Spatial Semantic Features: \'Convergent\' vs. \'Divergent\' Verbs (PDF ...

Spatial Semantic Features: ‘Convergent’ vs. ‘Divergent’ Verbs Ramin Assadollahi ([email protected]) FB Psychologie / SFB 471, Universität Konstanz Universitätsstr. 10, 78457 Konstanz, Germany

Wilhelm Geuder ([email protected]), FB Sprachwissenschaft / SFB 471, Universität Konstanz Universitätsstr. 10, 78457 Konstanz, Germany

Matthias Weisgerber ([email protected]) FB Sprachwissenschaft / SFB 471, Universität Konstanz Universitätsstr. 10, 78457 Konstanz, Germany

Abstract In this study, we investigate two classes of spatial semantic features for verbs, using a priming paradigm. The verb classification is based on a distinction between the movement directions ‘convergent’ (reducing the distance) and ‘divergent’ (enlarging the distance between two objects). These directionality features were also applied to abstract conceptual domains like social contact or functional relation. The results confirm the concrete / abstract distinction for verbs and show that ‘convergent’ and ‘divergent’ are able to produce priming effects within, and possibly across, conceptual domains. Convergent and divergent did not behave symmetrically leading us to the conclusion that they may be two different features rather than values of a ‘direction’ feature.

Introduction The analysis of verbs of movement constitutes a research area at the intersection of linguistic semantics and spatial cognition. Current research in this field is exploring spatial ontologies of abstract entities like paths, directed axes, vectors, regions etc. in order to come up with a rigorous model of the spatial relations that are named by movement verbs (cf. e.g. Eschenbach, Tschander, Habel & Kulik 2000; Schmidtke, Tschander, Eschenbach & Habel, 2003; Zwarts 2003). A second strand of research in semantics, “cognitive semantics”, has developed representations in terms of socalled topological/kinetic schemas for representing verb meanings (cf. Langacker, 1990; Talmy, 2000). These are usually treated as holistic spatial arrangements that may undergo schema transformations (cf. Lakoff, 1987; Tyler & Evans, 2003) and may be transferred to different conceptual domains; in this way, certain types of meaning• variation with verbs are accounted for. An example of schema transformation with movements are the two different uses of roll in The ball rolled downhill (roll as a form of movement) vs. She rolled a blanket around the doll (transformation of the movement schema into a change-of-shape schema) (Iwata, 2002). An example for a domain change would be between literal giving and giving orders (an application of the transfer schema in the domain of verbal communication; Newman, 1996: 136). While the formal approaches have hardly been extended to non-spatial domains so far, it is obvious that

the application of abstract spatial entities (like paths and vectors) in non-spatial conceptual domains emerges as a research question that is of interest across different schools of thought in semantics. In this paper, we investigate verbs which all denote a change in distance but involve a distance measure in different conceptual domains (like change in spatial or in social distance). Our question is whether these different types of change in distance have a unified semantic and mental representation as abstract direction vectors. In other words, the question is whether elementary direction vectors (as elements of a spatial ontology) can be shown to be active as semantic features of these verbs. We are aware of but a single empirical study with a somewhat similar aim in the literature, namely Richardson, Spivey, Barsalou & McRae (2003). Using an interference paradigm, these authors found that various verbs belonging to non-spatial domains involved a representation of directional gradients that were either vertical or horizontal (e.g. respect evoked a vertical spatial relation, but hunt an horizontal one). However, unlike Richardson et al., we investigate a relation that applies to spatial and abstract domains alike, via generalisation of the same relation; we are not concerned with metaphorical transfers between spatial and other meanings (this difference will be explained below).

‘Convergent / Divergent’ Direction as a Semantic Feature Relative Directions in Verb Semantics The two spatial situation types under investigation here are: a thing moving towards another one (‘convergent’ verbs), or a thing moving away from another one (‘divergent’ verbs). For example, to approach would be classified as a convergent verb because the situation involves a reduction of distance between the entities denoted by subject and object of the verb, while to spread qualifies as a divergent verb because it involves an increase in distance among the parts of a substance. In a broad classification of movement events in terms of their reference frame, the convergent / divergent distinction belongs to a class ‘movement with respect to a reference

object’ (in contrast to, say, fall [movement in an absolute direction], or to t u r n [movement wrt. intrinsic orientation]). Obviously, the classification into convergent vs. divergent verbs does not distinguish whether the entities in question make contact or not. Moreover, given the hypothesis that relative direction of movement is represented via abstract directional features (classes of direction vectors), the classification should be independent of the classification of the objects involved and independent of the relation between movement vectors and objects. Hence, in our stimulus set we generalise over various mappings between direction vectors and objects, so we may have relative movement of a pair of separate objects (to approach), or movement among connected parts of a thing or within groups and masses (to spread). Taking these classification criteria together, the feature ‘convergent direction’ may comprise the following scenarios: • Reduction of distance between objects • Reduction of distance among parts of one object (change of shape) • Making contact with another object • Entering into an inclusion relation • Fusion of two objects to a new one In a completely parallel fashion, we take divergent situations to involve either: • Increase of distance between two objects • Increase of distance among parts of one object (change of shape) • Separation of contact • Leaving an inclusion relation • Splitting up an object into parts In this way, the classification is sensitive purely to the existence of vectors with a particular relative orientation, and generalises over everything else. If convergent and divergent verbs, so defined, turn out to behave as a consistent class in a behavioural study, then the existence of abstract spatial entities as semantic primitives (in this case, direction vectors) would be supported. Verbs that do not denote movement in space may exhibit convergence / divergence with respect to conceptual domains that still involve some type of distance relation in a literal sense. The most important ones are changes with respect to: • Social distance or inclusion (to ally with vs. excommunicate) • Possession / rights (to award vs. expropriate) • Functional association (to substantiate vs. disallow) Note that the classification as an abstract situation type exclusively derives from the fact that the change denoted by the verb is not a movement in space; abstract situations may well involve concrete objects, though. For instance, awarding something to someone is an abstract relation, since it involves non-spatial concepts like honour, possession, etc. Hence, a verb like award always expresses a change in an abstract domain, even if the situation involves handing over a medal, for instance. It can now also be seen that the abstract verb types listed above are not spatial metaphors, at least not in the

same strict sense as the stimuli in the experiment by Richardson et al. (2003) mentioned in the introduction. This may be seen as a purely terminological point, but we want to emphasise that this distinction should be made. The difference, we maintain, is that the interpretation of metaphors has to be mediated not only by a relational structure in word meanings, but also by additional world knowledge. For example, the association of a vertical direction with the verb respect arguably derives from our knowledge about certain effects of the forces of gravity: a difference in vertical position is known to facilitate force exertion in downward direction, and hence higher position comes to be associated metaphorically with (social) power, bringing about attitudes like reverence and respect. The investigation of metaphorical links between word meanings is a major topic in cognitive semantics (cf. Lakoff 1987, who is, however, not interested in the distinction between effects of structural similarity and world knowledge), and the Richardson et al. study can be seen as an attempt at empirically substantiating this line of research. The present study, in contrast, is concerned with basic structural properties that are contained in the very lexical meanings, and with describing them in terms of elementary geometrical relations. Hence, we argue that such structural generalisations must be kept separate from the notion of metaphor, since the latter also involves factors extraneous to word meaning.

Testing the Semantic Analysis Our setup yields two classes (each) of convergent and of divergent verbs, namely ‘concrete’ cases, involving purely spatial movement, and ‘abstract’ cases, involving a change in a distance relation with respect to social or functional domains. If the latter domains are modelled in terms of spatial entities, too, both concrete and abstract convergent verbs should show parallel behaviour in particular tasks. We investigated these spatial semantic features in a priming experiment using a lexical decision task. Subjects saw words either from the four stimulus categories (concrete convergent, concrete divergent, abstract convergent, abstract divergent) or pseudo-words. In four different runs, we used one of the categories as primes and all categories plus pseudo-words as targets. Thus, every category served as prime and as target. We hypothesised that because of shared directional features, a verb from one class would prime targets from its own class, but not targets from other classes. For example, convergent concrete primes should lead to faster reaction times to other convergent concrete verbs and the response to no other class should be facilitated. Results from cross-modal studies (Zwaan group: Stanfield & Zwaan, 2001; Zwaan, Madden, Yaxley, & Aveyard, 2004; Zwaan et al., 2002) investigating matches and mismatches of movement underpin our hypothesis: Zwaan and colleagues showed that visual movements facilitate the processing of acoustically presented movement-words, but only if both movements have the same direction. Other experimental setups yielded similar findings (e.g. McDermott & Roediger, 1994). The hypothesis stated above takes convergent vs. divergent as different values of the same feature type

“direction”. This is not the only logical possibility, however. A concurring hypothesis might be that “convergent” and “divergent” are not values (maybe in the form of opposite ends of a scale) of the feature “direction”, but they may constitute qualitatively different, separate f e a t u r e types. (This form of conceptualisation would imply that a word can have both features with independent +/- values). Consequently, the concept of “divergent” and “convergent” as features rather than values would lead to the somewhat weaker hypothesis that priming could occur for one of the two features only. The stronger hypothesis would be predicted from convergence and divergence as values of the feature “direction”: Either priming effects should occur for both or for none. Investigating a second verb property, we hypothesised that the distinction between spatial and non-spatial (e.g. social, functional) directional changes may correspond to the concrete / abstract differentiation discussed in the literature. In the lexical decision task literature, there is ample evidence for slower reaction times to abstract words as compared to concrete words (e.g. deGroot, 1989; Bleasdale, 1987; Howell & Bryden, 1987; Kroll & Merves , 1986; Rubin, 1980; Whaley, 1978, James, 1975; see Schwanenflugel 1991 for a review). Also, in a priming task using verbs and nouns, Tyler et al (2002) showed that concrete primes lead to a stronger benefit in reaction time than abstract primes. In our present study, we thus hypothesised, that reaction times to “abstract” verbs should be slower than to our class of “concrete”, i.e. spatially defined, verbs.

Experiment Stimuli 50 German verbs for each of the four stimulus categories were selected from the CELEX corpus (Bayaan et al. 1993). Care was taken to exclude verbs from the stimulus set that could be interpreted as being ambiguous or vague with respect to spatial vs. abstract readings. Verbs were matched for length and frequency (mean length 9.2 letters, std dev. 1.6; mean frequency 6.5 per mio., median 3). All verbs were in third person, singular, indicative, present, active form. Moreover, we generated pseudo-words for each of the stimulus categories by interchanging morphemes or syllables between words of one category. Therefore, these words were not only pronounceable, but also included meaningful parts, hence can be described as “high-level” pseudo-words.

Setup & Procedure The experiment consisted of four separate runs. In each run, one of the stimulus categories was used as priming category. All four categories were used as target category. In addition, the respective pseudo-word set was used. The stimuli of all these five sets were presented in pseudorandomised order as targets 500ms after the prime. The inter-stimulus interval varied between 1600 and 2400ms. Stimuli were presented tachistoscopically for 100ms in

white (targets) or pink (primes) uppercase letters on a black screen 1.4m away from the subjects’ eyes.

Task The task for the 21 subjects (healthy, right-handed, native speakers of German with normal or corrected to normal eye-sight) was to press a button using the index finger of one hand when they read a pseudo-word and to press a button using the index finger of the other hand when they read a real word (response hands were balanced over conditions and subjects). Note that it was highly improbable that a verb would appear as prime and as target within one trial (p = 1/2500). The sequence of the runs was counterbalanced over subjects. Subjects were naïve with respect to the verb classes and the variation of their abstractness.

Analysis Reaction time data was trimmed to at least 300ms and at most 2000ms after the target. Moreover, the 5% of the fastest button presses as well as 5% of the slowest button presses were excluded. For each subject, mean reaction time per run/prime, condition/category (i.e. four values) went into statistical analyses. We were interested in how the reaction times to targets would depend on the different primes. Thus, for all of the four prime configurations (convergent, concrete; convergent, abstract; divergent, concrete; divergent, abstract), the reaction time data was submitted to a twoway repeated measures ANOVA (target: concrete|abstract) x (target: convergent|divergent). Newman-Keuls tests were used for investigating betweencategory differences.

Results For the sake of brevity, we will only report the results of the priming experiment. The lexical decision task yielded strong differences between real and pseudo-words, with reaction times to pseudo-words being at least 200ms slower. Prime concrete convergent: Difference in RT was only significant for the concrete class (interaction: F(1,20) = 5.5, p = 0.029, cf. fig. 1). RT to convergent verbs was faster than RT to divergent verbs. Post-hoc tests revealed that convergent concrete verbs indeed led to faster RTs than divergent concrete (p = 0.014), convergent abstract (p = 0.0002) and divergent abstract (p = 0.0003) verbs.

Figure 1. Reaction times to targets after convergent concrete primes, lines indicate the standard error. Prime divergent concrete: No difference between categories. Prime convergent abstract: There was a difference in RT only for the abstract category (interaction: F(1,20) = 8.8, p = 0.007, cf. fig. 2). RTs to abstract divergent verbs were slower than to abstract convergent verbs (p = 0.02), to concrete convergent verbs (p = 0.02) and to concrete divergent verbs (p = 0.005). On the first glance, it seems that this should be interpreted as slower RTs for abstract divergent verbs. However, a main effect for abstract/concrete (F(1,20) = 9.5, p = 0.005) revealed that generally concrete verbs led to faster RTs – an effect superimposed on the interaction in fig. 2. Thus, we interpret the result that in fact, the abstract convergent category led to faster RTs compared to the abstract divergent category and the concrete categories did not differ from the convergent abstract category because they were generally faster. Prime divergent abstract: There was a marginally significant main effect for divergence: RTs to divergent verbs were faster than to convergent verbs, irrespective of abstractness (p = 0.08).

Figure 2. Reaction times to targets after convergent abstract primes, lines indicate the standard error.

Discussion Our investigation looked for priming effects of semantic features of relative direction (convergent vs. divergent) in verbs belonging to different conceptual domains (spatial distance vs. social or functional distance).

First of all, the results confirm the expectations concerning the concrete / abstract distinction for verbs: verbs denoting a change of distance in space yielded significantly faster reaction times than verbs denoting changes in social or functional distance relations. This mirrors the well-known concreteness effects from other word classes (cf. Schwanenflugel 1991). Although verbs, on the whole, are usually considered as a class of linguistic items at the abstract end of the spectrum, the concrete-abstract distinction must still be made within this category. Our finding that, in particular, movement in space qualifies as a concrete type of verb meaning is plausible because movement can be defined in terms of elementary sensory categories, as opposed to abstract functional and social relations. It is true that the class of abstract verbs we investigated was semantically more heterogeneous than the concrete class (since the abstract class comprised several conceptual domains, and the concrete class only a single one, namely movement in space). If one were to compare priming effects within these two classes, this heterogeneity alone would already give the abstract verbs a disadvantage. Therefore, we should point out that semantic heterogeneity was definitely not the source of the abstractness effect in our study: it could be shown that the effect was unconnected to priming and only depended on the recognition of the verbs individually as targets. Next, we have demonstrated that “convergence” and “divergence” are able to produce priming effects. The data we obtained confirmed this hypothesis concerning the class of ‘convergent’ verbs, in the concrete domain as well as in the abstract domain. Also, some priming effects were found for divergent verbs, although here the pattern was different: there was no priming after concrete divergent verbs. However, we found a tendency for abstract divergent verbs to prime all divergent verbs (i.e. concrete as well as abstract targets). As mentioned at the outset, several semantic theories in linguistics favour representations of verb meanings in terms of modality-neutral geometrical entities like vectors, or topological schemata, or similar devices. This would lead to the expectation that cross-domain priming should occur in our experiment, e.g. from concrete divergent verbs to abstract divergent verbs etc., because they share the same abstract spatial element of a divergent direction. Our findings indicate that one has to be wary about such generalisations, although firm conclusions have to await further research. One of the four experimental conditions we investigated did produce such an effect of cross-domain priming (abstract divergent primes facilitated all kinds of divergent verbs). While this is a positive finding, the fact that it stands somewhat isolated presents a problem. With convergent primes, priming was only within the concrete domain and within the abstract domain, respectively. And concrete divergent primes did not produce any facilitation at all. Still, it must be pointed out that no effects occurred that would run counter to our hypotheses; in particular, convergent verbs never produced effects on divergent verbs or vice versa. And at any rate, the consistent priming effects obtained for convergent stimuli within each abstractness class

establishes that the directional feature ‘convergent’ does play a role in the conceptual representation of these verbs. A striking result that follows from our data is that the feature ‘divergent’ did not behave simply as the opposite or inverse of the feature ‘convergent’. For example, we observed cross-domain priming for abstract divergent verbs, but only within-domain priming for convergent verbs; and also, there were clear priming effects for convergent concrete verbs, but none for divergent concrete verbs. Thus, although semantic classification criteria for convergent and divergent stimulus classes were fully symmetric (cf. the list above), they led to different effects. This finding, then, rather supports the alternative hypothesis that “divergent” and “convergent” may be separate feature types in their own right, rather than values of a unified feature type “direction”. Only under this view would it seem plausible that one of the features (convergent) should produce priming effects whereas its opposite (divergent) does not. Supportive literature is discussed further below. One might ask whether the results may rather be due to some confounding variable hidden in the class of divergent verbs which disturbed the effects of the semantic classification. One thing that comes to mind is the occurrence of verbal prefixes. Since the verb lexicon of German makes heavy use of prefixes, many of them having a directional meaning, one might ask whether the similarity between classes might be influenced by the number of prefixes shared, over and above the semantic factor of convergence / divergence. The following table shows, however, that this suspicion is not borne out: Table 1: Distribution of prefixes over verb classes: [convergent, concrete]: 18 be-, [convergent, abstract]: 12 be-, [divergent, concrete]: 15 ent-, [divergent, abstract]: 14 ent-,

10 ver-, 0 er- 19 none 8 ver-, 6 er- 24 none 7 ver-, 14 zer- 14 none 12 ver-, 1 zer- 19 none

(sums of rows may not add up to 50 due to „other“ prefixes, mostly single occurrences)

As can be seen, there is no conspicuous difference in the number of shared prefixes between the two convergent and the two divergent classes. Moreover, the prefix vercreates some amount of similarity between the classes [convergent, concrete] and [divergent, abstract], but there were no corresponding priming effects. And finally, note that the class [divergent, concrete] stands out as the one with the largest number of prefix verbs. Hence, this class has the highest probability for prime and target to exhibit identical prefixes. Notably, however, this is the only class that did not exhibit class-internal priming in our data. Our data suggest, therefore, that convergent and divergent stimuli may indeed constitute two qualitatively different features, instead of being symmetrically opposing values of one feature (direction). ‘Divergence,’ as we have seen, interacts differently with the abstractness of the stimuli, as compared to ‘convergence’. The asymmetry between the two directional classes reminds us of findings in the psychological and linguistic literature to similar effect. For example, Regier (1997) and Lakusta &

Landau (in press) find asymmetries between goal-defined paths and source-defined paths in that goal-paths are more prominent and easier to name. This connection remains to be established by future research, but there is some initial plausibility, since convergent directions clearly represent goal-defined paths and divergent directions represent source-defined paths.

Conclusion In the present study, we started by defining semantic classes of verbs according to the spatial feature of convergent and divergent directionality, applied to purely spatial movement as well as to abstract domains. These semantic features indeed yielded priming effects. We also investigated whether convergent and divergent directions produced priming across the spatial and abstract conceptual domains; mostly, priming turned out to be domain-specific. Notably, the directional opposites convergent / divergent did not behave symmetrically. This result led us to the speculation that “convergent” and “divergent” may be features rather than values in the lexical entry of a word. Finally, our study established an abstract / concrete distinction for verbs, according to which spatial movement behaves as a concrete property.

References Baayan, H., & Piepenbrock, R., & van Rijn, H. (1993). The CELEX lexical database. Linguistic Data Consortium, University of Pennsylvania, PA. Bleasdale, F. A. (1987). Concreteness dependent associative priming: Separate Lexical organization for concrete and abstract words. Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 582-594. DeGrmt, A. M. B. (1989). Representational aspects of word imageability and word frequency as assessed through word association. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 824-845. Eschenbach, C., & Tschander, L., & Habel, C., & Kulik, L. (2000). Lexical specifications of paths. In W. B. Freksa, C. Habel & K. Wender (Eds.), S p a t i a l Cognition II - Integrating Abstract Theories, Empirical Studies, Formal Methods, and Practical Applications. Berlin: Springer. Howell, J. R., & Bryden, M. P. (1987). The effects of word orientation and imageability on visual half-field presentations with a lexical decision task. Neuropsychologia, 25, 527-538. Iwata, S. (2002). Does Manner count or not? Manner-ofmotion verbs revisited. Linguistics, 40, 61-110. James, C. T. (1975). The role of semantic information in lexical decisions. Journal of Experimental Psychology: Human Perception and Performance, 104, 130-136. Kroll, J. F., & Merves, J. S. (1986). Lexical access for concrete and abstract words. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 92107. Lakoff, J. (1987). Women, Fire and Dangerous Things. What Categories Reveal about the Mind. Chicago: University of Chicago Press.

Lakusta, L., & Landau, B. (in press). Starting at the end: the importance of goals in spatial language. To appear in Cognition. Langacker, R. (1990). Concept, image, and symbol. Berlin: Mouton de Gruyter McDermott, K., & Roediger, H. (1994). Effects of imagery on perceptual implicit memory tests. Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 418-434. Newman, J. (1996). Give. A cognitive linguistic study. Berlin: Mouton de Gruyter Regier, T. (1997). Constraints on the learning of spatial terms: A computational investigation. In R. Goldstone, P. Schyns & D. Medin (Eds.), Psychology of learning and motivation, Vol. 36: Mechanisms of perceptual learning. San Diego CA: Academic Press. Richardson, D., & Spivey, M., & Barsalou, L. W., & MacRae, K. (2003). Spatial representations activated during real-time comprehension of verbs. Cognitive Science, 27, 767-780. Rubin, D, C. (1980). 51 properties of 125 words: A unit analysis of verbal behavior. Journal of Verbal Learning and Verbal Behavior, 19, 736-755. Schmidtke, H., & Tschander, L., & Eschenbach, C., & Habel, C. (2003). Change of orientation. In E. van der Zee & J. Slack (Eds.), Representing direction in language and space. Oxford: Oxford University Press. Schwanenflugel, P. J. (1991). Why are abstract concepts hard to understand? In P. J. Schwanenflugel (Ed.), The psychology of word meanings Hillsdale, NJ.: Erlbaum. Stanfield, R. A., & Zwaan, R. A. (2001). The effect of implied orientation derived from verbal context on picture recognition. Psychological Science, 12, 153156. Talmy, L. (2000). Toward a cognitive semantics. Cambridge, MA: MIT Press Tyler, A., & Evans, V. (2003). The semantics of English prepositions. Cambridge: Cambridge university Press. Tyler, L., K., & Moss, H. E., & Galpin, A., & Voice, J.K. (2002). Activating meaning in time: The role of imageability and form-class. Language and Cognitive Processes, 17 (5), 471-502. Whaley, C, P. (1978). Word-nonword classification times. Journal of Verbal Learning and Verbal Behavior, 17, 143- 154. Zwaan, R. A., & Madden, C. J., & Yaxley, R. H., & Aveyard, M. E. (2004). Moving words: Dynamic mental representations in language comprehension. Cognitive Science, 28, 611-619. Zwaan, R. A., & Stanfield, R. A., & Yaxley, R. H. (2002). Language comprehenders mentally represent the shape of objects. Psychological Science, 13, 168-171. Zwarts, J. (2003). Vectors across Spatial Domains: From place to size, orientation, shape, and parts. In E. van der Zee & J. Slack (Eds.), Representing direction in language and space. Oxford: Oxford University Press.

Appendix: Examples from the Stimulus Set Note: The English translations are approximations. Even though some items below are ambiguous between concrete and abstract readings in English, this would be excluded for their German counterparts. [convergent, concrete] umwickelt(wrap, swathe), verquirlt (whisk), pflastert (pave), apportiert (retrieve, as by a dog), kontrahiert (contract, as of a muscle), bekleckert (make stains on something), trommelt (drum), implantiert (implant), landet (land) [convergent, abstract] akkumuliert (accumulate, as of capital), adoptiert (adopt child), besoldet (salary), erringt (gain, win; e.g. victory), immigriert (immigrate), einigt (unite), konvergiert (converge), betraut (consign, entrust), behelligt (molest, bother), verlobt (affiance) [divergent, concrete] zerbröckelt (crumble), entlaust (delouse), sekretiert (secrete), spreizt (spread, splay), versprüht (disperse / spray), raspelt (grate), entsaftet (extract juice from), amputiert (amputate), portioniert (portion out), zerplatzt (burst) [divergent, abstract] entwöhnt(wean), splittet (split [figurative senses]), selektiert (select), aberkennt (deprive / disallow, e.g. a title), dekomponiert (decompose), enthebt (divest of office), vergrault (roughly: put off, make someone avoid social contact), entstammt (emanate from), emanzipiert (emancipate), opfert (sacrifice)