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.
This paper aims to contribute to our understanding of word order universals related to “grammatical weight” in OV vs. VO languages, by presenting data from Persian, an SOV language with mixed head direction. Grammatical weight (or heaviness) refers to the structural complexity and/or the length (number of words) of a constituent in relation to other constituents of a sentence. The first generalization on the matter was originally formulated by Behaghel (
In psycholinguistics and cognitive sciences, this phenomenon favored availability-based incremental models of sentence production that assumed a universal “end-weight” (or short-before-long, more commonly used in this literature) ordering preference (e.g. De Smedt (
These mirror-image preferences presently seem to be well-established in the literature and are broadly assumed to result from a general/universal principle, according to which minimizing dependency distance facilitates language processing and comprehension (e.g. Hawkins (
A production-oriented account has also been proposed by Yamashita & Chang (
In this paper, we present sentence production data from Persian, studying the effect of the length in interaction with factors that are related to conceptual accessibility. While previous studies investigated the effect of length in transitive and ditransitive constructions, they do not take into account factors such as definiteness and animacy: 1) In experimental studies, namely on Japanese (
Indeed, interactions between the effect of these factors are not expected if we assume that length-based preferences derive from the tendency to minimize the distance between the verb and its complements, or, similarly, to avoid center-embedded constructions (
A length-based effect corresponding to the long-before-short preference has been documented for Persian by corpus-based (
Another issue is to determine what metric of the head-dependents distance provides more accurate predictions. Interestingly, syntactic properties of Persian allow us to tease apart the two main available measures, that is, Temperley’s “Dependency Length Minimization”
We have conducted two complementary experimental studies in order to explore cases that are problematic for a dependency-distance minimization account. Based on our findings, we will argue that both a parsing-oriented account in terms of dependency-distance minimization and a production-oriented account in terms of the conceptual accessibility hypothesis can be advantageously recruited to explain word order preferences in Persian, and arguably in other languages.
In the following sections, we will first provide an overview of the relation between grammatical weight and word order, along with the available accounts of the long-before-short preference in SOV languages. In Section 3, we will introduce Persian and its interest for the study of length-based effects on word order, while discussing the findings of previous studies. We will present our two experimental studies in Section 4 and discuss our results in Section 5.
The study of the role of grammatical weight (or heaviness) on the linear order of the constituents in a sentence is an important research topic in language sciences, including theoretical linguistics, psycholinguistics, as well as linguistic typology. It is an old topic, yet continues to motivate ongoing theoretical debates. Interest in the topic within general syntax goes far back. The first generalization on the matter, known as the “end-weight principle”, was originally formulated by Behaghel (
Early accounts of end-weight are formulated in terms of language processing and comprehension (
Indeed, a straightforward account of the end-weight effect can be framed in terms of availability-based incremental models of sentence production (e.g. Garrett (
However, this account implies that the end-weight preference should be universal, as it has indeed been explicitly or implicitly assumed in many studies (see Hawkins (
Two types of hypotheses are available to account for the long-before-short preference in OV languages: 1) parsing-oriented and 2) production-oriented.
Dependency-based distance-minimizing accounts motivated by efficient processing, such as the Early Immediate Constituent (EIC) or the Minimize Domains (MiD) principles (
Hawkins (
DLT and DLM, on the other hand, calculate the cumulative distance between the verb and the heads of all its dependents, referred to as the dependency length, and predict a preference for the sentence with a shorter dependency length. The two models differ substantially. While Gibson’s DLT only takes into account constituents with “new discourse referents” (
Recall that accessibility-based incremental models of sentence production predict a universal end-weight preference. To account for both long-before-short and short-before-long preferences in a unified manner, that is, via the same model of sentence production, Yamashita & Chang (
Their rationale is that: 1) since both conceptual and form-related factors are shown to influence word order preferences (see
Each of these two hypotheses (parsing-oriented or production-oriented) tackles the problem from a different angle. Hence, they are not contradictory and can be assumed to hold simultaneously, as long as they are not falsified by the data. Furthermore, in previous studies, the predictions of these hypotheses converge for the majority of data discussed – coming from strictly head-final languages such as Japanese (
Parsing-oriented (distance-minimizing) accounts nevertheless seem to have gained more approval among researchers because these accounts maintain the same mechanism for VO and OV languages. Moreover, a couple of recent large-scale cross-linguistic corpus studies conducted on available dependency tree-banks support the universality of dependency-distance minimizing hypothesis (e.g. Futrell et al. (
Yamashita & Chang’s (
However, there are also studies that pay credit to this language-specific availability-based model and adhere to the conceptual accessibility hypothesis. For instance, Kempen & Harbusch (
It is important to bear in mind that salience is a complex and multi layered notion, and it is crucial to capture its various dimensions and the ways in which they interact. Conceptual accessibility of constituents, for instance, remains a vague and sometimes even confusing notion in the literature, especially with regard to its relation with discourse-related accessibility and the information structure of the sentence (e.g. topics are accessible entities and are generally realized by short/simple constituents such as pronouns).
There however seems to be an agreement on the definition of accessibility in psycholinguistic literature, as “the ease with which the concept associated with a noun phrase (NP) can be retrieved from memory”, as well as a consensus that the latter is “one of the most influential factors” in the processing and resolution of ambiguous pronouns (
Persian is an SOV and a three-way
In addition, Persian exhibits differential object marking (DOM),
(1)
a.
Mahsā
Mahsa
(in)
(this)
medād(=e
pencil(=
qermez)=rā
red)=
xarid
buy.
‘Mahsa bought (this/) the (red) pencil.’
b.
Mahsā
Mahsa
medād(=e
pencil(=
qermez)
red)
xarid
buy.
‘Mahsa bought (red) pencils/a (red) pencil.’
c.
Mahsā
Mahsa
yek
a
medād(=e
pencil(=
qermez)
red)
xarid
buy.
‘Mahsa bought a (red) pencil.’
DOM has important bearings on the canonical order between the NP and PP complements. While, in previously studied OV languages, the canonical order is given as S-IO-DO-V with no nuances or controversies, positing a canonical order in Persian ditransitive constructions is neither straightforward nor uncontroversial. In this paper, we rely on the following generalizations supported by quantitative studies (e.g. Faghiri et al. (
Interestingly, as we will show in the next section, different distance-minimizing accounts discussed in Section 2.2 do not predict the same length-based ordering preferences for Persian, contrary to previously investigated OV languages such as Japanese, Korean or Basque. Hence, the Persian data provide us with cases that make it possible to tease apart between, on one hand, Hawkins’s MiD and, on the other, DLT and DLM. We limit our discussion to the predictions of DLM as a less restrictive version of Gibson’s DLT and will not discuss DLT separately.
As noted by Temperley (
In this section, we present the predictions of MiD and DLM for Persian data discussed in this paper. In Section 2.2, we saw that these models both account for length-based ordering preferences in OV and VO languages from a processing point of view by positing that some orders are less complex to process. However, they build on different measures of complexity. While both measures depend on the relative length between the constituents involved, they differ in the way the dependency distance is operationalized.
Recall that MiD depends on the size of the Constituent Recognition Domain (CRD) and takes into account the number of words needed to be parsed in order to obtain all immediate constituents (IC) of the sentence. Between two alternative word orders, MiD votes for the sentence that yields a greater IC-to-Word ratio, that is, for an equal number of ICs, a sentence with a smaller CRD is preferred. This is illustrated by the pair of examples in (2) from Hawkins (
In (2a), Verb, PP1 and PP2 can be recognized on the basis of five words (italicized in the example). The CRD contains 5 words and hence the sentence has an IC-to-Word ratio of 3/5 (60%). (2b), on the other hand, has an CRD of 9 words and an IC-to-Word ratio of 3/9 (33%). Consequently, (2a), which reflects a short-before-long ordering, is considered to be less complex and easier to process than (2b), and thus should be preferred.
(2) | a. | |
b. |
Hawkins has also proposed a more fine-grained metric that allows the ranking of two sequences with the same CRD score.
(3) | a. | |
b. |
DLM, on the other hand, represents complexity of a sentence as the total sum of all the lengths of its dependencies. The length of a dependency is defined as the number of words spanned; a dependency connecting adjacent words is considered to have a length of 1 (
(4) | a. | |
b. |
In a consistently head-final language such as Japanese, as illustrated by schematic examples in (5) from Temperley (
(5) | a. | |
b. |
Finally, it is worth highlighting that besides the direction of length-based preferences, both models predict that the strength of preference depends on the size of reduction in complexity (or the gain in efficiency), which amounts to the difference between the complexity/efficiency measures of the two alternative orderings, that is, the difference between total dependency lengths in DLM, and the difference between CRDs or IC-to-Word ratios in MiD. This difference is in turn directly related to the relative length of the constituents involved. In other words, the rate of length-based shifts is expected to increase with the relative length of the constituents.
In what follows, we present the predictions of these two models for a number of cases of constituent ordering variation in Persian where the constituents differ in length. Distinct predictions are provided for
For illustration, we take a length difference of 4 words between the two constituents and consider the two possibilities where one constituent has a phrasal length of 6 and the other 2 and vice versa. We use Hawkins’s more fine-grained metric when relevant. Note that the latter is not taken into account by Temperley (
We calculate the measures in minimal-pair sentences using schematic examples similar to (4) and (5) for simplification. Each sentence contains 3 ICs including the V(erb). We use bold to mark the head (constructing category) of the other constituents. In PPs, this would be the P (reposition), and in NPs, the first/leftmost element of the NP, which can be a determiner or a noun (represented by X that stands for any word). For
1) | Non- |
In (6) and (7), we consider the relative order between an NP and a PP complement in the preverbal domain. In (6), we consider the case where the PP is longer than the NP and in (7) the reverse.
(6) | a. | |
b. |
(7) | a. | |
b. |
Predictions:
In both (6) and (7), both pairs have the same CRD but (b) has a greater aggregate IC-to-Word ratio. The gain in efficiency is the same (10.6%) in each pair; (a) has a shorter total dependency length. The reduction is the same (4 words) in each pair.
MiD predicts a preference for (b) that reflects a short-before-long ordering.
DLM predicts a preference for (a) that reflects a long-before-short ordering.
The strength of preference depends on the relative length between the two constituents – regardless of the direction of the difference (NP>PP or NP<PP).
In (8), we consider the relative order between the subject and the DO in the preverbal domain. Note that here we provide one pair of examples to illustrate both the case where the subject is longer than the DO and the case where the subject is shorter. In each example, NP1 and NP2 can represent respectively the subject and the DO (Subj>DO) and vice versa (Subj<DO).
(8) | a. | |
b. |
Predictions:
Both sentences have the same CRD but (b) has a greater aggregate IC-to-Word ratio. The gain in efficiency is the same in each case; (a) has a shorter total dependency length. The reduction is the same (4 words) for each case.
MiD predicts a preference for (b) that reflects a short-before-long ordering.
DLM predicts a preference for (a) that reflects a long-before-short ordering.
The strength of preference depends on the relative length between the two constituents – regardless of the direction of length difference (Subj>DO or DO>Subj).
2 |
In (9) and (10), we consider the relative order between an NP and a PP complement in the preverbal domain. In (9), we consider the case where the PP is longer than the NP and in (10) the reverse. Recall that in the latter case the NP contains a relative clause of 4 words that appears after
(9) | a. | |
b. |
(10) | a. | |
b. |
Predictions:
Sentence (9b) has a smaller CRD and hence a greater IC-to-Word ratio; Sentence (10a) has a smaller CRD and hence a greater IC-to-Word ratio. The gain in efficiency is the same in each case; Sentences (9a) and (10a) have shorter total dependency lengths. The reduction is the same (4 words) in each pair.
MiD predicts a preference for NP-PP-V order regardless of the relative length.
DLM predicts a preference for (a) that reflects a long-before-short ordering.
The strength of the preference depends only on the relative length between the two constituents – regardless of the direction of length difference.
In (11) and (12), we consider the relative order between the subject and the DO in the preverbal domain. (11) illustrates the case where Subj>DO and (12) the case where Subj<DO.
(11) | a. | |
b. |
(12) | a. | |
b. |
Predictions:
Sentence (11b) has a smaller CRD and hence a greater IC-to-Word ratio; Sentence (12b) has a smaller CRD and hence a greater IC-to-Word ratio. The gain in efficiency is the same in each pair; Sentences (11a) and (12b) have shorter total dependency lengths. The reduction is the same (4 words) in each pair.
MiD predicts a preference for the DO-first order regardless of the relative length.
DLM predicts a preference for a long-before-short ordering.
The strength of the preference depends on the relative length between the two constituents – regardless of the direction of length difference.
To sum up: in different pair-wise comparisons examined here, DLM consistently predicts a long-before-short preference that increases with the relative length between the two constituents, regardless of the choice of analysis for
A number of quantitative studies have already addressed the issue of grammatical weight in Persian. Rasekh-Mahand et al. (
In line with studies on the heavy NP shift in different SOV and SVO languages, Faghiri and colleagues (
Faghiri & Samvelian (
In the vein of constrained production experimental studies on the heavy NP shift, Faghiri et al. (
Furthermore, Faghiri et al. (
Faghiri et al. (
To resume, previous studies report length-based ordering variations corresponding to a long-before-short preference in the relative order between the DO and the PP argument, limited to non
In their attempt to provide an explanation for these observed length-based effects, Faghiri and colleagues favor a production-oriented account over parsing-oriented accounts. They argue that Yamashita & Chang’s (
Length-based ordering preferences depicted by available studies so far present some potential challenges for dependency-distance minimizing accounts.
The cases for which previous studies report zero length-based effects run counter to the predictions of DLM. According to the latter the long-before-short preference is expected to trigger ordering variations for all types of DOs.
Although MiD may account for the absence of a length-based effect in the case of
The relatively important rise in the rate of shifted orders observed for a two-word length difference, when the relative length is increased by adding two attributive modifiers to bare DOs, is intriguing.
We have conducted two more production experiments to follow up on these observations. These experiments are carried out in order to 1) pin down the effect of phrasal length in terms of conceptual enrichment vs. dependency-distance by comparing simple/short vs. modified/long non-
In this section, we present the results of two constrained sentence production experiments carried out to study the ordering preferences of Persian native speakers. The task used in these experiments is identical to the one used by Faghiri et al. (
All our data are analyzed via statistical open access software R (
Our experiments are implemented via web-based self-administrated questionnaires
This production paradigm is inspired by the
In the design used in our experiments, participants are asked to make a sentence to complete a preamble with phrases that appear on the screen. Participants see simultaneously a preamble, an incomplete sentence represented by blank boxes and a vertical list of phrases (that appear in blue). A screenshot of a (filler) item is given in Figure
Screenshot of an experimental item (English equivalents are added).
Each experimental item contains a sentence in which three constituents are missing, represented by three blank boxes. To complete the sentence, participants are provided with four phrases (presented in a counterbalanced order). Participants are instructed to 1) read the preamble and the list of phrases, 2) complete the sentence with the most natural continuation that comes to their mind using three of the four given phrases, 3) fill in the blanks accordingly, and 4) click on “Continue” to go to the next sentence.
The list of options contains one element more than the number of blanks in order to prevent participants from guessing the purpose of the experiment and to push them to concentrate on the content of each sentence, in order to produce reasonably natural sentences. The relative order between these constituents (in the final sentence) is left to the participants and constitutes the response or the dependent variable.
This experiment targets the effect of phrasal length by adding an attributive (restrictive) modifier to non
Furthermore, in order to disentangle the effect of conceptual enrichment from the effect of increasing the dependency-distance, we manipulate the length by adding only a one-word attributive modifier. Recall that the strength of a dependency-length preference depends on the size of reduction in the length difference between two alternatives. A one-word length difference presents the least possible reduction in the dependency length. Note that the lowest rate of length-based shifts observed so far (in Persian data) is about 10% and was triggered by a three-word length difference (see Section 3.4). Consequently, a comparably strong effect in this configuration can hardly be viewed as a dependency-minimizing effect.
Finally, in Faghiri et al.’s study, PPs were inanimate in all sentences and the results showed an overall higher rate of NP-PP-V order with respect to another experiment with bare DOs but including only animate PPs (see
A set of 16 sentences was created for this experiment, following a 2 × 2 design with DO type (bare vs. indefinite) and DO length (simple vs. modified with a one-word attributive modifier) as within-item variables, where we prepared 4 versions of each sentence, as in (13). In half of sentences the PP was human (construed as a goal or a source argument and involving various prepositions, e.g.
Each experimental item was preceded by a preamble containing the subject and a (vertical) list of four constituents: a PP, two choices of formally identical NPs and a verb. The order of the list was counterbalanced (between PP over NP and NP over PP) across items. The dependent variable is the order between the three remaining constitutes (PP, NP and Verb) in sentences filled out by participants. However, expecting non-verb final orders to be scarce, the comparison will be limited to NP-PP-V vs. PP-NP-V.
(13)
Ali hamiše …
Ali always …
[NP1
(yek
a
mošt)
handful
gerdu(=ye
walnut=
tāze)]
fresh
[PP
tu=ye
in=
sālād]
salad
[V
mi-riz-ad]
[NP2
(yek
a
meqdār)
quantity
serke(=ye
vinegar=
sib)]
apple
‘Ali always puts (a handful of) (fresh) walnuts / (a small quantity of) (apple) vinegar in the salad.’
These 16 experimental items were combined with 24 filler items. The final list of items, in which target items were spaced by at least one filler, was randomized for each participant individually. It contained an additional filler item appearing as the first item for all participants.
80 native speakers of Persian (39 women and 41 men; mean and median age: 33 and 31.5 years) volunteered to complete our web-based questionnaire (40 for each sub-experiment) – the exact number of participants was 97 but we discarded data from bi/multilingual speakers that did not declare Persian as their dominant language. Data from two participants in the indefinite sub-experiment was excluded from the final dataset because they did not fill out sentences according to the instructions. There were also a few erroneous answers, which we marked as NA.
Table
Distribution of word order for Experiment 1.
Bare Dos | |||||
---|---|---|---|---|---|
NPV | NVP | PNV | NA | Total | |
Short | 26 | 0 | 301 | 1 | |
Long | 55 | 1 | 268 | 4 | |
Anim | 25 | 0 | 300 | 3 | |
Inan | 56 | 1 | 269 | 2 | |
Total | |||||
Short | 154 | 1 | 148 | 1 | |
Long | 198 | 1 | 104 | 1 | |
Anim | 164 | 0 | 139 | 1 | |
Inan | 188 | 2 | 113 | 1 | |
Total |
NP-PP-V rate by DO type and DO length in Experiment 1 (the error bars present 95% confidence intervals).
NP-PP-V rate by DO type and animacy of PP in Experiment 1.
The overall rates for NP-PP-V are 12.5% for bare DOs, and 58.3% for indefinite DOs. These rates are in line with the rates reported for these DO types in previous experimental studies. We observe that the rate of NP-PP-V significantly increases with longer/modified NPs for both bare and indefinite types, respectively, from 8.0% to 17.0% (
We analyzed all the data (a total of 1254 data points excluding miscellaneous orders) using an GLMM model including items and participants as random effects and DO type, DO length and PP animacy (sum-coded) as fixed effects. The NP-PP-V order is coded as success. We find a significant effect for the DO type (Est. = –1.94, SE = 0.22, p < 0.001) as well as for length (Est. = 0.55, SE = 0.11, p < 0.001), but no significant effect for animacy nor any significant interactions between the variables (ps > 0.10).
We then fitted two models separately for bare and indefinite DOs (a total of, respectively, 650 and 604 data points), to see whether there is a numerical difference between the estimated coefficients and in what direction. The estimated coefficient of length is greater for bare DOs (Est. = 0.64, SE = 0.17, p < 0.001) than for indefinite DOs (Est. = 0.38, SE = 0.11, p < 0.01).
In sum, our data show a robust effect of phrasal length for both non-
With respect to the animacy of the PP, in line with the general animate-before-inanimate preference, we observe that overall in our data the rate of PP-NP-V orders increases with animate PPs. We are not going to comment further on the effect of animacy which is beyond the scope of this paper.
In this experiment, we study the effect of the relative length (by adding a relative clause to the PP argument) for
A set of 16 sentences was created for this experiment. Following a 2 × 2 design with DO type (
(14)
Parvin …
Parvin …
[NP1
yek
a
arusak
doll
/ arusak=rā]
/ doll=
[PP
barāye
for
Zohre
Zohreh
(ke
that
dāšt
gerye
cry
mi-kard)]
[NP2
yek
a
ābnabāt
candy
/ ābnabāt=rā]
/ candy=
[V
xarid]
buy.
‘Parvin bought the/a doll/candy for Zohreh (who was crying).’
These 16 experimental items were combined with 24 filler items. The final list of items, in which target items were spaced by at least one filler, was randomized for each participant individually. It contained an additional filler item appearing as the first item for all participants.
34 native speakers of Persian (16 women and 18 men; mean and median age: 32 and 33 years) volunteered to complete our web-based questionnaire – the exact number of participants was 36 but we did not include bi/multilingual speakers who did not declare Persian as their dominant language. There were also a few erroneous or incomplete answers that we marked as NA.
The overall rates of NP-PP-V order are 80.4% for
NP-PP-V rate by DO type and Preposition in Experiment 2.
Distribution of word order for Experiment 2.
NPV | NVP | PNV | NA | Total | |
---|---|---|---|---|---|
Short PP | 71 | 0 | 64 | 1 | 136 |
Long PP | 43 | 1 | 92 | 0 | 136 |
Short PP | 118 | 0 | 18 | 0 | 136 |
Long PP | 100 | 0 | 35 | 1 | 136 |
NP-PP-V rate by DO type and PP length in Experiment 2.
We analyzed all the data (a total of 541 data points excluding miscellaneous orders) using a GLMM model including preposition types, items and participants as random effects and DO type and PP length (both sum-coded) as fixed effects. The NP-PP-V order is coded as success. We find a significant effect for the DO type (Est. = 1.10, SE = 0.12, p < 0.001) as well as for PP length (Est. = –0.52, SE = 0.11, p < 0.01), but no significant interaction between the two variables (p = 0.85).
We also fitted the model separately for each DO type in order to check whether there is a numerical difference between the estimated coefficients of length between them. In addition, in order to have more homogeneous data for comparison, we also used a limited dataset including only items with the preposition
In sum, our data show a robust length-based effect corresponding to the same long-before-short preference for both DO types. This effect is in accordance with the predictions of DLM, and importantly, it contradicts Faghiri & Samvelian’s (
Experiment 1 shows that there are strong length-based word order variations for which an account in terms of distance minimization is irrelevant. The effect of distance minimization on parsing is expected to be proportional to the relative length. In this experiment, the relative length between the PP and the NP varies by only one word between the two conditions. Hence, dependency-distance minimizing models do not predict a large effect, if any.
Meanwhile, an account in terms of the contribution of length to semantic enrichment/informativity is more satisfactory. One could safely assume that a restrictive modifier adds additional information to a non-specific NP, making its reference more specified/salient. In addition, we note that this effect conforms to observations made by Faghiri et al. (
If we are on the right track, such manipulation of length should yield a fairly smaller effect in the case of (definite)
In Experiment 2, the relative length between the PP and the NP is manipulated by adding a (four-word length) relative clause to the PP. The results clearly show that a long-before-short preference exists regardless of markedness and/or definiteness of the DO. Hawkins’s EIC/MiD model falls short of accounting for the long-before-short preference altogether. The predictions of DLM, on the other hand, are met by the data: we find significant main effects for both relative length and DO type but no interactions between the two variables. This implies that the long-before-short preference is independent of the DO type.
It is worth noting that although distance minimization is relevant, we nevertheless observe that relative length has a much smaller effect size than DO type, which further supports the claim that the relative order between the PP argument and the DO is mainly determined by the DO type. Hence, it is not surprising that the rate of conformity to DLM, that is, the ratio of observed vs. expected shifts, is only 31% (13.6% and 48.1% respectively for
At this point, it is interesting to compare Persian with other languages for which similar data are available. Below, we summarize data from studies on two SVO languages, French and English, and two SOV languages, Japanese and Basque. All these studies report data from comparable production experiments. The idea is to compare the rate of DLM-triggered shifts, that is, cases where there is a default/canonical word order and a heavy constituent shift (leftward or rightward) from its default position. We calculated the rate as the ratio of observed shifted cases, considering the expected order to be the non-shifted order. In Persian data, this rate, calculated for
In a study on French focusing on indefinite DOs, the rate of conformity to DLM is about 70%, with about 53.7% of shifts for heavy NPs (
In a study on heavy NP shift in English by Stallings & MacDonald (
In the experimental study on Basque, the overall rate of shifts (to the left) with heavy NPs is above 60.1% (
In the experimental study on Japanese, the overall rate of shifts (to the left) with heavy NPs is 48.4% (
With respect to the relative order between the subject and the object in transitive sentences, based on the available data (see Section 3.3), we can safely conclude that the magnitude of the DLM effect is smaller in Persian transitive sentences compared to other SOV languages so far investigated. Recall that in their well-powered production experiment manipulating the length of the DO by adding a relative clause, Faghiri et al. (
We can conclude from this comparison that the effect of DLM in Persian is less strong than in these languages. A similar observation has been made by Gildea & Temperley (
Ros et al. (
In this paper, we have studied the effect of phrasal length on word order in Persian. Our data confirm a general long-before-short preference in line with other studies on SOV languages investigated so far (e.g. Basque, Korean and Japanese), contra the universal end-weight principle supported by availability-based models. This is important, because unlike previously studied OV languages, Persian is not consistently head-final and displays a mixed head direction.
We have provided solid experimental evidence for a dependency-distance-minimizing effect in Persian that previous corpus and experimental studies failed to detect (
Furthermore, we have shown that there is also enough empirical evidence to maintain the conceptual accessibility hypothesis (
Finally, we have pointed out the fact that while dependency distance minimization is relevant for Persian, it is reflected less strongly in this language compared to other languages for which comparable data is available. In particular, in transitive sentences no dependency-distance-minimizing effect has been detected so far, which contrasts with what is reported for other studied SOV languages.
This is intriguing because Persian is considered an SOV flexible language, and, importantly, allows for different constituents to be placed in the postverbal domain. These findings may entail that the SOV order is more grammaticalized in Persian (than in these other languages) and thus constitutes a stronger parsing cue in this language.
Another difference between Persian and other investigated SOV languages is its “aberrant” properties with respect to word order typological universals, namely the fact that it displays also head-initial structures. Crucially, clausal verbal complements always occur post-verbally and a relative clause that modifies a preverbal constituent can be placed after the verb. Consequently, there is also enough evidence for a short-before-long preference in the postverbal domain that also shows a solid tendency for leftward ordering of heavy constituents. It could be the case that distance minimization is stronger in the postverbal domain that in the preverbal domain in Persian.
More experiments are required to test these assumptions and to pin down the respective contribution of different parsing cues. Also, crosslinguistic studies involving similar languages will certainly be promising in order to investigate the respective role of these parsing cue in relation to language specific typological properties.
The additional files for this article can be found as follows:
List of experimental items. DOI:
Full summary of the GLMM models presented in Section 4. DOI:
3 = third person,
It should be noted that the cross-linguistic generalizations regarding head-final vs. head-initial languages made by Hawkins are called into question by other more recent typological studies involving African languages in particular (see e.g. Dimmendaal (
There are of course a number of detailed corpus studies on different weight-related ordering phenomena in a given language that take into account a bundle of factors, such as animacy, definiteness, givenness, semantic relatedness, alongside constituent weight/length. In particular, in English, a number of studies (corpus-based and/or experimental) have shown that the end-weight preference applies notwithstanding other factors (e.g. Arnold et al. (
With respect to grammatical roles, it is worth mentioning that in Japanese and Basque, longer DOs are more likely to shift over an IO than over a subject (
It should be noted that the tendency for the postposition of heavy constituents after the verb has also been reported in a corpus study by Rasekh-Mahand et al. (
Note that some studies (e.g. Futrell et al. (
A similar account attributes both heavy-first and heavy-last preferences to the tendency to avoid center-embedded complex structures that cause extra difficulty for both production and processing/comprehension (
Hawkins initially defines the EIC principle as “The human parser prefers to maximize the left-to-right IC-to-word ratios of the phrasal nodes that it constructs.” (
The MiD principle is defined as: “The human processor prefers to minimize the connected sequences of linguistic forms and their conventionally associated syntactic and semantic properties in which relations of combination and/or dependency are processed. The degree of this preference is proportional to the number of relations whose domains can be minimized in competing sequences or structures, and to the extent of the minimization difference in each domain.” (
Hawkins defines CRD as follows: “The CRD for a phrasal mother node M consists of all non-terminal and terminal nodes dominated by M on the path from the terminal node that constructs the first IC on the left to the terminal node that constructs the last IC on the right.” (
Temperley (
Moreover, the corpus study by Rasekh-Mahand et al. (
Indeed, different semantic and pragmatic properties related to conceptual accessibility are known to influence ordering preferences, including animacy (e.g. Bock et al. (
Note that pronouns are generally used by speakers, as a choice of referring expression, to refer to highly accessible references, that is, referents that are (supposed to be) already activated (in the memory of listeners/comprehenders) and/or entities that are easily identified in the context.
In other words, it is common in studies on pronoun ambiguity resolution to assess the accessibility of a referring expression by its potential to be identified as the antecedent of a pronoun by listeners/comprehenders (see Karimi & Ferreira (
In Persian NPs, unbound determiners, quantifiers as well as classifiers precede the head noun and all dependents (adjectives or adjective phrases, PP modifiers, the possessor NP, and the relative clause) follow the head noun.
Coined by Bossong (
DOM is a well-known feature of Persian, yet the object of ongoing controversial debate with no uncontroversial or straightforward account available in the literature (see Samvelian (
(i) a. in this medād=rā pencil= [ke that qermez red ast] is xarid-am buy. b. in this medād pencil [ke that qermez red ast]=rā is= xarid-am buy. ‘I bought this pencil which is red.’
The noun remains in the singular form even when it denotes more than one entity.
The enclitic =(y)e, the Ezafe, links the head noun to its modifiers and to the possessor NP (see Samvelian (
Theoretical studies, however, have grouped indefinite DOs with bare DOs, claiming that the former occur in the same linear position as the latter, which is adjacent to the verb. This has been argued to constitute strong evidence in favor of the hypothesis of two distinct syntactic positions for the DO in Persian, depending on its markedness (see Faghiri & Samvelian (
Recall that these two only differ with respect to the words that should be included when measuring the dependency length (all words or only those introducing a new discourse reference).
Temperley (
“[T]he data reveal a consistent preference for ‘short-long’ ordering, contrary to both theories. We should note, however, that the body of data available here is very small (87 cases). While this particular test is inconclusive, it points to a possible way of testing the DLT against the EIC theory, given further data.” (
The constructing category of an NP constituent is considered to be the determiner or the head noun (see
Hawkins argues that “head categories such as P immediately project to mother nodes such as PP, enabling the parser to construct and recognize them online.” (
In Hawkins’s subsequent work, this proposition is subsumed in a broader principle, namely the Maximize Online Processing (MaOP) principle (see
Recall that in languages that display case-marking, Hawkins assumes the case-marker to be the constructing category of a case-marked NP (
Recall that in Persian, a relative clause modifying a preverbal noun can be extraposed after the verb:
(i) yek a mard man āmad come. [ke that hame=rā all= mi-šenāxt] ‘A man came who knew everybody.’
The estimated coefficient is 0.844 (SE = 0.261, p < 0.01), with an intercept of 1.593 (SE = 0.295, p < 0.001), when NP-PP-V is coded as success (
Note that for simple bare (single word) nouns, not included in Faghiri & Samvelian’s (
Note that it very difficult to manipulate the length of indefinite and bare DOs in Persian by adding a relative clause, since the latter usually triggers
The estimated coefficient for the main effect of length is not significantly different from zero (p > 0.05).
The effect size (Cohens w) of length, which we have calculated from the contingency table provided by Faghiri et al. (
A two- to four-word difference between the NP>PP and NP<PP conditions (by adding attributive modifiers to the NP and an RC to the PP) in Faghiri et al.’s (
Models that did not converge or had perfect correlations between random effects were reduced.
This means that these experiments have been designed specifically to be completed by a respondent (on their own computer) without intervention of the researchers.
Also, as mentioned by an anonymous reviewer, another rather straightforward way to test whether semantic richness per se plays a role on word order preferences is to manipulate the degree of informativity of nouns: e.g. “person” vs. “doctor” vs. “pediatrician”.
It is important to bear in mind that in Basque (
Note that their comparison is based on overall (uncontrolled) corpus-based calculation of the dependency length. Using data from treebanks for each language they calculate average dependency lengths for hypothetical optimized and random linearizations and compare them to the actual average dependency length.
Their comparison with Korean is based on data from a similar production experiment by Dennison (
Note that this observation does not contradict the fact that word orders that comply with DLM are likely to become grammaticalized in a given language – for instance, one can argue that the postposition of subordinate clauses in Persian mentioned earlier (see Section 2.2 page 9) presents such a case.
Importantly, for Basque as well, Ros et al. (
We wish to express our gratitude to all participants who volunteered to take part in our experiments, as well as to friends who helped diffusing our call for participants in social networks. We are thankful to four anonymous reviewers whose comments and suggestions greatly improved the paper.
This work is partly supported by a public grant funded by the French National Research Agency (ANR) as part of the “Investissements d’Avenir” program (reference: ANR-10-LABX-0083).
The authors have no competing interests to declare.