formulation process consists basically of instantiating one or more syntactic segments and attaching these to the present syntactic structure by means of a general unification operation. A shared memory provides the necessary (and only) interaction between parallel processes and allows the integration of segments created by different processes into one syntactic structure. The race between parallel processes in time may partly explain some variations in word order and lexical choice. The system is provided with a Dutch grammar and is implemented in the object-oriented language CommonORBIT on a Symbolics machine with simulated parallelism.
Ambiguity resolution is a central problem in language comprehension. Lexical and syntactic ambiguities are standardly assumed to involve different types of knowledge representations and be resolved by different mechanisms. An alternative account is provided in which both types of ambiguity derive from aspects of lexical representation and are resolved by the same processing mechanisms. Reinterpreting syntactic ambiguity resolution as a form of lexical ambiguity resolution obviates the need for special parsing principles to account for syntactic interpretation preferences, reconciles a number of apparently conflicting results concerning the roles of lexical and contextual information in sentence processing, explains differences among ambiguities in terms of ease of resolution, and provides a more unified account of language comprehension than was previously available.
An interactive 2-step theory of lexical retrieval was applied to the picture-naming error patterns of aphasic and nonaphasic speakers. The theory uses spreading activation in a lexical network to accomplish the mapping between the conceptual representation of an object and the phonological form of the word naming the object. A model developed from the theory was parameterized to fit normal error patterns. It was then "lesioned" by globally altering its connection weight, decay rates, or both to provide fits to the error patterns of 21 fluent aphasic patients. These fits were then used to derive predictions about the influence of syntactic categories on patient errors, the effect of phonology on semantic errors, error patterns after recovery, and patient performance on a single-word repetition task. The predictions were confirmed. It is argued that simple quantitative alterations to a normal processing model can explain much of the variety among patient patterns in naming. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Long, complex phrases tend to come at the ends of clauses; this is called “endweight.” A variety of characterizations of weight have been proposed in the literature, but none has been sufficient to cover the full range of attested cases of end-weight. Corpus data on heavy NP shift, the dative alternation, and particle movement indicate that there are several structural measures of weight that are highly correlated with constituent ordering. Proposed explanations for endweight have been based on parsing considerations, largely ignoring the speaker; but what facilitates parsing does not always help in production. Examination of phenomena where these interests do not coincide indicates that the demands of sentence planning provide a better explanation for end-weight than parsing. Finally, accounts of end-weight cannot be purely structure-based, but must take lexical factors into consideration.
Natural languages contain probabilistic constraints that influence the resolution of ambiguities. Current models of sentence processing agree that probabilistic constraints affect syntactic ambiguity resolution, but there has been little investigation of the constraints themselves-what they are, how they differ in their effects on processing, and how they interact with one another. Three different types of probabilistic constraints were investigated: “pre-ambiguity” plausibility information, information about verb argument structure frequencies, and “post-ambiguity” constraints that arrive after the introduction of the ambiguity but prior to its disambiguation. Reading times for syntactically ambiguous sentences were compared to reading times for unambiguous controls in three self-paced reading experiments. All three kinds of constraints were found to be helpful, and when several constraints converged, ambiguity resolution was facilitated compared to when constraints conflicted. The importance of these constraint interactions for ambiguity resolution models is discussed.
A computer simulation model of the human speaker is presented which generates sentences in a piecemeal way. The module responsible for Grammatical Encoding (the tactical component) is discussed in detail. Generation is conceptually and lexically guided and may proceed from the bottom of the syntactic structure upwards as well as from the top downwards. The construction of syntactic structures is based on unification of so-called syntactic segments.
Immediate effects of verb-specific syntactic (subcategorization) information were found in a cross-model naming experiment, a self-paced reading experiment, and an experiment in which eye movements were monitored. In the reading studies, syntactic misanalysis effects in sentence complements (e.g., "The student forgot the solution was...") occurred at the verb in the complement (e.g., was) for matrix verbs typically used with noun phrase complements but not for verbs typically used with sentence complements. In addition, a complementizer effect for sentence-complement-biased verbs was not due to syntactic misanalysis but was correlated with how strongly a particular verb prefers to be followed by the complementizer that. The results support models that make immediate use of lexically specific constraints, especially constraint-based models, but are problematic for lexical filtering models.
In this paper I want to describe a universal theory of “grammatical relations” that arises naturally within the formal theory of syntax and semantics of natural language developed originally by Richard Montague (1970, 1973) and subsequently extended in a certain series of treatments of Montague’s ideas that begins with Thomason 1976 (first written in 1972) and includes Thomason 1974, Dowty 1979a (written in 1975), Dowty 1975, Dowty 1978, Bach 1979 (written in 1977), Bach 1980, and Schmerling 1979. I believe this analysis of grammatical relations is supported by some of the same observations as are the related theories of Relational Grammar and Arc Pair Grammar developed by Perlmutter, Postal, and others (Perlmutter and Postal 1977, Johnson and Postal forthcoming, Perlmutter and Postal to appear, Perlmutter this volume, Postal this volume).
In a variety of situations in psychological research, it is desirable to be able to make statistical comparisons between correlation coefficients measured on the same individuals. For example, an experimenter may wish to assess whether two predictors correlate equally with a criterion variable. In another situation, the experimenter may wish to test the hypothesis that an entire matrix of correlations has remained stable over time. The present article reviews the literature on such tests, points out some statistics that should be avoided, and presents a variety of techniques that can be used safely with medium to large samples. Several illustrative numerical examples are provided.
The question addressed in this study is how lexical information stored with verbs is used during the earliest stages of syntactic analysis. The properties of a novel type of verb we term ''reciprocal verbs" were exam ined in three experiments. Reciprocal verbs include hug and kiss - verbs that alternate between a transitive (e.g.Bill hugged Mary ) and an intransitive form (Bill and Mary hugged), but whose intransitive form requires a conjoined subject. In three experiments, readers were presented with garden-path sentences containing reciprocal verbs, and their eye movements were m onitored. The experiments demonstrate that information from the reciprocal verb influenced the earliest stages of syntactic parsing. Implications for models of parsing are discussed, focusing in particular on the argument structure of reciprocal verbs as well as the role of argument-structure frequency in parsing.
The crosslinguistic study of sentence processing - Volume 20 Issue 2 - Jean Aitchinson
Four experiments demonstrate effects of prosodic structure on speech production latencies. Experiments 1 to 3 exploit a modified version of the Sternberg et al. (1978, 1980) prepared speech production paradigm to look for evidence of the generation of prosodic structure during the final stages of sentence production. Experiment 1 provides evidence that prepared sentence production latency is a function of the number of phonological words that a sentence comprises when syntactic structure, number of lexical items, and number of syllables are held constant. Experiment 2 demonstrated that production latencies in Experiment 1 were indeed determined by prosodic structure rather than the number of content words that a sentence comprised. The phonological word effect was replicated in Experiment 3 using utterances with a different intonation pattern and phrasal structure. Finally, in Experiment 4, an on-line version of the sentence production task provides evidence for the phonological word as the preferred unit of articulation during the on-line production of continuous speech. Our findings are consistent with the hypothesis that the phonological word is a unit of processing during the phonological encoding of connected speech.
This introduction reviews some history of how lexical representations have come to play an important role in sentence processing research. It discusses a number of issues related to lexical representations and sentence processing, including the importance of timecourse information in theorising, the importance of frequency information in theories of sentence processing and the question of the grain of frequency inform ation.
Ferreira and Clifton (1986, Experiment 1) found that readers experienced equal difficulty with temporarily ambiguous reduced relatives clauses when the first noun was animate (e.g., "The defendant examined by the lawyer was . . .") and when it was inanimate and thus an unlikely Agent (e.g., "The evidence examined . . ."). This data pattern suggested that a verb′s semantic constraints do not affect initial syntactic ambiguity resolution. We repeated the experiment using: (1) inanimate noun/verb combinations that did not easily permit a main clause continuation, (2) a baseline condition with morphologically unambiguous verbs (e.g., "stolen"), (3) a homogeneous set of disambiguating prepositional phrases, and (4) a display in which all of the critical regions were presented on the same line of text. In two eye-movement experiments, animacy had immediate effects on ambiguity resolution: only animate nouns showed clear signs of difficulty. Post-hoc regression analyses revealed that what little processing difficulty readers had with the inanimate nouns varied with the semantic fit of individual noun/verb combinations: items with strong semantic fit showed no processing difficulty compared to unambiguous controls, whereas items with weak semantic fit showed a pattern of processing difficulty which was similar to Ferreira and Clifton (1986). The results are interpreted within the framework of an evidential (constraint-based) approach to ambiguity resolution. Analyses of reading times also suggested that the millisecond per character correction for region length is problematic, especially for small scoring regions. An alternative transformation is suggested.
constraints, abstract underlying forms, phonological rules, and syntactic and semantic features. Furthermore, such errors reveal that linguistic performance is highly rule-governed, and that in many cases it is grammatical rules which con- strain or monitor,actual speech,production. While a model,of linguistic compe- tence is independent of temporal constraints, a model of linguistic performance must,provide information,as to the sequencing,of events in real time. To explain the occurrence of particular kinds of errors, a specific ordering of rules is posited, which ordering may,or may,not coincide with the organization,of a grammar. 1. In current linguistic and psychological literature a sizable number,of arti-
Thirteen papers in this book illustrate MacWhinney and Bates's Competition Model (CM), with a focus on cross-linguistic processing. Studies in this volume show that (1) the CM is useful in predicting certain gross cross-linguistic differences of comprehension, particularly in relation to actor assignment and (2) children's processing strategies change over the years. (JP)
focus on evidence from transient troubles in access that rise normal conversation and from there long-lasting troubles found in aphasic patients
two-stage models of lexical access
closed-class versus open-class words
retrieval of full-form versus retrieval of stem
interactive activation models
studies of picture description
neuropsychological evidence for two lexical levels
priming in the semantic lexicon
evidence for the temporal disjunction of levels / temporal properties of speech (PsycINFO Database Record (c) 2012 APA, all rights reserved)
In psychological research, it is desirable to be able to make statistical comparisons between correlation coefficients measured on the same individuals. For example, an experimenter (E) may wish to assess whether 2 predictors correlate equally with a criterion variable. In another situation, the E may wish to test the hypothesis that an entire matrix of correlations has remained stable over time. The present article reviews the literature on such tests, points out some statistics that should be avoided, and presents a variety of techniques that can be used safely with medium to large samples. Several numerical examples are provided. (18 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
This article compares the predictions of two models of grammatical encoding in language production. The basis of one model is that alternative syntactic structures compete to determine which structure is eventually used. The second model is incremental: Utterances are gradually built up, and the structure emerges from the construction process itself. If grammatical encoding is competitive, syntactic choices should pose difficulties; if incremental, syntactic choices should ease the creation of speech. These predictions were tested in three experiments where speakers created utterances which sometimes required a syntactic decision. When constructing a sentence allowed a syntactic choice, speakers generally constructed that utterance with fewer errors and more quickly. This finding supports the notion that language production operates incrementally. (PsycINFO Database Record (c) 2012 APA, all rights reserved)
Theories of sentence production based on speech errors divide lexical-syntactic integration processes into 2 components. The first involves formulating an abstract structural representation that includes semantically specified lexical items. The second involves placing phonologically specified content words into a syntactic frame whose configuration is determined by the initial structural representation. Syntactic form thus may be influenced directly by variations in the semantic processing of words, but not by variations in phonological processing. This hypothesis was tested and supported in 2 experiments with a total of 160 undergraduates who produced extemporaneous picture descriptions. Production of each description was preceded by the presentation of a priming word that was semantically or phonologically related to a target word likely to occur in the description. Semantically primed targets tended to appear as the Ss of active and passive sentences, whereas the same targets when they were not primed were more likely to appear as the objects. Phonological priming, although equal to semantic priming in ability to elicit the target words, was not reliably related to syntactic form. (34 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)
discuss sentence comprehension, focusing on the relation between language structure and language processing at this level of analysis / presents a history of sentence comprehension over the past 35 yrs that reviews now classic experiments and theories that drove these research efforts / summarize recent experimental efforts and the evolving theories and metatheories that have driven these experimental investigations; this discussion includes a consideration of the important evolving notions of the lexical/syntactic interface (PsycINFO Database Record (c) 2012 APA, all rights reserved)
In three experiments, I examined initiation times for memorized utterances. Sentences varied in phonological word length, syntactic complexity, or semantic plausibility. The experiments demonstrated that number of phonological words and syntactic complexity (as measured by the number of nodes in a phrase structure tree) affected the time it took subjects to initiate the utterance. If a sentence had a syntactically complex subject and a syntactically complex object, speakers tended to pause at the subject-verb phrase boundary, and pause duration increased with upcoming complexity, just like initiation times. Semantic plausibility had no measurable effect. I argue that these results reflect the process of translating a semantic/syntactic representation of a sentence into a sound-based structure ultimately useable by the speech apparatus. For long and complex sentences, these resource-intensive processes cannot occur over the domain of the entire sentence, and so the production system must divide the sentence into two performance units, with the boundary between them occuring at a syntactically prominent location.
A connectionist account of lexical ambiguity resolution is presented in which the nonlinear dynamics that arise during processing is emphasized. In the implementation, the spelling, pronunciation, part of speech, and meaning of both ambiguous and unambiguous words are represented as a distributed pattern of activity over a set of simple processing units in a fully recurrent network. After the network is trained on this lexicon using an error-correction algorithm, the performance of the network is assessed by presenting just part of a lexical entry (e.g., the spelling). The number of processing cycles to activate all the units representing the spelling, the pronunciation, or the meaning to their minimal or maximal activation are used as indices of lexical decision times, naming times, or reading times. respectively. Consistent with empirical results, the simulations demonstrate the effect of frequency and context on the processing of unambiguous words, as well as the effect of relative dominance and context on the time course of the activation of meanings of ambiguous words. Advantages of a distributed representation are also discussed.
This chapter presents an analysis of sentence production, by distinguishing between the general problem of language production, which must include message formulation, and the specific problem of sentence production, which may be viewed as translation process. The chapter assumes that somewhere in the recesses of the central nervous system, an interaction takes place among the current motor and perceptual experiences, stored information, motivational systems, and various other variables. This interaction gives rise to a communicative intention, also termed as message. This message has to be translated into a set of instructions sufficient to guide the articulatory apparatus. The final point that should be made from the analysis of sentence production is that even if one to has good evidence about the vocabulary of the computational system that mediates the translation from messages into their realization as instructions to the articulatory system, one would still be unsatisfied. The reason of this dissatisfaction is that one would want to be able to characterize the information flow in the system in terms of interactions between the various structural types represented in the computational vocabulary.
It is widely acknowledged that characteristics of the general information-processing system in which sentence formulation occurs may provide constraints on syntax in language use. The author proposes one possible source of such constraints. Evidence is reviewed indicating that the syntax of sentences may to some degree reflect the transient processing demands of lexical retrieval, suggesting an interaction between syntactic and lexical processing. Specifically, the syntactic structure of utterances appears to be sensitive to the accessibility of lexical information, with phrases containing more accessible information occurring earlier in sentences. The existence of such an interaction argues that the utterance formulation system is not strictly hierarchical, as most current approaches to sentence production imply. A broad framework for models of production is outlined that incorporates these interactions within a limited-capacity processing system. This framework also permits a resolution of contradictions in the literature on pragmatic determinants of constitutional order in adult language use. (6 p ref)
Eyetracking and the self-paced moving-window reading paradigm were used in two experiments examining the contributions of both frequency-based verb biases and the plausibility of particular word combinations to the comprehension of temporarily ambiguous sentences. The temporary ambiguity concerned whether a noun following a verb was its direct object (The senator regretted the decision immediately.), or instead the subject of an embedded clause (The senator regretted the decision had been made public.). The experiments crossed the plausibility of the temporarily ambiguous noun as a direct object (e.g.,The senator regretted the decision . . .vsThe senator regretted the reporter . . .) with verb bias, eliminating a confound present in earlier research and allowing an examination of interactions between the two factors. Unbiased verbs were included as well to evaluate the role of plausibility in the absence of verb bias. The results generally replicated Trueswell, Tanenhaus, and Kello's (1993) finding that verb bias has rapid effects on ambiguity resolution, and showed in addition that verb bias and plausibility interact during comprehension. The results are most consistent with parallel interactive models of language comprehension such as constraint satisfaction models.