On the learnability of frequent and infrequent word orders: An artificial language learning study

Most languages have a basic or "canonical" word order, which determines the relative positions of the subject (S), the verb (V), and the object (O) in a typical declarative sentence. The frequency of occurrence of the six possible word orders among world languages is not distributed uniformly. While SVO and SOV represent around 85% of world languages, orders like VSO (9%) or OSV (0.5%) are much less frequent or extremely rare. One possible explanation for this asymmetry is that biological and cognitive constraints for structured sequence processing make some word orders easier to be processed than others. Therefore, the high frequency of these word orders would be related to their higher learnability. The aim of the present study was to compare the learnability of different word orders between groups of adult subjects. Four artificial languages with different word orders were trained: two frequent (SVO, SOV) and two infrequent (VSO, OSV). In a test stage, subjects were asked to discriminate between new correct sentences and syntax or semantic violations. Higher performance rates and faster responses were observed for more frequent word orders. The results support the hypothesis that more frequent word orders are more easily learned.     

When Cars Hit Trucks and Girls Hug Boys: The Effect of Animacy on Word Order in Gestural Language Creation

A well‐known typological observation is the dominance of subject‐initial word orders, SOV and SVO, across the world's languages. Recent findings from gestural language creation paradigms offer possible explanations for the prevalence of SOV. When asked to gesture transitive events with an animate agent and inanimate patient, gesturers tend to produce SOV order, regardless of their native language biases. Interestingly, when the patient is animate, gesturers shift away from SOV to use of other orders, like SVO and OSV. Two competing hypotheses have been proposed for this switch: the noisy channel account (Gibson et al., 2013) and the role conflict account (Hall, Mayberry, & Ferreira, 2013). We set out to distinguish between these two hypotheses, disentangling event reversibility and patient animacy, by looking at gestural sequences for events with two inanimate participants (inanimate‐inanimate, reversible). We replicated the previous findings of a preference for SOV order when describing animate‐inanimate, irreversible events as well as a decrease in the use of SOV when presented with animate‐animate, reversible events. Accompanying the drop in SOV, in a novel condition we observed an increase in the use of SVO and OSV orders when describing events involving two animate entities. In sum, we find that the observed avoidance of SOV order in gestural language creation paradigms when the event includes an animate agent and patient is driven by the animacy of the participants rather than the reversibility of the event. We suggest that findings from gestural creation paradigms are not automatically linkable to spoken language typology.     

Cognitive systems struggling for word order

We argue that the grammatical diversity observed among the world’s languages emerges from the struggle between individual cognitive systems trying to impose their preferred structure on human language. We investigate the cognitive bases of the two most common word orders in the world’s languages: SOV (Subject–Object–Verb) and SVO. Evidence from language change, grammaticalization, stability of order, and theoretical arguments, indicates a syntactic preference for SVO. The reason for the prominence of SOV languages is not as clear. In two gesture-production experiments and one gesture comprehension experiment, we show that SOV emerges as the preferred constituent configuration in participants whose native languages (Italian and Turkish) have different word orders. We propose that improvised communication does not rely on the computational system of grammar. The results of a fourth experiment, where participants comprehended strings of prosodically flat words in their native language, shows that the computational system of grammar prefers the orthogonal Verb–Object orders.     

The semantic origins of word order

Where do the different sentence orders in the languages of the world come from? Recently, it has been suggested that there is a basic sentence order, SOV (Subject–Object–Verb), which was the starting point for other sentence orders. Backup for this claim was found in newly emerging languages, as well as in experiments where people are asked to convey simple meanings in improvised gesture production. In both cases, researchers found that the predominant word order is SOV. Recent literature has shown that the pragmatic rule ‘Agent first’ drives the preference for S initial word order, but this rule does not decide between SOV and SVO. This paper presents experimental evidence for grounding the word order that emerges in gesture production in semantic properties of the message to be conveyed. We focus on the role of the verb, and argue that the preference for SOV word order reported in earlier experiments is due to the use of extensional verbs (e.g. throw). With intensional verbs like think, the object is dependent on the agent’s thought, and our experiment confirms that such verbs lead to a preference for SVO instead. We conclude that the meaning of the verb plays a crucial role in the sequencing of utterances in emerging language systems. This finding is relevant for the debate on language evolution, because it suggests that semantics underlies the early formation of syntactic rules.     

Investigating Constituent Order Change With Elicited Pantomime: A Functional Account of SVO Emergence

One of the most basic functions of human language is to convey who did what to whom. In the world's languages, the order of these three constituents (subject [S], verb [V], and object [O]) is uneven, with SOV and SVO being most common. Recent experiments using experimentally elicited pantomime provide a possible explanation of the prevalence of SOV, but extant explanations for the prevalence of SVO could benefit from further empirical support. Here, we test whether SVO might emerge because (a) SOV is not well suited for describing reversible events (a woman pushing a boy) and (b) pressures to be efficient and mention subjects before objects conspire to rule out many other alternatives. We tested this by asking participants to describe reversible and non‐reversible events in pantomime, and we instructed some participants to be consistent in the form of their gestures and to teach them to the experimenter. These manipulations led to the emergence of SVO in speakers of both English (SVO) and Turkish (SOV).     

The role of frequency in the acquisition of English word order

Akhtar [Akhtar, N. (1999). Acquiring basic word order: Evidence for data-driven learning of syntactic structure. Journal of Child Language, 26, 339–356] taught children novel verbs in ungrammatical word orders. Her results suggested that the acquisition of canonical word order is a gradual, data-driven process. The current study adapted this methodology, using English verbs of different frequencies, to test whether children's use of word order as a grammatical marker depends upon the frequency of the lexical items being ordered. Ninety-six children in two age groups (2;9 and 3;9) heard either high frequency, medium frequency or low frequency verbs that were modeled in SOV order. Children aged 2;9 who heard low frequency verbs were significantly more likely to adopt the weird word order than those who heard higher frequency verbs. Children aged 3;9 preferred to use SVO order regardless of verb frequency. Furthermore, the younger children reverted to English word order using more arguments as verb frequency increased and used more pronouns than their older counterparts. This suggests that the ability to use English word order develops from lexically specific schemas formed around frequent, distributionally regular items (e.g. verbs, pronouns) into more abstract, productive schemas as experience of the language is accrued.     

Sentence comprehension in aphasia in two clear case-marking languages

Studies of aphasia in Indo-European languages point to a selective vulnerability of morphological case marking in sentence comprehension. However, in case-marking languages such as German and Serbo-Croatian, the use of case marking to express formal grammatical gender diminishes the clarity of grammatical role marking. In Hungarian and Turkish, there are simple and reliable markings for the direct object. These markings are not linked to grammatical gender. Compared to Hungarian, the Turkish accusative marking is somewhat lower in availability, but somewhat higher in detectability. The processing of these cues by aphasics was tested using the design of MacWhinney, Pléh, and Bates (1985. Cognitive Psychology, 17, 178-209). Simple sentences with two nouns and one transitive verb were read to Broca's and Wernicke's aphasics, anomics, and control subjects in both Turkey and Hungary. The main effect of case marking was extremely strong. However, this was not true for all groups. The aphasics used the case cue far less than the normals, with the Hungarian Wernicke's group showing the greatest loss. Word order variations were largely ignored in all groups whenever the case-marking cue was present. When case marking was absent, Turkish subjects had a clear SOV interpretation for NNV sentences and Hungarians had a clear SVO interpretation for NVN sentences, in accord with basic patterns in their languages. When there was a contrast between the animacy of the two nouns, subjects choose the animate nouns significantly more often. The effect of animacy was particularly strong in Turkish, in accord with basic facts of Turkish grammar. In Hungarian, VNN sentences without case marking were interpreted as VOS when the first noun was inanimate. In Turkish, VNN sentences without case marking were often interpreted as VSO. In general, the aphasic subjects showed a clear preservation of virtually all aspects of their native languages, albeit in a much noisier form. Despite the high reliability of the case-marking cue, it was damaged more than the word order cue in English subjects. The near-chance processing of the case cue by the Wernicke's aphasics in Hungarian can probably be attributed to the relatively greater difficulty involved in detecting the Hungarian accusative suffix.     

Word Order Typology Interacts With Linguistic Complexity: A Cross-Linguistic Corpus Study

Much previous work has suggested that word order preferences across languages can be explained by the dependency distance minimization constraint (Ferrer-i Cancho, 2008, 2015; Hawkins, 1994). Consistent with this claim, corpus studies have shown that the average distance between a head (e.g., verb) and its dependent (e.g., noun) tends to be short cross-linguistically (Ferrer-i Cancho, 2014; Futrell, Mahowald, & Gibson, 2015; Liu, Xu, & Liang, 2017). This implies that on average languages avoid inefficient or complex structures for simpler structures. But a number of studies in psycholinguistics (Konieczny, 2000; Levy & Keller, 2013; Vasishth, Suckow, Lewis, & Kern, 2010) show that the comprehension system can adapt to the typological properties of a language, for example, verb-final order, leading to more complex structures, for example, having longer linear distance between a head and its dependent. In this paper, we conduct a corpus study for a group of 38 languages, which were either Subject–Verb–Object (SVO) or Subject–Object–Verb (SOV), in order to investigate the role of word order typology in determining syntactic complexity. We present results aggregated across all dependency types, as well as for specific verbal (objects, indirect objects, and adjuncts) and nonverbal (nominal, adjectival, and adverbial) dependencies. The results suggest that dependency distance in a language is determined by the default word order of a language, and crucially, the direction of a dependency (whether the head precedes the dependent or follows it; e.g., whether the noun precedes the verb or follows it). Particularly we show that in SOV languages (e.g., Hindi, Korean) as well as SVO languages (e.g., English, Spanish), longer linear distance (measured as number of words) between head and dependent arises in structures when they mirror the default word order of the language. In addition to showing results on linear distance, we also investigate the influence of word order typology on hierarchical distance (HD; measured as number of heads between head and dependent). The results for HD are similar to that of linear distance. At the same time, in comparison to linear distance, the influence of adaptability on HD seems less strong. In particular, the results show that most languages tend to avoid greater structural depth. Together, these results show evidence for “limited adaptability” to the default word order preferences in a language. Our results support a large body of work in the processing literature that highlights the importance of linguistic exposure and its interaction with working memory constraints in determining sentence complexity. Our results also point to the possible role of other factors such as the morphological richness of a language and a multifactor account of sentence complexity remains a promising area for future investigation.     

Interaction Between Syntactic Structure and Information Structure in the Processing of a Head-Final Language

The effects of syntactic and information structures on sentence processing load were investigated using two reading comprehension experiments in Japanese, a head-final SOV language. In the first experiment, we discovered the main effects of syntactic and information structures, as well as their interaction, showing that interaction of these two factors is not restricted to head-initial languages. The second experiment revealed that the interaction between syntactic structure and information structure occurs at the second NP (O of SOV and S of OSV), which, crucially, is a pre-head position, suggesting the incremental nature of the processing of both syntactic structure and information structure in head-final languages.     

The action-sentence compatibility effect in Japanese sentences

The action-sentence compatibility effect (ACE) is a phenomenon in which a reader's response to a sentence is made faster when there is congruity between the action described in the sentence and the action that makes up the response. Previous studies showed the ACE occurs in action-related sentences in several languages. However, all these were SVO (verb-object) languages, in which verbs are placed before object nouns; this order is reversed in SOV languages. Moreover, those studies investigated hand responses. This study assessed the existence of the ACE in Japanese, an SOV language, and in foot responses. 24 female participants judged the sensibility of Japanese sentences that described actions and responded with either their foot or hand as an effector. Reaction times were significantly faster when there was congruity between the effector described in the sentences and the effector actually used for the response. However, sentence dependency was also found in the foot responses.     

Effects of Case Marking and Word Order on Sentence Parsing in Finnish: An Eye Fixation Analysis

Effects of case marking and word order on syntactic parsing in Finnish were examined by registering readers' eye fixation patterns while they read single sentences for comprehension. Target nouns appearing towards the beginning of the sentence took one of three grammatical roles: subject, object, or adverbial. The subject phrase in the sentence-initial position is the canonical order in Finnish, but the two other word orders are less frequent. In one experimental condition, the grammatical role of the target noun was signalled by a case inflection attached to the preceding adjective modifier; in the second condition this was not the case. The results showed a facilitation effect in sentence parsing due to case marking. Similarly, there was an effect of word order, where the canonical SVO order was associated with greater processing ease than were non-canonical word orders. The two factors interacted so that there was no effect of case marking for the SVO order, but a significant case marking effect for the two marked word orders. The same pattern of results showed up as both immediate and delayed effects. The results speak against the notion of head licensing proposed by Abney (1989) and Prichett (1991).     

Functional constraints on sentence processing: A cross-linguistic study

English and Italian provide some interesting contrasts that are relevant to a controversial problem in psycholinguistics: the boundary between grammatical and extra-grammatical knowledge in sentence processing. Although both are SVO word order languages without case inflections to indicate basic grammatical relations, Italian permits far more variation in word order for pragmatic purposes. Hence Italians must rely more than English listeners on factors other than word order. In this experiment, Italian and English adults were asked to interpret 81 simple sentences varying word order, animacy contrasts between the two nouns, topicalization and contrastive stress. Italians relied primarily on semantic strategies while the English listeners relied on word order—including a tendency to interpret the second noun as subject in non-canonical word orders (corresponding to word order variations in informal English production). Italians also made greater use of topic and stress information. Finally, Italians were much slower and less consistent in the application of word order strategies even for reversible NVN sentences where there was no conflict between order and semantics. This suggests that Italian is ‘less’ of an SVO language than English. Semantic strategies apparently stand at the ‘core’ of Italian to the same extent that word order stands at the ‘core’ of English. It is suggested that these results pose problems for claims about a ‘universal’ separation between semantics and syntax, and for theories that postulate a ‘universal’ priority of one type of information over another. Results are discussed in the light of the competition model, a functionalist approach to grammar that accounts in a principled way for probabilistic outcomes and differential ‘weights’ among competing and converging sources of information in sentence processing.     

Structural and experiential factors in newborns’ preference for speech sounds

It is well established that newborns prefer some sounds over others. It is widely believed that these preferences reflect past experience, either prenatally or in the immediate postnatal environment. However, it is also possible that structural factors are involved. The frequency of occurrence of different sounds in the languages of the world may provide evidence of such structural factors. If structural factors are important, and if frequency of occurrence of sounds in the world’s languages reflects structural factors, then we might expect newborn preference for sounds to reflect their frequency across languages. In this study, newborns were presented with two cv pairs consisting of a trilled /r/ and a bunched /r/ followed by the vowel //. The newborns were either of English‐speaking or Spanish‐speaking origin. The bunched /r/ occurs in American English and is not frequent in the languages of the world. It does not occur in Spanish. The trilled /r/ occurs in Spanish and is a frequent consonant in the languages of the world. It does not occur in American English. Results showed that newborns of both linguistic backgrounds preferred to listen to the trilled /r/. It is hypothesized that the frequency of occurrence of a sound in the languages of the world may reflect structurally based ease of production and ease of perception, the latter leading to the newborn preference.     

Asyntactic thematic role assignment by Mandarin aphasics: a test of the Trace-Deletion Hypothesis and the Double Dependency Hypothesis

This study examines the comprehension patterns of various sentence types by Mandarin-speaking aphasic patients and evaluates the validity of the predictions from the Trace-Deletion Hypothesis (TDH) and the Double Dependency Hypothesis (DDH). Like English, the canonical word order in Mandarin is SVO, but the two languages differ in that the head noun precedes the relative clause in English, but it follows the relative clause in Chinese. According to the Default Principle as stated in the TDH, the word order discrepancy will make subject relative clauses more difficult to comprehend for Mandarin agrammatics than object relative clauses, but the DDH predicts that agrammatic patients from the two languages have the same pattern of selective deficits. The results of this study support the prediction of the TDH.     

BASIC RELATIONS IN CHILD LANGUAGE AND THE WORD ORDER MYTH

The research involved an evaluation of the psychological reality of the syntactic functions of subject and object, the semantic functions of agent and patient, and the discourse functions of given and new. The domain of analysis was from the early two-word into the early three-word phase (approximately 1;7 to 2;5) of the acquisition of the relatively synthetic language of Polish. An analysis of the inflectional morphology revealed the appropriate use of all seven cases and contrasts within three dimensions of agreement before 1;9. These varied inflectional forms were used with sufficient frequency to argue that the nominative case and the agreement relations are productive subjecthood properties at this phase of development. The semantic function of the subject noun phrase was either an agent or a patient. Hence, case and agreement cannot be reduced to properties of semantic functions. Since both subject and object were rarely realized simultaneously in the surface structure in the early phase of two word utterances, the set of verb plus subject and verb plus object patterns (i.e., SV, VS, OV, and VO) were analyzed for the discourse functions of given and new. Neither discourse function nor cannonical (SV and VO) form control word order. As both subject and object came to be realized in the surface structure, subject before object priority was revealed as a subjecthood property but not in a rigid word order (i.e., both SVO and SOV patterns occurred). The evidence demonstrates the independent status of the syntactic function of subject.     

The interaction of word frequency and word class: a test of the GO model's account of the missing-letter effect

When asked to detect target letters while reading a text, participants miss more letters in frequent function words than in less frequent content words. In this phenomenon, known as the missing-letter effect, two factors covary: word frequency and word class. According to the GO model, there should be an interaction between word class and word frequency with more omissions for function than for content words only among high-frequency words. This pattern would be due to the fact that function words could only assume a structure-supporting role if they are identified rapidly, which is only possible for high-frequency words. These predictions were tested by assessing omission rate for frequent and rare function and content words. Results lend support to the GO model with more omissions for frequent than for rare words, and more omissions for the function than for the content word among high-frequency words, but not among low-frequency words. These results were observed both in English (Experiment 1) and in French (Experiment 2).     

When it helps to be misled: Facilitative effects of increasing the frequency of conflicting stimuli in a Stroop-like task

Three experiments are reported that involve responding to the meaning or position of a word (Above or Below) presented above or below a fixation point. Position and word meaning conflicted (Above/below or Below/above) or were compatible (Above/above or Below/below), and the relative frequency of conflicting trials was varied. Experiment 1 required responses to the word and its position. Compatibility and frequency had no effect in the spatial task, but interacted strongly in the word task: Compatible stimuli were processed faster when conflicting trials were rare (20% conflicting), but conflicting stimuli were processed faster when they were frequent (80% conflicting). Experiments 2 and 3 used the word task only and extended these findings to intermediate (20%, 40%, 60%, and 80% conflicting) and more extreme (10%, 20%, 80%, and 90% conflicting) frequencies, respectively. The advantage for conflicting stimuli when they were frequent was taken as evidence for a strategy involving dividing attention between reported and unreported dimensions.     

Interactions between sentence context and word frequencyinevent-related brainpotentials

Event-related brain potentials (ERPs) were recorded as subjects silently read a set of unrelated sentences. The ERP responses elicited by open-class words were sorted according to word frequency and the ordinal position of the eliciting word within its sentence. We observed a strong inverse correlation between sentence position and the amplitude of the N400 component of the ERP. In addition, we found that less frequent words were associated with larger N400s than were more frequent words, but only if the eliciting words occurred early in their respective sentences. We take this interaction between sentence position and word frequency as evidence that frequency does not play a mandatory role in word recognition, but can be superseded by the contextual constraint provided by a sentence.     

The interaction of word frequency and word class: A test of the GO model's account of the missing-letter effect

When asked to detect target letters while reading a text, participants miss more letters in frequent function words than in less frequent content words. In this phenomenon, known as the missing-letter effect, two factors covary: word frequency and word class. According to the GO model, there should be an interaction between word class and word frequency with more omissions for function than for content words only among high-frequency words. This pattern would be due to the fact that function words could only assume a structure-supporting role if they are identified rapidly, which is only possible for high-frequency words. These predictions were tested by assessing omission rate for frequent and rare function and content words. Results lend support to the GO model with more omissions for frequent than for rare words, and more omissions for the function than for the content word among high-frequency words, but not among low-frequency words. These results were observed both in English (Experiment 1) and in French (Experiment 2).