Abstract
The telicity behavior of degree achievements has been a puzzling problem to many linguists. The most successful and currently standard theory (Kennedy & Levin 2008) treats them as degree expressions lexicalizing different types of scales, which in turn influence the resulting evaluative or non-evaluative interpretation. While it may account for English, this theory does not hold up cross-linguistically. We challenge the scalar theory with new Slavic data and show that verbal prefixes influence the (non-)evaluative interpretation of degree achievements more than their underlying scales do. This proposal is formalised as an addition of two type shifters, morphosyntactically realised as prefixes, which, in result, have an evaluative/non-evaluative effect on the given degree achievement.
1 Introduction
In our paper, we discuss the interaction of degree achievements with the derivational morphology of Czech (as a representative example of Slavic languages). Degree achievements are verbs generally based on adjectives, such as the English widen based on wide. Degree achievements denote scalar change: if it is true that the river widened, then the river underwent a positive scalar change on the scale of width. One of the most prominent problems in the degree achievement literature has been the issue of their telicity behavior (see at least Dowty 1979; Abusch 1986; Hay et al. 1999; Kearns 2007; Kennedy & Levin 2008).
We will draw our data from Czech, and we will focus on the compositional contribution of Czech prefixes on the telicity of degree achievements. By way of example, consider (1) with the degree achievement cool with two possible interpretations as witnessed by the standard test of adverbial modification. The adverbial in an hour emphasizes the telic interpretation (in the end, the tea is at a contextually given degree of coldness). The adverbial for an hour highlights the atelic interpretation where the tea just decreases on the scale of temperature, but it does not reach any goal of being cold.
| a. | The tea cooled in one hour. | telic/evaluative |
| b. | The tea cooled for one hour. | atelic/non-evaluative |
In English (1) the standard adverbial modification test directly corresponds to an inference (or the lack of it) of such sentences to sentences where the degree achievement would be substituted with its base adjective in a positive form. Thus, from (1a) we can infer that the sentence The tea was cool is true but from (1b) such inference would be non-valid. We call such inference ‘evaluative’ and, following Brasoveanu & Rett's (2018) usage of this term, we call degree achievements evaluative if their corresponding adjectives instantiate a degree above a particular standard. The standard notion of telicity overlaps in many cases like English (1) with evaluativity, but once we move to Slavic degree achievements the situation gets more complex, which is the reason why we stick to the evaluative/non-evaluative distinction further. 1
The two interpretations are unambiguously marked by the different Czech prefixes in (2a) and (2b). But the situation is more complex than in English due to the interaction of the degree semantics with the aspectual system of Czech. 2 The verb vy-chladl ‘from-cooled’ in (2a) has the unambiguous evaluative interpretation and the verb o-chladl ‘around-cooled’ in (2b) signals the non-evaluative interpretation. The evaluation inference can be verified by the tests that we will introduce in Section 4.1.
| Čaj | vy-chladl | za | hodinu. | evaluative |
| tea | from-cooled.3sg | in | hour | |
| ‘The tea cooled in an hour.’ | ||||
| Čaj | o-chladl | za | hodinu. | non-evaluative |
| tea | around-cooled.3sg | in | hour | |
| ‘The tea cooled slightly in an hour.’ | ||||
At this point we want to draw attention to the fact that both Czech sentences in (2) are telic according to the standard adverbial test. Note that in both (2a) and (2b), the predicates are perfective verbs (under the standard tests for grammatical aspect) and, following the standard approach to the relationship between grammatical and lexical aspect in Slavic languages (exemplified by Brecht 1985 a.o.), we believe that the standard adverbial test in (2) detects the telicity, which stems from the perfective grammatical aspect. 3 But despite the telicity of both (2a) and (2b), native speakers of Czech would infer that a Czech sentence corresponding to English The tea was cool would follow only from (2a), not (2b). Notice also that the telicity of (2b) is of a different kind than the telicity in English (1a), Czech (2b) only expresses that a slight degree change occurred, not that the contextual standard was reached. And since all the current standard approaches to degree achievements (like Kennedy & Levin 2008; Kennedy 2012; Piñón 2005) build on the degree analysis of degree achievements emphasizing the link between degree achievements, their base adjectives, and their particular standards, we use the evaluativity terminology and utilize the appropriate tests. In sum: for Czech degree achievements, the decisive factor, at least in examples like (2), seems to be the nature of the prefix, which predicts the evaluativity/non-evaluativity of the degree achievement. And even if the Czech perfective degree achievements are telic according to the standard adverbial test, this telicity is in a sense trivial because it only signals the inclusion of the event time in the reference time, in which the increase was either non-evaluative (as in (2b)) or evaluative (as in (2a)). Therefore we utilize the evaluativity dichotomy further, since it is the appropriate tool for teasing apart those Czech degree achievements which in English fall under the rubric of telic degree achievements from the non-evaluative Czech perfective degree achievements.
This paper is organized as follows. In Section 2, we begin with the inference behavior of degree achievements and their different classes. We show that many distinctions of adjectival meaning are maintained in the distinctive patterns of absolute and relative degree achievements and then further zoom in on absolute degree achievements. In Section 3, we introduce the scalar framework of Kennedy & Levin (2008). Section 4 then shows the application of the framework to the Czech data: we first begin with the standard evaluativity tests (Section 4.1), which we apply to sub-classes of degree achievements, and discuss the wrong predictions of Kennedy & Levin (2008) with respect to the Czech data. Section 4.2 presents our solution to the problem couched in the extended version of Kennedy and Levin's framework. Finally, Section 5 summarizes the article.
2 Degree achievements and evaluativity
The majority of degree achievements is based on adjectives, and many facets of adjectival classification have their clear counterparts in the different sub-types of degree achievements. Let us start with one such difference, the distinction between relative adjectives (those whose positive form is interpreted with respect to a contextually provided standard, e.g., wide, deep, old) and absolute adjectives (adjectives with a lexically fixed standard, e.g., dry, full, empty). The distinction carries over to degree achievements and gives rise to different entailment behavior of absolute and relative degree achievements. While absolute degree achievements, as in (3), allow inferences to a positive form of their corresponding adjective (called evaluativity), the relative degree achievements, as in (4), do not.
| a. | The hall emptied. |
absolute |
| b. | The hall is empty. |
| a. | The river widened. |
relative |
| b. | The river is wide. |
Furthermore, it is not only the distinction between the absolute and relative degree achievements which gives rise to different inferential behavior of degree achievements. 4 The more nuanced nature of the scale corresponding to the degree achievements also plays an important role. The absolute adjectives are usually divided into three sub-classes (see Rotstein & Winter 2004 a.o.): (i) lower-bounded adjectives, like wet or dirty; (ii) upper-bounded adjectives, like dry or clean; (iii) closed-scale adjectives, like opaque or transparent. The standard test used to classify the adjectives into the sub-classes is their distribution with different types of modifiers. The lower-bounded adjectives can be modified by slightly-type of modifier but not by almost: #almost wet/slightly wet. The upper-bounded adjectives act exactly the other way round: almost dry/#slightly dry. The totally closed adjectives allow modification by both: almost/slightly transparent, whereas the relative adjectives cannot be modified by either: #almost/#slightly tall. The distinction between the four sub-classes of gradable adjectives is then explained as the difference in the structure of the scale denoted by the particular adjective (see Kennedy & McNally 2005). All the three sub-classes of absolute adjectives are described as lexicalizing the lower, upper, or both bounds on their scales. The relative adjectives do not do that and hence must use contextual standards. Since the degree achievements and their source adjectives seem to share their scalar semantics, the distinctions discussed above carry over to degree achievements.
The relevant facts about English degree achievements are at least the following (see Kennedy & Levin 2008 for a detailed discussion): (i) upper-bounded degree achievements strongly prefer evaluative interpretations, see (3); (ii) open-scale (i.e., relative) degree achievements are by default interpreted as non-evaluative (with only two known exceptions in English: cool and hot), see (4); 5 (iii) lower-bounded degree achievements prefer non-evaluative interpretation; (iv) totally closed-scale degree achievements are reported to behave as upper-bounded (see Kennedy & Levin 2008). While we will dispute the applicability of these predictions to non-English data below, we agree that they show the importance of the scalar properties for the proper treatment of degree achievements and their evaluativity behavior.
Another important fact, discussed by Pedersen (2015), concerns the different types of ambiguities triggered by again with respect to relative and absolute degree achievements. For English, the absolute adjectives combined with again yield restitutive/repetitive reading demonstrated in (5). The repetitive/restitutive ambiguities of again have been discussed in the literature for a long time (see at least von Stechow 1995, 1996; Beck 2005), but here we will focus only on a small part of the again ambiguities – those that arise in the context of degree achievements. The repetitive reading of the degree achievement dried in (5) means that the whole event of drying had to happen before; one such context is exemplified in (5a). The restitutive meaning, (5b), does not require the repetition of the whole drying event; it merely demands the restitution of the (previously held) result state – the shirt being dry. And as (5) shows, both readings are available when used with absolute degree achievements like dry.
| The shirt dried again. | ||
| a. | I washed the shirt, and it dried. I washed the shirt for the second time, and it dried again. | repetitive |
| b. | The shirt got wet, but soon after, it dried again. | restitutive |
The relative degree achievements differ from the absolute ones: for relative adjectives, we can get a so-called successive increase reading with again demonstrated in (6), which is not an accessible reading for absolute adjectives, see (7). The reason is, of course, the obligatory evaluative inference of the absolute degree achievement dry, which is incompatible with the gradual increase scenarios.
| At 13:00, the shadow of the tree lengthened and reached the table. At 16:00, it lengthened again and touched the house. | successive |
| I washed the shirt and let it hang in the sun for one hour. It dried somewhat but was still wet. Then I put the shirt on the radiator, and it dried #again. |
3 Scalar approaches
The inference patterns of degree achievements, particularly their evaluativity behavior (and its dependence on the type of the scale the degree achievements associate with) introduced above, lead many researchers to adopt the scalar approach to degree achievements (see, among others, Kennedy 1997; Hay et al. 1999; Svenonius & Kennedy 2006; Kennedy & Levin 2008; Sawada & Grano 2011; Grano & Kennedy 2012). The starting point of such analyses is the interpretation of adjectives as measure functions of the type 〈e, d〉 – for example, the adjective long is represented with a measure function long returning the degree of an object on the particular scale along the particular dimension (i.e., length). Therefore, the analysis takes the measure function as the core of the meaning. It is then type shifted to a property of objects with – in most cases – morphologically null element pos (first introduced by Kennedy 1997). The type shifter pos supplies the contextual standard needed for the interpretation of relative adjectives.
The degree achievements are then modeled as a measure of change function, which is built on top of the ‘regular’ measure function and returns the degree of change on the appropriate scale that the particular object underwent during the event. As an example, the core meaning of the degree achievement lengthen is then claimed to be a difference function long ↑, which returns the difference between the degree at the initial phase of an event e and the degree in the final phase of e. The verbal measure of change function, (8), is then notated as long Δ, as a mnemonics for the difference core of its meaning. We follow Henderson (2013) in extending Kennedy & Levin's (2008) notation, which allows the verbal measure function to access its arguments via theta-roles.
| Measure of change |
| For any measure function m, |
The same type of difference function is used for modeling the meaning of the comparative form of adjectives. In that case, the difference function returns the degree of difference between the correlate and the standard, the scale used is the difference scale, e.g., long ↑. Typological support for a similar treatment of comparatives and degree achievements can be found in the syncretism patterns: in the majority of cases where the adjective is suppletive, the syncretism holds between the comparative and the degree achievement (like English worse A and worsen V ) but not between the positive form of the adjective and the degree achievement (bad A and worsen V ; see Bobaljik 2015). As an example, the degree achievement widen would denote such a degree by which a bearer of the theta-role changed between the initial and final phases of an event (along the dimension of width). The difference scale wide ↑ would be the common meaning core of the degree achievement widen and a comparative form of the adjective wider than.
|
|
Still, (9) is not a set of entities but a measure function like in the case of adjectives before the type shift with pos null morpheme. The meaning of degree achievements (and their contribution to sentence meaning) is the result of pos being applied to the measure of change function (analogically to the application of pos to the 〈e, d〉 regular measure function in case of adjectives). The result for open-scale and upper-bounded English degree achievements is exemplified below in (10) and (11), respectively. Since in both cases pos is sensitive to the Interpretive Economy principle, (12), the stnd in both formulas picks up the minimum and maximum degree of the corresponding scale, respectively. The idea behind the Interpretive Economy principle is the following: as the lexical meaning of adjectives is based on scales and different types of scales come with natural transition points (the maximum in case of upper-bounded adjectives like dry, both minimum and maximum in case of totally closed adjectives like full), the pos morpheme, which shifts the measure functions into a set of entities, is sine qua non predetermined to use the transition points and not, e.g., contextually supplied standards.
| The shadow of the tree lengthened. |
| ∃e[long
|
| The shirt dried. |
| ∃e[dry
|
| Interpretive Economy (Kennedy & Levin 2008, ex. 18)Maximize the contribution of the conventional meanings of the elements of a sentence to the computation of its truth conditions. |
The approach of Kennedy & Levin (2008) directly predicts some correlations of scalar structure and evaluativity behavior of degree achievements. Or the other way round, it is designed in a way that predicts the patterns found among English degree achievements. We summarize the predictions below, but for the full description and details, see Kennedy & Levin (2008).
-
Open scales: only non-evaluative reading is predicted for open scale degree achievements (with the only exception: ‘conventionalized’ cool). This follows from the nature of the difference scale: since it is lower-bounded, at the start of the event, it begins at the minimal degree, and any increase makes a sentence like (10) true. Since any increase verifies open-scale degree achievements, they behave as non-evaluative.
-
Upper-bounded scales: since upper-bounded degree achievements like dry in (11) are based on a measure function with a maximum, only evaluative reading is derived – no smaller increase than the maximum increase would make sentences like (11) true.
-
Lower-bounded scales: they are predicted to behave like the open-scale degree achievements and for such (English verbs like wet or impure), only the non-evaluative reading is expected.
-
Totally closed scales: here, the predictions of the scalar framework are not totally clear – in Kennedy & Levin (2008, fn. 11), it is claimed that some total adjectives can use both minimum and maximum standards (e.g., opaque); others should prefer only the maximum standard (e.g., full). However, when later discussing the degree achievement straight, they claim that, since the maximum standard is logically stronger than the minimum standard, the evaluative interpretation should be at least the more preferred one. In sum, it seems that the pragmatical stance predicts at least a strong preference for the evaluative interpretation, similar to the upper-bounded degree achievements.
4 Czech degree achievements
In this section, we will show that counterexamples to all predictions of Kennedy & Levin (2008) can be found. In a nutshell, we can find natural examples of Czech degree achievements with evaluative or non-evaluative interpretation for all types of scales. Consequently, Czech does not seem to show the behavior of English, where the Interpretive Economy principle dictates the interpretation of a degree achievement based on the scalar type it associates with. We will go through all types of scales and show pertinent examples of our claim.
4.1 Tests and scales
Within all types of scales, we tested the behavior of degree achievements with the two following tests: (i) the inference pattern of the degree achievement and its corresponding adjective; (ii) the presupposition behavior of the degree achievement with respect to Czech znovu ‘again’.
-
The inference/contradiction test: if the sentences containing the degree achievement allow a continuation with a base adjective of the opposite polarity without a contradiction, the degree achievement is interpreted as non-evaluative (it is interpreted without evaluativity inferences). On the other hand, evaluative degree achievements lead to a contradiction.
-
The presupposition test: if the degree achievement modified with again results in a successive increase reading, it is interpreted as non-evaluative. On the other hand, evaluative degree achievements give rise to restitutive or reversal interpretations.
Let us now show the application of both tests on the open-scale Czech degree achievements:
| Pokoj | se | o-hřál. | non-evaluative |
| room | refl | around-warmed.3sg | |
| ‘The room warmed.’ | |||
| …ale | pořád | byl | studený. | no contradiction |
| …but | still | was | cold | |
| ‘…but it was still cold.’ | ||||
| …a | pak | se | o-hřál | znovu. | successive |
| …and | then | refl | around-warmed.3sg | again | increase |
| ‘…and then it warmed again.’ | |||||
| Pokoj | se | vy-hřál. | EVALUATIVE |
| room | refl | from-warmed.3sg | |
| ‘The room warmed.’ | |||
| #…ale | pořád | byl | studený. | contradiction |
| …but | still | was | cold | |
| ‘…but it was still cold.’ | ||||
| …a | pak | se | vy-hřál | znovu. | repetitive/ |
| …and | then | refl | from-warmed.3sg | again | reversal |
| ‘…and then it warmed again.’ | |||||
The examples in (13) and (14) represent a minimal pair differing only in the verbal prefix. While o- ‘around’ in (13) prompts the non-evaluative interpretation of the degree achievement, vy- ‘from’ in (14) gives rise to an evaluative one. 6 As discussed already in Section 1, Czech perfective degree achievements discussed in this article are telic according to the standard adverbial modification test but show different evaluativity/non-evaluativity profiles according to the appropriate tests. Furthermore, notice that we translated the Czech prefixes as English prepositions; this is because we assume that Slavic prefixes and prepositions are syntactically and semantically identical (see Matushansky 2002; Asbury et al. 2006 a.o. for the empirically well-supported claims of their identity). We follow Zwarts (2005) in formalizing prepositions as sets of abstract paths (or vectors) that can be semantically divided into locative (e.g., at, in) and directional; the latter ones split further into source (e.g., from, off), route (e.g., through, around), and goal (e.g., to, into) prepositions. The key property of the prepositions/prefixes for our account is their algebraic structure, namely the possibility to concatenate their trajectories (in other words, their cumulativity). Take oběhl ‘around-ran’ from (15) as an example; o- ‘around’ denotes a trajectory around the house, i.e., a set of vectors that begin and end in the general spherical area around the house. If two such vectors are concatenated, the resulting trajectory necessarily begins and ends in the same general area around the house, thus also qualifying for a denotation by o- ‘around’. Following Zwarts (2005), we classify these vectors as atelic. On the other hand, the source trajectory denoted by vyběhl ‘from-ran’ in (16) requires the set of vectors to start in the house and end outside. However, two such vectors cannot be concatenated because the endpoint of the first vector (outside the house) must be the starting point of the second one (in the house), which is unattainable. Such vectors are telic. Now, with respect to the discussion at hand, it is important to note that atelic prefixes usually lead to the non-evaluative interpretation of degree achievements and, vice versa, telic prefixes lead to the evaluative interpretation.
| Petr | o-běhl | dům. |
| Petr | around-ran.3sg | house.acc |
| ‘Peter ran around the house.’ | ||
| Petr | vy-běhl | z | domu. |
| Petr | from-ran.3sg | from | house.gen |
| ‘Peter ran out of the house.’ | |||
Returning to the first out of the four types of degree achievements, Kennedy & Levin (2008) predict only the non-evaluative interpretation for the open-scale degree achievements, which is clearly a wrong prediction for the Czech data: both tests agree on classifying the degree achievement o-hřát ‘around-warm’ and vy-hřát ‘from-warm’ as non-evaluative and evaluative, respectively. Moreover, the (non-)evaluative interpretation of Czech degree achievements seems to be derivable from the algebraic properties of their prefixes. We will carry more empirical material for the support of this claim below.
As for upper-bounded degree achievements, where Kennedy & Levin (2008) predict only the evaluative interpretation, we see that Czech degree achievements can be either evaluative or non-evaluative again, depending on the prefix they combine with. With the same two prefixes as discussed above, the upper-bounded adjective suchý ‘dry’ can be interpreted either as non-evaluative in (17) or evaluative in (18).
| Louže | o-schly. | non-evaluative |
| puddles | around-dried.3pl | |
| ‘The puddles dried.’ | ||
| …ale | pořád | byly | mokré. | no contradiction |
| …but | still | were | wet | |
| ‘…but they were still wet.’ | ||||
| …a | pak | o-schly | znovu. | successive increase |
| …and | then | around-dried.3pl | again | |
| ‘…and then they dried again.’ | ||||
| Louže | vy-schly. | evaluative |
| puddles | from-dried.3pl | |
| ‘The puddles dried.’ | ||
| #…ale | pořád | byly | mokré. | contradiction |
| …but | still | were | wet | |
| ‘…but they were still wet.’ | ||||
| …a | pak | vy-schly | znovu. | repetitive/reversal |
| …and | then | from-dried.3pl | again | |
| ‘…and then they dried again.’ | ||||
For the lower-bounded degree achievements, Kennedy & Levin (2008) predict only non-evaluative reading, and exactly as in the previous two cases Czech offers both evaluative and non-evaluative ones, depending on the prefix. In this case, the prefixes used here are different: the non-evaluative interpretation in (19) results from the prefixation by the locative prefix na- ‘on’. 7 The evaluative interpretation in (20) is signaled with the route denoting prefix pro- ‘through’.
| Košile | na-vlhla. | non-evaluative |
| shirt | on-got.wet.3sg | |
| ‘The shirt got wet.’ | ||
| …ale | pořád | byla | většinou | suchá. | no contradiction |
| …but | still | was | mostly | dry | |
| ‘…but it was still mostly dry.’ | |||||
| …a | pak | na-vlhla | znovu. | successive increase |
| …and | then | on-got.wet.3sg | again | |
| ‘…and then it got wet again.’ | ||||
| Košile | pro-vlhla. | evaluative |
| shirt | through-got.wet.3sg | |
| ‘The shirt got wet.’ | ||
| #…ale | pořád | byla | většinou | suchá. | contradiction |
| …but | still | was | mostly | dry | |
| ‘…but it was still mostly dry.’ | |||||
| …a | pak | pro-vlhla | znovu. | repetitive/ |
| …and | then | through-got.wet.3sg | again | reversal |
| ‘…and then it got wet again.’ | ||||
Finally, according to Kennedy & Levin (2008), the totally closed degree achievements are either interpreted only as evaluative, or the evaluative reading should be strongly preferred. Again, in Czech, both evaluative and non-evaluative interpretations of these degree achievements can be found. In this particular case, the evaluative interpretation is a product of the above-discussed source prefix vy- ‘from’, see (22). And the non-evaluative interpretation in (21) is a contribution of the locative prefix u- corresponding to the English at.
| Sál | se | u-prázdnil. | non-evaluative |
| hall | refl | at-emptied.3sg | |
| ‘The hall emptied.’ | |||
| …ale | pořád | byl | plný. | no contradiction |
| …but | still | was | full | |
| ‘…but it was still full.’ | ||||
| …a | pak | se | u-prázdnil | znovu. | successive |
| …and | then | refl | at-emptied.3sg | again | increase |
| ‘…and then it emptied again.’ | |||||
| Sál | se | vy-prázdnil. | evaluative |
| hall | refl | from-emptied.3sg | |
| ‘The hall emptied.’ | |||
| #…ale | pořád | byl | plný. | contradiction |
| …but | still | was | full | |
| ‘…but it was still full.’ | ||||
| …a | pak | se | vy-prázdnil | znovu. | repetitive/ | |||||
| …and | then | refl | from-emptied.3sg | again | reversal | |||||
| ‘…and then it emptied again.’ | ||||||||||
4.2 Tentative solution
The summary of the empirical findings discussed in the last section is the following: Czech degree achievements can be evaluative or non-evaluative, depending on the type of the prefix they combine with. Any type of scale can be changed by the prefix; open scales can be made into upper-bounded, and upper-bounded scales can be changed into open scales via the prefixes with evaluative/non-evaluative effect. In contrast, English seems to be relying on the lexical semantics (and the scalar properties) much more, while in Czech, the evaluativity behavior of degree achievements is determined in morphosyntax, namely via different types of prefixes.
We follow Kennedy & Levin (2008) in their claims concerning English: in English, the lexical/pragmatic information is the main factor in the interpretation of degree achievements (as reflected via stnd(g) in pos v and in the Interpretive Economy principle). But at least in Czech and Slovak (and arguably other Slavic languages that commonly make use of verbal prefixes), the degree achievements behave differently, and they seem to rely much less on the Interpretive Economy principle. Technically, we cash out this English/Czech difference as follows: for Czech, we propose semantically much-loaded type shifters, which are morphosyntactically realized as prefixes. Based on the evaluativity patterns of degree achievements and the independent algebraic properties of the prefixes, we claim that there are two additional difference-function type shifters.
First, for telic prefixes, we propose the type-shifting evaluative semantics in (23). Notice that, unlike in pos type shifter, there is no space in (23) for stnd. Consequently, the Interpretive Economy principle, which relates stnd to the different scalar transition points, is neglected. 8
|
|
Second, for atelic prefixes, we propose the non-evaluative semantics in (24). Again, the type shifter in (24) has the scale minimum hard-wired into the definition. Consequently, no lexical or pragmatic manipulation via the Interpretive Economy principle is possible.
|
|
We acknowledge that our treatment of evaluative and non-evaluative type shifters is a continuation of Kennedy & Levin's (2008) treatment of degree modifiers like completely or partially, which seem to have a similar effect on English degree achievements as the Czech prefixes have on the Czech degree achievements. The crucial difference between English degree modifiers and Czech prefixes, though, is that while English degree modifiers are totally optional, Czech prefixes are, in the majority of cases, obligatory. For some degree achievements, the unprefixed imperfective version is unattested, e.g., unlike vy-prázdnit ‘from-empty’ or u-prázdnit ‘at-empty’, there is no *prázdnit ‘empty’. Or, in other cases, the perfective prefixed version is at least the unmarked one and much more frequent than the corresponding imperfective degree achievement. 9
It seems to us that there are two types of languages with respect to degree achievements: (i) the English type of language where the interpretation of degree achievements is regulated by the Interpretive Economy principle; (ii) the Czech type of language where the role of the Interpretive Economy principle is taken up by morphosyntactic processes.
A natural question to ask is what the combinatorial possibilities of Czech degree achievements with various kinds of modifiers are. At first blush, it seems that the combinatorial pattern follows the predictions of our approach. By way of example, upper-bounded modifiers are much more frequent with Czech degree achievements prefixed with evaluative prefixes and lower-bounded modifiers prefer non-evaluative prefixed degree achievements. However, a full account of degree modifiers and their interaction with verbal prefixes cannot be tackled here but awaits future research.
Returning to the analysis at hand, for the discussed cases of prefixes it would mean that o- ‘around’, na- ‘on’ and u- ‘at’ will pick up the non-evaluative version of the type shifter in (24) and the prefixes vy- ‘from’ and pro- ‘through’ will select the evaluative version in (23). The (non-)evaluative type shifter is selected by the algebraic properties of the prefix: if the prefix denotes a concatenable trajectory, it selects the non-evaluative type shifter, if the trajectory is not concatenable, the prefix picks up the evaluative type shifter. The application of the type shifters to the non-evaluative and evaluative interpretation of the open-scale degree achievements is illustrated below in (25) and (26), respectively. The application of the type shifters (23) and (24) to other types of scales would be analogical.
| Pokoj | se | o-hřál. | non-evaluative |
| room | refl | around-warmed.3sg | |
| ‘The room warmed.’ | |||
| ∃e[warm
|
|||
| Pokoj | se | vy-hřál. | evaluative |
| room | refl | from-warmed.3sg | |
| ‘The room warmed.’ | |||
| ∃e[warm
|
|||
We are aware that our proposed solution faces some difficulties (as also pointed out by one of the anonymous reviewers). One of them can be exemplified with (26). Our evaluative type shifter requires the degree of change to reach a maximum. But for relative adjectives like hot or cold, the usual treatment (after Kennedy & McNally 2005 a.o.) predicts that their scale is open without any clearly identifiable maximum, which then holds of degree achievements as well. This is not the case for absolute adjectives, where our semantics would work without problems. 10 Nevertheless, the missing maximum of relative degree achievements seems like a serious problem of our approach since in (26) we refer to it. The solution to this problem could be proposed as follows. First, notice that for adjectives like hot, cold and dry, a so-called scalar variability is usually assumed (see Kennedy & McNally 2005 for adjectives and Kennedy & Levin 2008 for degree achievements). Let us illustrate the scalar variability with an adjective cold: in its (default) relative scale usage, it behaves as a standard relative adjective, allowing modification with very and lacking evaluative inference in comparatives: a sentence John's tea was colder than yours does not imply that John's tea was cold. But in its absolute scale usage, the adjective allows modification by completely and in a sentence like After two hours, John's tea was colder than yours, the comparative can be interpreted evaluatively (the max for the absolute scale would be, in most cases, the room or other contextually given temperature). This reasoning is confirmed by a short query into COCA (Davies 2009), where we found 1445 sentences where cold was modified with very but also 31 sentences where its degree modifier was completely and the adjective was evidently interpreted evaluatively with respect to a contextually given maximum. Consider one of the examples in (27). Not all adjectives show the scalar variability behavior though, adjectives like tall seem to be stubbornly relative and we did not find any absolute degree modifier with tall in COCA.
| Phoebe took a sip of her tea, which must have been completely cold by now. | COCA |
Going back to the problematic case in (26), we now propose the following solution. In the case of relative degree achievements prefixed with evaluative prefixes, the underlying scale is shifted to the absolute one. The maximum then comes from the context, e.g., comfortable temperature in (26), room temperature for cold. In this way, our proposal does not run into problems and makes an interesting prediction: the perfective evaluative degree achievements based on relative adjectives should be grammatical only for such degree achievements that allow the scalar variability. So far, this prediction seems to us to be correct – we did not find any evaluative degree achievements based, e.g., on the adjective dlouhý ‘long’ – but we have to check the prediction thoroughly in a future work.
Next to the problems discussed above, a lot of empirical work lies ahead: we plan to run some experiments which will give us a more precise measure of the two evaluativity tests with various classes of degree achievements. Then, it is an open question whether the pattern observed above can be scaled up to all prefixes. This, too, is another area where more systematic data work awaits us.
5 Summary
In this article, we presented systematic data patterns of Czech degree achievements. The behavior of Czech degree achievements is puzzling from the point of the current standard scalar theory of degree achievements. We presented the entailment behavior of Czech degree achievements and supported our judgments with ambiguities triggered by again. We proposed that the scalar theory is correct, but it has to be (conservatively) amended with two additional type shifters, which formalize the more grammaticalized nature of Czech degree achievements (compared to the more lexical semantics-based nature of English degree achievements).
Certainly, many open questions and plans for future work remain. The first package of problems and questions concerns the perfective degree achievements based on relative scales, which we discussed in Section 4.2. Only thorough empirical work can bring more support for the proposed solution.
Another issue concerns the relationship between degree achievements and non-culminating accomplishments like English Mary ate the pizza in five minutes (although not completely) from Martin (2019) a.o. At first sight, their ability to be interpreted either – by default – as telic but in some context as atelic is very reminiscent of the problems we discussed here. But the current literature on Slavic non-culminating accomplishments (like Tatevosov & Ivanov 2009 a.o.) suggests that Slavic degree achievements and non-culminating accomplishments should be treated separately. Nevertheless, we noticed that the Czech prefix o-, which shows the non-evaluative interpretation (like in (17)) seems to enable the non-culminating accomplishment interpretation in a sentence like Petr o-jedl pizzu (ale většinu nechal) ‘Petr ate the pizza but left most of it’. This generally points at another possible theoretical explanation of the patterns discussed in this article, namely relating evaluativity with the maximal interpretation of the argument undergoing the scalar change. This issue is also something we want to pursue in future work.
Acknowledgments
The authors would like to thank the organizers and the audience of SinFonIJA 13 in Budapest, Hungary at the Research Institute for Linguistics for the opportunity to present the research and a stimulating debate.
We further would like to acknowledge three anonymous reviewers whose comments resulted in a number of improvements to this paper. As usual, remaining errors and shortcomings are our own responsibility.
References
Abusch, Dorit . 1986. Verbs of change, causation and time. Technical report. Center for the Study of Language and Information, Stanford University, Stanford, CA.
Asbury, Anna , Berit Gehrke and Veronika Hegedűs . 2006. In C. Keskin (ed.) One size fits all: Prefixes, particles, adpositions and cases as members of the category P. Uil OTS Yearbook, 2006. 1–17. Available at: http://www.beritgehrke.com/uploads/1/0/6/4/106403025/psasbgehheg3011.pdf.
Beck, Sigrid . 2005. There and back again: A semantic analysis. Journal of Semantics 22(1). 3–51. https://doi.org/10.1093/jos/ffh016.
Bobaljik, Jonathan David . 2015. Suppletion: Some theoretical implications. Annual Review of Linguistics 1(1). 1–18. https://doi.org/10.1146/annurev-linguist-030514-125157.
Brasoveanu, Adrian and Jessica Rett . 2018. Evaluativity across adjective and construction types: An experimental study. Journal of Linguistics 54(2). 263–329. https://doi.org/10.1017/S0022226717000123.
Brecht, Richard D. 1985. The form and function of aspect in Russian. Issues in Russian Morphosyntax 10. 9–34.
Davies, Mark . 2009. The 385+ million word corpus of contemporary American English (1990–2008+): Design, architecture, and linguistic insights. International Journal of Corpus Linguistics 14(2). 159–190.
Dowty, David R. 1979. Word meaning and Montague grammar (Synthese Language Library, Vol. 7). Dordrecht, Boston, MA & London: Reidel. https://doi.org/10.1007/978-94-009-9473-7.
Filip, Hana . 2008. Events and maximalization: The case of telicity and perfectivity. In S. Rothstein (ed.) Theoretical and crosslinguistic approaches to the semantics of Aspect. Amsterdam/Philadelphia: John Benjamins. 217–256. https://doi.org/10.1075/la.110.10fil.
Filip, Hana and Susan Rothstein . 2006. Telicity as a semantic parameter. In J. Lavine , S. Franks , H. Filip and M. Tasseva-Kurktchieva (eds.) Formal approaches to Slavic linguistics, Vol. 14. Ann Arbor, MI: Michigan Slavic Publications. 139–156. Available at: https://www.semanticscholar.org/paper/Telicity-as-a-Semantic-Parameter-Filip/20ca0e497b77934d97fa98d06a0b9a95fc9adb6c?p2df.
Grano, Thomas and Chris Kennedy . 2012. Mandarin transitive comparatives and the grammar of measurement. Journal of East Asian Linguistics 21(3). 219–266. https://doi.org/10.1007/s10831-012-9090-y.
Hay, Jennifer , Christopher Kennedy , and Beth Levin . 1999. Scalar structure underlies telicity in “degree achievements”. Semantics and Linguistic Theory 9. 127–144. https://doi.org/10.3765/salt.v9i0.2833.
Henderson, Robert . 2013. Quantizing scalar change. Semantics and Linguistic Theory 23. 473–492. https://doi.org/10.3765/salt.v23i0.2674.
Kearns, Kate . 2007. Telic senses of deadjectival verbs. Lingua 117. 26–66. https://doi.org/10.1016/j.lingua.2005.09.002.
Kennedy, Christopher . 1997. Comparison and polar opposition. Semantics and Linguistic Theory 7. 240–257.
Kennedy, Christopher . 2012. The composition of incremental change. In V. Demonte and L. McNally (eds.) Telicity, change, and state. Oxford: Oxford University Press. 103–121. https://doi.org/10.1093/acprof:oso/9780199693498.003.0004.
Kennedy, Christopher and Beth Levin . 2008. Measure of change: The adjectival core of degree achievements. In L. McNally and C. Kennedy (eds.) Adjectives and adverbs: Syntax, semantics and discourse. Oxford: Oxford University Press. 156–182. Available at: http://semantics.uchicago.edu/kennedy/docs/kl07-measure.pdf.
Kennedy, Christopher and Louise McNally . 2005. Scale structure, degree modification, and the semantics of gradable predicates. Language 81(2). 345–381. https://doi.org/10.1353/lan.2005.0071.
Křen, Michal , Václav Cvrček , Tomáš Čapka , Anna Čermáková , Milena Hnátková , Lucie Chlumská , Tomáš Jelínek , Dominika Kováříková , Vladimír Petkevič , Pavel Procházka , Hana Skoumalová , Michal Škrabal , Petr Truneček , Pavel Vondřička and Adrian Zasina . 2015. SYN2015: Reprezentativní korpus psané čeştiny. Prague: Ústav Českého národního korpusu (Czech National Corpus), FF UK. Available at: http://www.korpus.cz.
Martin, Fabienne . 2019. Non-culminating accomplishments. Language and Linguistics Compass 13(8). e12346. https://doi.org/10.1111/lnc3.12346
Matushansky, Ora . 2002. On formal identity of Russian prefixes and prepositions. MIT Working Papers in Linguistics 42. 217–253.
Pedersen, Walter A . 2015. A scalar analysis of again-ambiguities. Journal of Semantics 32(3). 373–424. Available at: https://doi.org/10.1093/jos/ffu003.
Piñón, Christopher . 2005. Adverbs of completion in an event semantics. In H.J. Verkuyl , H. de Swart and A. van Hout (eds.) Perspectives on aspect. Dordrecht: Springer. 149–166. https://doi.org/10.1007/1-4020-3232-3_8.
RNC . 2003–2020. Russian National Corpus – Nacional’nyj korpus russkogo jazyka. Moscow: Institut russkogo jazyka imeni V.V. Vinogradova, RAN. Available at: http://ruscorpora.ru/new/index.html.
Rotstein, Carmen and Yoad Winter . 2004. Total adjectives vs. partial adjectives: Scale structure and higher-order modifiers. Natural Language Semantics 12(3). 259–288. https://doi.org/10.1023/B:NALS.0000034517.56898.9a.
Sawada, Osamu and Thomas Grano . 2011. Scale structure, coercion, and the interpretation of measure phrases in Japanese. Natural Language Semantics 19(2). 191–226. https://doi.org/10.1007/s11050-011-9070-1.
SNK . 2020. Slovenský národný korpus (Slovak National Corpus) – prim-9.0. Bratislava: Jazykovedný ústav L’. Štúra, SAV. Available at: https://korpus.juls.savba.sk.
Svenonius, Peter and Christopher Kennedy . 2006. Northern Norwegian degree questions and the syntax of measurement. In M. Frascarelli (ed.) Phases of Interpretation (Studies in Generative Grammar 91). Berlin: Mouton de Gruyter. 133–161. https://doi.org/10.1515/9783110197723.3.133.
Tatevosov, Sergei and Mikhail Ivanov . 2009. Event structure of non-culminating accomplishments. In L. Hogeweg , H. de Hoop and A.L. Malchukov (eds.) Cross-linguistic semantics of tense, aspect, and modality. Amsterdam & Philadelphia, PA: John Benjamins. 83–130.
von Stechow, Arnim . 1995. Lexical decomposition in syntax. Amsterdam Studies in the Theory and History of Linguistic Science Series 4. 81–117. https://doi.org/10.1075/cilt.114.05ste.
von Stechow, Arnim . 1996. The different readings of wieder ‘again’: A structural account. Journal of Semantics 13(2). 87–138. https://doi.org/10.1093/jos/13.2.87.
Zwarts, Joost . 2005. Prepositional aspect and the algebra of paths. Linguistics and Philosophy 28(6). 739–779. https://doi.org/10.1007/s10988-005-2466-y.
Another related terminology used in the literature is a ‘positive’ (≈ telic/evaluative) vs. ‘comparative’ (≈ atelic/non-evaluative) interpretation of degree achievements, but since we will use the variation of the inference to the positive form of an adjective test, we use the evaluative terminology.
As all three anonymous reviewers correctly point out.
We are aware of the fact that grammatical perfective aspect does not always coincide with telicity as correctly emphasized by Filip & Rothstein (2006), Filip (2008) and others who argue for the non-reducibility of the Slavic lexical aspect to the grammatical aspect. In fact, our data concerning Slavic degree achievements point in the same direction: evaluativity is not directly derived from the grammatical or lexical aspect since both examples in (2a) and (2b) contain perfective verbs in telic interpreted sentences but just (2a) is evaluative.
The distinction between absolute and relative adjectives is valid only in gradable adjectives, but since non-gradable adjectives, such as childless or mammalian, cannot be a source for degree achievements, we put them aside.
There is a clear distinction between English open-scale degree achievements, where the non-evaluative distinction is default only and can be overridden, and the Czech type of these degree achievements, where the evaluativity is decided by the prefix. See our discussion of examples (2a) and (2b) in the previous section. Thanks to an anonymous reviewer for raising the importance of this distinction.
In the following paragraphs, we sketch our approach in a way where the evaluative/non-evaluative interpretation is derived from the topological properties of a verbal prefix. This is, of course, one of the possible approaches. We touch upon an alternative theoretical solution in Section 5.
An anonymous reviewer asks about the role of the Czech adverbial většinou ‘mostly’ in (19a). For our ears, the sentence would still be acceptable even without the adverbial. Nevertheless, it would sound slightly pragmatically odd since the predicate suchá ‘dry’ can be, sine qua non, interpreted exhaustively concerning the affected object. The adverbial většinou ‘mostly’ facilitates a non-maximal interpretation of the affected argument (see also our discussion of non-maximal interpretation stemming from non-evaluative prefixes in Section 5). Be it as it may, the crucial observation is that (18a), repeated below as (i), remains a contradiction even with added adverbial většinou ‘mostly’. This holds analogically for (22a) as well. Generally: Czech ‘mostly’ helps the non-evaluative prefixes to pass the contradiction test but doesn't save the evaluative prefixes in the same configuration.
| #Louže | vy-schly, | ale | pořád | byly | většinou | mokré. | contradiction |
| puddles | from-dried | but | still | were | mostly | wet.3pl | |
| ‘The puddles dried but they were still mostly wet.’ | |||||||
As one of the anonymous reviewers correctly remarks, our type-shifters in (23) and (24) are of the type 〈〈e, 〈v, d〉〉, 〈d, 〈e, 〈v, t〉〉〉〉, unlike Kennedy & Levin's (2008) pos v , which is of the type 〈〈e, 〈v, d〉〉, 〈e, 〈v, t〉〉〉. In fact, the logical type of our type-shifter is identical to the logical types of Kennedy & Levin's (2008) degree modifiers and also their functional projection μ, which can accommodate degree phrases like the one in The shadow lengthened about 1 meter, where the arguments to the type-shifter μ are the difference function and the degree argument 1 meter. We consider the degree argument in our type-shifters a good feature since Slavic perfective degree achievements are very open for degree modification like in (ii).
| Stín | se | prodloužil | o | jeden | metr. |
| shadow | SE | lengthened | about | one | meter |
| ‘The shadow lengthened about one meter.’ | |||||
We compared the prefixation tendencies of different types of verbs in the national corpora of Czech (Křen et al. 2015), Slovak (SNK 2020), and Russian (RNC 2020). The preformed Fisher test for Czech (p < 2.2e − 16, OR = 10.6), Slovak (p < 2.2e − 16, OR = 9.5), and Russian (p < 2.2e − 16, OR = 10.9) concludes that degree achievements are approximately 10 times more likely to be prefixed (and, therefore, perfective) than other types of verbs (e.g., activities, accomplishments, unaccusatives, unergatives).
There is an intriguing issue raised by one of the anonymous reviewers with respect to our evaluative/non-evaluative semantics and absolute scales: in cases where the absolute degree achievements combine with non-evaluative prefixes, like in (21), the proposed semantics predicts a non-zero increase of a zero degree on the appropriate scale. We assume that a solution to this problem lies in the mapping to the object where non-evaluative prefix signals non-maximal affectedness of the object undergoing the scalar change. See also the discussion of non-maximality effects in sentences like The sky darkened in an hour, but it wasn't completely dark in Kennedy & Levin (2008, ex. 13).
