Cross-linguistic comparisons on distributive universal quantification: Each vs. every vs. mei

. This paper discusses differences between each and every with regard to (a) pair-list readings; (b) subject/object asymmetries seen with every but not with each ; and (c) the long-held intuition that each is more individualistic whereas every is friendlier to groups. We propose that these phenomena can be captured by prior accounts of the Mandarin Chinese distributive universal quantifier mei . In particular, we consider the double variable hypothesis (the idea that in DUQ, for every x, there must be a y ) (S.-Z. Huang 1995; 1996), and the Skolemized topicality hypothesis (the idea that topical quantifiers are Skolemized, resulting in the required x-y pairings) (S.-Z. Huang 2022b). We argue that (a’) pair-list answers to questions with quantifiers are derivable from the double variable hypothesis; (b’) the subject/object asymmetry seen in every is due to its positionally-varied association with the double variable hypothesis, while each is always subject to Skolemized topicality due to its inherent topicality; and (c’) the individualistic interpretation of each can be described as stemming from its intrinsically Skolemized topicality as well. Results from experimental works will be brought to bear on the theoretical proposals.


Introduction.
Distributive universal quantification (DUQ), exemplified by English each and every, has been studied in a rich body of literature (see citations in Table 1).The Mandarin Chinese (henceforth Chinese) universal quantifier mei also has been investigated in numerous works (e.g., Lee 1986;Liu 1990;S.-Z. Huang 1996;Li 1997;Lin 1998;Cheng 2009;Li 2014).Here, we investigate DUQ through a comparison between English and Chinese.We sketch how some general patterns of each and everysummarized in Table 1 can be viewed afresh through the lens of Chinese, based on two related proposals for mei: the double variable hypothesis (DVH) on DUQ, which states that in DUQ, for every x, there must be a y such that the interpretation of y depends on that of x (S.-Z.Huang 1995;1996); and the Skolemized topicality hypothesis (STH), which states that topical quantifiers are Skolemized, and that Skolemization leads to the required x-y pairings (S.-Z.Huang 2022b).Given the complexity of these proposals and the richness of the cross-linguistic data used to motivate them, we can only offer a first sketch of how the views that were initially developed to account for the behavior of mei can fruitfully be applied to the behavior of DUQ in English.
Section 2 introduces the double variable hypothesis (DVH) on DUQ and its motivation from Chinese.Section 3 supplements the DVH with the Skolemized topicality hypothesis (STH), on the basis of additional Chinese data.Section 4 illustrates how pair-list readings in English DUQ can be seen as the DVH at work.In Section 5, we enhance Krifka (2001)'s intrinsic topicality treatment of each to bring it under the STH and discuss differences between each and every along these lines.The leading idea is that each is intrinsically topical, and thus intrinsically Skolemized.Section 6 reviews results from recent psycholinguistic experiments (vi-viii in Table 1) and their relation to the DVH and the STH.Section 7 reports a pilot judgment study of tenseless English sentences (Joe wash dishes?Not in a million years!), which offers initial empirical support for the applicability of the DVH in English.Which book did you lend to each/every student?Krifka (2001); Beghelli (1997); Surányi (2003); Szabolcsi (2010)

The double variable hypothesis (DVH).
In a nutshell, the DVH states that DUQ implicates a double-variable pairing, namely, that for every x (in the restriction of the quantifier) there must be a y (within the scope of the quantifier), such that the value of y depends on the value of x.The data that initially motivate the DVH come from the Chinese quantifier mei (S.-Z.Huang 1995;1996).As seen in ( 1), mei generally requires an accompanying instance of the adverb dou.
(1) Mei-ge xuesheng *(dou) biye-le.every-CL student DOU graduate-ASP (ASP = aspect, CL = classifier, DOU = dou) 'Every student graduated.'However, as seen in (2-3), mei can occur without dou if there is a variable within the predicate in the form of a singular indefinite (either an argument or an adjunct).1 (2) Mei-ge haizi (dou) mi yi-ge gexing.every-CL child DOU fancy one-CL pop.star 'Every child fancies a pop star.' (3) Mei-ge gexing (dou) hong-le yi nian.every-CL pop-star DOU popular-ASP one year 'Every pop star was popular for a year.'So, dou, while generally required to license mei, becomes optional in the presence of an indefinite (or reflexive; see footnote 2) in its scope.The DVH proposes that DUQ requires two variables in a dependent relationship.The second variable can be introduced, e.g., by an indefinite as in (2-3), but otherwise has to be provided by dou, as in (1).If we take the DVH to be general across languages, including English, the question arises why no additional dou-like element is needed to license the use of each or every, as seen in the English translations of (1-3).
S.-Z.Huang (1995Huang ( , 1996) ) suggests that the English translation of (1) meets the DVH due to tense, which constrains the event variable for quantificational purposes, in the current case, as the variable y for DVH. 2 In Chinese, which has no systematic overt tense marking (Chao 1968;Li & Thompson 1981), the unconstrained event variable is not available for quantification unless the adverb dou "activates it" to participate in the x-y pairing.To formally capture this idea, S.-Z.Huang (2005) treats dou as a sum operator marking the least upper bound of the summation of events which are then mapped into the denotation of the predicate.But setting aside the formal details, the general idea is that the contrast between (1) and (2-3) in Chinese, as well as the contrast between Chinese and English versions of (1), can be accommodated by adopting the two hypotheses in (4) and ( 5).
(4) English (1) satisfies the double variable requirement due to overt tense marking, which constrains the event variable and makes it available for the x-y pairing.
(5) Having no systematic overt tense marking, the event variable in Chinese (1) is not available unless the adverb dou is present to make it so; therefore mei requires dou or a different source of the second variable (i.e., a variable-introducing element in its scope).
One might wonder what happens when both dou and an indefinite object phrase are present with a mei subject.Consider again the sentence in (2), which one might initially expect to have a reading on which every child out-scopes a pop star (for each child, there's some pop star that child likes) and another on which a pop star out-scopes every child (e.g., Taylor Swift is liked by every child).3But in fact, this scopal ambiguity is possible only when dou is present.Without it, (2) only has a narrow scope reading of the indefinite object, as predicted by the DVH: Given the need for an x-y pairing, the DVH requires narrow scope for the indefinite as it is the only source of the second variable y.On the other hand, when dou is present, it makes the event variable available, so the indefinite is not restricted to narrow scope and can have a wide scope vis-à-vis mei.Note that the general point here holds independent of whether one takes the wide scope reading of the indefinite object to involve a specific reading: the fact remains that the "widescope" reading is absent when dou is absent, because the indefinite is needed for contributing variable y for mei under the mechanism of the DVH.In other words, when dou is absent, the indefinite object cannot have a specific reading.4

The Skolemized topicality hypothesis (STH).
Another aspect of mei's behavior is a subject/object asymmetry, which the DVH does not account for.In particular, while the double variable condition holds for mei in (1-3), it does not apply to mei in object position: (6) Zongtong (*dou) jiejian-le mei yi-wei xinren dashi.president DOU receive-ASP every one-CL newly-appointed ambassador 'The President received every new ambassador.' In fact, when mei is used in the object position, dou is ungrammatical.5There are differences in meaning as well: It is generally assumed that object mei has a strong group reading (see, e.g., Zhang & Pan 2019 and references cited therein).To account for the asymmetry, S.-Z.Huang (2022b) introduces the Skolemized topicality hypothesis (STH), which centrally claims that topichood is intrinsically associated with a Skolem function for topical quantifiers. 6By way of illustration, consider what happens when a topical indefinite undergoes Skolemization on standard theories: the Skolem function replaces the variable introduced by the indefinite with a Skolem constant, so that a topicalized indefinite results in a specific reading that meets the referentiality condition of a topic.This accounts for the general pattern that although Chinese disfavors indefinites in the subject/topic position (Zhu 1982), when they do appear in the subject/topic position (often with topic markers), they receive specific readings.7Skolemization of a topical universal quantifier is similarly defined by the Skolem function "looking for" a variable y to replace, effectively "pairing" it with the variable x, creating a dependency between the two.If being in the subject/topic position inherently invites Skolemization of this sort, then it follows that mei only needs to meet the double-variable condition when it appears in that position;8 when mei is not topicalized, it is subject to no such condition.
We hope we have made the central idea of the STH clear: that the subject/topic position in Chinese is intrinsically associated with a Skolem function, which in turn gives rise to subjectmei's double variable requirement.With the DVH and STH in place, we now turn to how the two apply to English DUQ.
4. Pair-list readings and the DVH.Since May (1985), a large body of work has discussed the availability of pair-list answers to questions like (7).9 (7) Which item did every guest bring?a.
Every guest brought food.b.
Amy brought cake, Bill brought wine, … Of particular interest has been the apparent subject/object asymmetry: while ( 7) permits a pairlist response, like (7b), ( 8) seems to resist such an answer: (8) Which guest brought every item?a.
George brought every item.b.
#Amy brought cake, Bill brought wine, … Though also of note is the observation that replacing every with each increases the acceptability of (8b) (and, for that matter, of (7b) too) (see e.g., Beghelli 1997;Surányi 2003;Szabolcsi 2010;Achimova et al. 2016).There are thus two facts to explain: first, that pair-list answers to questions with subject-every are permitted to a greater extent than pair-list answers to questions with object-every; and second, that pair-list answers are, in general, more acceptable given each than given every.
If pair-list answers for questions with quantifiers are viewed as a subcase of x-y pairings induced by the double variable requirement on DUQ, then the first patternthe subject/object asymmetrycan easily be derived from the DVH.10In (7), subject-every has in its scope a variable y in the form of the trace/copy of the wh-phrase that covaries with it, as is required under the DVH.Thus, an x-y pairing obtains for (7), which is informally represented in (9).
(9) For every x, x is a guest, there is a y, y is something x brought, what is y?
With every in the object position in (8), there is no variable y in its scope that can covary with it, and thus there is no x-y pairing.As a consequence, pair-list answers are ruled out as an option.And as the DVH predicts, the subject/object asymmetry holds only if there is no variable in the scope of object-every.If there is a variable open for covariation with every x, then the ability to offer a pair-list answer is viable, even when every is in object position: (10) What did Bella give every guest?a.
Bella gave every guest a party favor.b.
Bella gave Zoe flowers, Abigail wine, Paul a book… But there is more to the story of questions with quantifiers than just this asymmetry.Bumford (2015:1) offers the following characterizations of "distributional universals": (i) matrix interrogatives that contain them accept pair-list answers; (ii) indefinites and disjunctions in their scope may assume arbitrary functional readings; and (iii) they permit sentence-internal interpretations of a wide range of comparative adjectives, like new and different.
As we have seen, (i) has been directly confirmed as derivable from DVH, with subject/object asymmetry accounted for.Would DVH account for Bumford's (ii) and (iii) as well?While there is no space for a full account, the quick answer seems to be "yes".Just like indefinites, disjunctions are also viable sources for the second variable in the scope of the universal quantifier for distributive interpretation (to form an x-y pairing).For one thing, indefinites and disjunctions are treated as the same type in the algebraic model of semantics under Keenan & Faltz (1986): both are interpretable by join.Under the DVH/STH, pair-list answers are derivable from the functional readings, a position similar to Chierchia (1993), but under different mechanisms.The sentence-internal interpretations of comparative adjectives, which are treated as anaphora in Bumford (2015), also fall under the DVH/STH, as anaphoric expressions are bound by their antecedents.With a universal quantifier as the binder, the anaphoric expressions covary with them (S.-Z.Huang 1996Huang , 2022aHuang , 2022b)), ensuring an x-y pairing.11Turning to the each/every contrast, the fact to be explained is the lack of subject/object asymmetry for each with regard to pair-list answers: each allows pair-list readings even in cases where every does not.This contrast does not fall out of the DVH directly.But with a further assumption, outlined in the next section, the behavior of both each and every can be captured by the DVH.

Intrinsic topicality (each) vs. positional topicality (every and mei).
The crucial assumption, initially defended in Krifka (2001), is that each is intrinsically topical.Under the STH, this means each is defined in terms of an x-y pairing through a Skolem function, which guarantees individual interpretation of each member of the restrictor set, regardless of its syntactic position.Put another way, the requirement for an x-y pairing is inherent in the semantic DNA of each; topicalityand the Skolemization that comes with itis part of its lexical meaning.In contrast, every is a distributive universal quantifier, but gets its Skolemized interpretation only when in topic position (as per the STH).This puts every on par with Chinese mei, which also shows a similar subject/object asymmetry (as discussed in Section 3).
As a consequence of its intrinsic Skolemized topicality, each is guaranteed to give rise to pair-list readings, even when it occupies a syntactic position that would otherwise not invite such readings.This idea could also make sense of various other each/every differences noted in Table 1.For instance, consider the apparent fact that each seems to prefer to always take widest scope, out-scoping other operators such as questions, and even scoping out of finite and relative clauses (Kroch 1974;Beghelli & Stowell 1997;Szabolcsi 2010).The reason that each does so, under the DVH/STH, is that each needs a variable y in its scope, so has no choice but to scope high.This explanation receives indirect support from Chinese, where topics have also been argued to be able to escape relative clauses (C.-T.J. Huang 1984).
By the same token, consider Landman (2003)'s contrast between *In this talk I combine each theory of plurality and In this talk I combine every theory of plurality.When in object position, the restrictor of every seems to be "summed up" into a group, allowing it to combine with verbs like combine, a fate to which each is seemingly immune.Under the DVH/STH, the behavior of each in contrast to every is due to each's intrinsic association with Skolemized topicality, something every only acquires in preverbal positions and only under certain conditions.
We adopt this intrinsic topicality view from Krifka (2001), with one modification about how to meet the referentiality requirement of topichood.For Krifka, distributive universal quantifiers meet the referentiality requirement on topics because "the predication is about the elements in the restrictor set of the quantifier, and the universal determiner expresses that each element is subjected to the predication expressed in the comment.Hence the determiner denotes a relation between the topic set and the predication expressed by the comment" (Krifka 2001:26).This would work if such a quantifier is each since its ability to denote a relation between the topic set and the predication doesn't vary with its position in a sentence (and we know that each tends to take the widest scope so its intrinsic topicality is matched with its structural position).But this way of placing the "referentiating" power on the quantifier determiner doesn't work with every.We have seen that every N can gain topical readings in preverbal positions, but not so in postverbal positions.This variability of topicality for every is the reason we have treated each with intrinsic Skolemized topicality and every with positional Skolemized topicality.But if not through the quantificational determiner, how do topical universal quantifier phrases headed by each and every meet the referentiality condition?S.-Z.Huang (2022b) places the "referentiating" power of creating topic worthy quantifier phrases (both existential and universal) on the Skolem function that a topic position is intrinsically equipped with.Skolemizing a universal quantifier means "individuating" each member of the restrictor set of the quantifier, as the Skolem function goes through this set one member at a time "looking for" a second variable to match with it.While each N obtains an individualistic reading fulfilling the referentiality condition unfailingly because of its intrinsic topichood, every N meets the referentiality only if it is in the right position (a preverbal position).

Individualistic vs. group-friendly representations.
A long-held intuition, originating with Vendler (1962), is that each is in some sense 'more individualistic' whereas every is somehow 'friendlier to groups'.A recent body of psycholinguistic work demonstrates a particular ramification of this intuition: processing sentences with each causes speakers to deploy a cognitive system for individuation, whereas processing matched sentences with every causes them to instead deploy a cognitive system for grouping individuals.We briefly summarize these results before asking how they might be captured under the DVH/STH.Knowlton et al. (2022) presented participants with scenes consisting of circles of various colors and sizes and asked them to evaluate simple sentences like each big circle is blue and every big circle is blue.After answering these questions, participants were then asked to recall how many circles there were, a property that holds of the group as a whole and not of any one individual circle.Participants who initially evaluated sentences with every were better able to recall the circles' cardinality than those who saw the very same images but initially evaluated sentences with each. 12Knowlton (2021) likewise asked children between 5 and 8 years of age to complete a similar task probing the group of circles' center of mass, another fundamentally group-based property.Again, those who initially evaluated a sentence with every showed superior memory for the group's center.Knowlton et al. (2023b) test the opposite prediction: evaluating an each sentence should make participants better at recalling individual properties than an every sentence.Indeed, they report that participants who evaluated sentences like each circle is green were better able to succeed at a subsequent change detection task in which one circle's hue changed.Ongchoco, Knowlton & Papafragou (2023) showed that these results also extend to auditory objects (by asking what participants recall after evaluating sentences like each/every sound is pleasant with respect to sequences of tones).
The effects of each biasing activation of the cognitive system for individuation and every biasing activation of the cognitive system for group-formation are not limited to sentence verification tasks.Knowlton, Trueswell & Papafragou (2023) show that participants prefer to use each to quantify over small domains, and prefer to use every to quantify over large domains in contexts where all else is held equal.And when asked to imagine the size of the domain, participants given a sentence in which DUQ is indicated with each reliably offer estimates of two or three items; participants given a matched sentence using every more often offer estimates of far higher cardinalities.Knowlton & Gomes (2022) show that this same pattern emerges in naturalistic speech to children: parents use each to quantify over small numbers of physically present things far more often than they do so with every or all.This set size difference is a wellknown signature of the difference between using the cognitive systems for individuation and for grouping: the system for individuation builds high-fidelity representations, and is thus subject to stricter working memory constrains.
Broadly speaking, then, the fact to account for is that each triggers activation of the mind's system for individuation whereas every triggers activation of the mind's system for group formation.We might suppose that under the STH, this contrast could be captured by proposing a link between each's intrinsic topicalityand hence Skolemizationand the cognitive system for individuation.However, there is a wrinkle.If preverbal every is always topical, then Skolemization applies to it, as it does to each intrinsically.Under the STH, both quantifiers would receive individualistic readings in such cases, not just each.In the contexts tested, both quantifiers were in subject position.If this further implies that both quantificational phrases were topical, then both should have given rise to the same performance.
A more nuanced differentiation between each and every thus has to be made to capture these experimental results under the DVH/STH.This can be done if we recognize a difference between topic and subject.While every is in subject position in the experiments cited above, it might not be topical.And if every is in the subject position but non-topical, this allows for the STH to account for why subject-every invites speakers to turn to the mind's system for group representation: Skolemization does not kick in to call for individuation.With each, being intrinsically topical, such individuation is unavoidable regardless of syntactic position.Of course, making this case would require independent evidence that every is indeed not topical in the cases tested.
One potential avenue for future work along these lines relates to the idea that every's topicalityor lack thereofmight be able to be manipulated based on whether Stage Level or Individual Level predicates are used.In the experiments cited above, Individual Level predicates were used (e.g., be green).Such predicates might not be conducive for coercing topical every, thus allowing subject-every to remain un-topicalized (and, as such, un-Skolemized) and retain its group-friendly interpretation.After all, Individual Level predicates have been argued to carry a generic operator (Chierchia 1995), and generic statements could be thought to be quintessential non-topic/comment structures (S.-Z.Huang 2024).So perhaps using Stage Level predicates would invite topicalizing every and make the observed each/every contrast disappear.We leave empirical testing of this suggestion for future work.
If such a prediction were borne out, it might also signal a new way of accounting for the fact that every is sometimes able to give rise to generic interpretations in situations where each cannot.For example, as Knowlton, Trueswell & Papafragou (2023) discuss, each martini needs an olive calls to mind (a small number of) particular cocktails in need of garnishes.But every martini needs an olive instead calls to mind something that one might encounter as part of a recipe: a claim about martinis in general.Knowlton & Lidz (2021) argue that a difference in usage along these lines is quite common in child directed speech: parents seem to use every mostly to make generic generalizations (e.g., every time we go to the store, you cry), which they seldom do for each.Both papers suggest that this apparently "generic" behavior of every stems from the fact that it triggers the cognitive system for group representation, which itself is friendlier to generalization than the system for individuation.An alternative view is that every sometimes receives generic readings because it scopes under the generic operator (Beghelli & Stowell 1997), whereas each, being intrinsically topical, always scopes above it.

Tenseless sentences and the applicability of DVH in English.
Having offered a sketch of how the English contrasts in Table 1 can be accounted for under the DVH/STH, we now return to the contrast between English and Chinese DUQ.Recall that in (1), repeated here as ( 11), the Chinese version requires dou, but the English translation requires no such additional element.
every-CL student DOU graduate-ASP 'Every student graduated.' Section 2 suggested that, in the English version, tense licenses the event variable, which in turn satisfies the double variable requirement of the DVH.In contrast, Chinese does not have the event variable at the ready, since the language does not have systematic tense marking, so dou (or some other source of the necessary second variable) is required.This account gives rise to a clear prediction: if an English DUQ sentence lacked overt tense marking, it would be unacceptable, just like Chinese, unless some other source of the second variable for DUQ is available.Though infrequent, English does have a sentence structure that fits the bill: so-called "Mad Magazine" (MM) sentences, 13 such as Joe wash dishes?!Not in a million years!Though this is a marked sentence type, which requires specific contexts (often including some kind of explicit denial like No way! or Yeah right!), it is perhaps the closest we can get to tenseless sentences in English.DUQ is predicted to interact with MM sentences in the following way, first outlined in S.-Z.Huang (2005): MM sentences with subject-every (e.g., Every doctor cure patients?Yeah right!) fail to meet DVH because the event variable is not licensed as the needed variable y.Such sentences should thus be ungrammatical unless another source of variable y (such as an indefinite) is present (e.g., Every doctor cure a patient?Yeah right!).Furthermore, temporal adverbs should improve acceptability, since, like dou, they can be expected to make the event variable available (e.g., Every doctor cure patients on time?Yeah right!).On the other hand, 13  Akmajian (1984) is credited to be the first that analyses Mad Magazine Sentences, so-called because such sentences were popular in a magazine called Mad.Muffy Siegel brought the MM sentence type to the attention of the first author, and Amanda Payne provided the reference.Their contributions are gratefully acknowledged.
additional non-temporal modification should be of little help (e.g., Every doctor cure patients thoroughly?Yeah right!).
To test these predictions, we asked 72 English-speaking participants (recruited on Prolific) to judge the acceptability of eight MM sentences on a 5-point scale.The MM sentences were presented as the final line of a short story that provided pragmatic support for the use of these marked constructions (see Figure 1).The untensed and unmodified (12a) is predicted to be less acceptable than the tensed (12d), as the latter meets the double variable requirement by virtue of overtly introducing an event variable with tense marking.Adding a non-temporal modifier to the untensed (12a), as in (12b), should do little to rescue the sentence and is thus also predicted to yield low acceptability, in line with (12a).In contrast, the addition of a temporal modifier, as in (12c), should "rescue" the tenseless sentence, thanks to the temporal adverb picking up on the predicate's event variable.In which case, (12c) should have higher acceptability, on par with (12d).These predictions were borne out (see Figure 2).Untensed sentences with a DUQ, like (6a), were judged as degraded by participants (2.6/5 on average).But such sentences could be rescued with the addition of a temporal adverbial, as in (6c), which were judged as more natural (3.2/5), on par with tensed sentences like (6d) (also 3.2/5).The addition of a non-temporal modifier, as in (6b), did not likewise rescue the tenseless sentences (2.8/5).We observe significant effects of tense (β = -.56 [-.84 to -.28], p < .05)and temporal modification (β = .58[.32 to .83],p < .05)but not non-temporal modification (p = .5).With another batch of participants, we elicited acceptability judgements for nonquantificational versions of (12a-d), which used proper nouns instead of every-phrases.These control sentences were predicted to yield higher acceptability ratings than the DUQ versions in (12), since proper nouns are not subject to the double variable requirement and should thus not be affected by lack of tense to the same degree as DUQ.The NP-version of (12a), for example, was rated as numerically more acceptable than the DUQ version, though this comparison was only significant (p=.054).Perhaps more importantly, we should expect that the addition of temporal or non-temporal modifications will have no additional effects on acceptability in the NP versions.This prediction was borne out: we observe no significant effects of tense (β = -.24[-.05 to -.44], p = .21),temporal modification (β = .04[-.19 to .27],p = .85),or non-temporal modification (β = -.14 [-.30 to .02],p = .40).
This overall pattern of results suggests that the presence of an event variableeither from overt tense marking or from the added temporal modificationis important for licensing DUQ in English.Untensed sentences with DUQ are rated as unacceptablepresumably because DUQ is not licensedand merely adding additional modification does nothing to make them more acceptable.These results thus lend initial support to the idea that the DVH is applicable in English.Future work in this line will test each, to see if it patterns similarly to every in this case, as well as other sources of the additional variable (e.g., indefinites and bound variables).
It might be worth noting that all the conditions used in the above study used Stage Level predicates, as Individual Level predicates seem even more degraded in an MM context (Every circle be green?Fat chance!).So the strongest empirical evidence for x-y pairings for topical every comes from the environment when the predicate is a Stage Level one.This lends credence to our supposition in the previous section that the Stage vs. Individual Level difference might impact whether every behaves more like each or more like mei.Again, putting this claim to the test is a target of future work.

Figure 1 .
Figure 1.Example story.Participants judged the bolded objection on a 5-point naturalness scale.

Figure 2 .
Figure 2. Acceptability rating by predicate type (untensed without a modifier; untensed with a non-temporal modifier; untensed with a temporal modifier; tensed without modification).Large solid points represent group means; small translucent points represent participant means.

Table 1
. Properties of each and every observed by various authors