• Dutch
  • Frisian
  • Saterfrisian
  • Afrikaans
Show all
1.2.4.Linking the syntactic and semantic classifications
quickinfo

The mental lexicon must encode in some way the form and meaning of the lexical items as well as certain syntactic information. We have seen, however, that there seem to be specific systematic relations between the relevant semantic and syntactic information; agents, for example, are normally external arguments and therefore typically appear as the subject of an active clause. Given that we do not want to include predictable information like this in the lexicon, it is an important question as to whether more of such correlations can be established. This section therefore aims at linking the syntactic classification in Section 1.2.2, sub II, to the aspectual event classifications based on participant roles in Section 1.2.3, sub II.

readmore
[+]  I.  Valuing classifications

An advantage of aspectual event classifications based on participant roles, such as the one in (77), repeated here as (94), is that they are explicitly linked to syntactic verb classifications of the kind sketched in Section 1.2.2. Van Voorst (1988), for instance, claims that originators and delimiters typically correspond to, respectively, external agent/cause and internal theme arguments. Such linking is a priori desirable because form and meaning can normally be considered two sides of the same coin.

94
Aspectual classification of event structure based on participant roles
Activity:
Achievement:
Accomplishment:

The requirement that the syntactic and semantic classifications should be linkable may also prevent these classifications from diverging too much, and can thus be used to evaluate individual proposals. The examples in (95), for instance, suggest that the traditional distinction between monadic (intransitive) and dyadic (transitive) verbs is incompatible with the aspectual event classification in (94) because it does not succeed in providing a natural account for the fact that while lachen'to laugh' denotes an activity, overlijden'to die' denotes an achievement.

95
a. Jan lacht.
activity
  Jan laughs
  'Jan is laughing.'
b. Jan verongelukte.
achievement
  Jan was.killed.in.an.accident

The alternative syntactic classification developed in Section 1.2.2, sub II, fares better in this respect, as it distinguishes two types of monadic verbs: the contrast between the two examples in (95) follows from Van Voorst's (1988) claim that external arguments of intransitive verbs like lachen'to laugh' typically function as originators, while internal theme arguments of unaccusative verbs like overlijden'to die' typically function as delimiters. This clearly favors the alternative classification in Table 3 of Section 1.2.2, sub III, which is repeated here as Table 6, over the traditional one.

Table 6: Classification of verbs according to the type of nominal arguments they take
name external argument internal argument(s)
no internal
argument
intransitive nominative (agent)
impersonal
one internal
argument
transitive nominative (agent) accusative (theme)
unaccusative nominative (theme)
two internal
arguments
ditransitive nominative (agent) dative (goal)
accusative (theme)
nom-dat verb dative (experiencer)
nominative (theme)
undative verb nominative (goal)
accusative (theme)

Dyadic verbs can likewise denote states, activities, achievements or accomplishments. The traditional classification with an undifferentiated set of dyadic verbs provides no means to describe these differences, whereas according to the alternative classification in Table 6 at least the verb hebben differs from all other verbs in (96) in that it is an undative verb and thus does not have an agentive argument. If it turns out that undative verbs typically denote states, this can again be seen as an argument in favor of the alternative classification.

96
a. De jongen heeft een kat.
state
  the boy  has  a cat
b. De jongen droeg een kat.
activity
  the boy  carried  a cat
c. De jongen ontdekte een kat.
achievement
  the boy  descried  a cat
d. De jongen verborg een kat.
accomplishment
  the boy  hid  a cat

Of course, it may be the case that the semantic and the syntactic classification do not reflect each other in all respects. The semantic distinctions between the examples in (96b-d), for instance, are reflected neither by the traditional nor by the alternative syntactic classification and may thus be due to additional restrictions imposed by the verb on their arguments in the way indicated in table (97): although originators and delimiters may typically correspond to, respectively, external agentive and internal theme arguments, it may be the case that external and internal arguments do not necessarily function as originators and delimiters; see also the linking rules in Levin and Rappaport Hovav (1995: Section 4.1).

97
lexical properties of transitive verbs
external argument = originator internal argument = delimiter
dragen'to carry' +
ontdekken'to discover' +
verbergen'to hide' + +

The discussion of the examples in (96) therefore suggests that the distinction between (96a) and (96b-d) is syntactic, whereas the distinctions between the examples in (96b-d) may be of a purely semantic nature. This may also account for the sharp contrast between the attributive (a)-examples in (98), on the one hand, and the remaining ones, on the other.

98
a. *? de een kat hebbende jongen
  the  a cat  having  boy
a'. * de gehadde kat
  the  had  cat
b. de een kat dragende jongen
  the  a cat  carrying  boy
b'. de gedragen kat
  the  carried  cat
c. de een kat ontdekkende jongen
  the  a cat  descrying  boy
c'. de ontdekte kat
  the  descried cat
d. de een kat verbergende jongen
  the  a cat  hiding  boy
d'. de verborgen kat
  the  hidden  cat
[+]  II.  Some correspondences

Subsection I has shown that the traditional syntactic classification based on the adicity of the verb cannot straightforwardly be linked to the aspectual event classifications of the type in (94) and that the alternative proposal in Table 6 based on both the number of arguments and the distinction between internal and external arguments fares much better in this respect. This subsection will show that, on the assumption that (depending on the semantic properties of the verb) external arguments are optionally interpreted as originators and internal theme arguments are optionally interpreted as delimiters, it is indeed possible to relate the syntactic classification in Table 6 to the aspectual event classification in (94). Given that goal, but not experiencer, arguments may function as the "new location" of a theme, we will also briefly consider whether the second internal argument can be interpreted as a terminus (a point of termination) in the sense of Tenny (1994); see the discussion of example (82) in Section 1.2.3, sub II, for this notion.
      In order to maximize contrasts and to highlight a number of potential problems, we will group the verbs on the basis of their adicity. We will not discuss impersonal verbs like regenen'to rain' and vriezen'to freeze', because we have little to say about them in this context. Note further that the discussion below is occasionally somewhat tentative in nature and presents a research program in progress rather than a set of well established facts/insights; the discussion below will therefore point out that there are still a number of questions that require further investigation.

[+]  A.  Verbs with one argument

At first sight the case of monadic verbs seems rather simple: as predicted, verbs with the behavior of prototypical intransitive verbs like lachen'to laugh' denote activities, whereas verbs with the behavior of prototypical unaccusative verbs like arriveren'to arrive' denote achievements.

99
Intransitive
Unaccusative
a. Jan heeft/*is gelachen.
  Jan has/is  laughed
a'. Jan is/*heeft gearriveerd.
  Jan is/has  arrived
b. * de gelachen jongen
  the  laughed  boy
b'. de gearriveerde jongen
  the  arrived  boy
c. Er werd gelachen.
  there  was  laughed
c'. * Er werd gearriveerd.
  here  was  arrived

There are, however, a number of monadic verbs exhibiting mixed behavior and seem to refer to states: this is illustrated for the verbs drijven'to float' and bloeden'to bleed' in (100). The selection of the auxiliary hebben as well as the impossibility of using the past participle attributively suggest that we are dealing with intransitive verbs, whereas the impossibility of impersonal passivization suggests that we are dealing with unaccusative verbs.

100
a. Jan heeft/*is gebloed.
  Jan has/is  bled
a'. Jan heeft/*is op het water gedreven.
  Jan has/is  on the water  floated
b. * de gebloede jongen
  the  bled  boy
b'. * de gedreven jongen
  the  floated  boy
c. * Er werd gebloed.
  there  was  bled
c'. * Er werd gedreven.
  there  was  floated

That we are not dealing with an activity is clear from the fact that the subject can be inanimate, whereas the subjects of verbs denoting an activity normally take animate subjects or a small set of inanimate subjects like computer that can be construed as performing the action. That we are not dealing with an achievement is clear from the fact that there is no logically implied endpoint.

101
a. Jan/de wond bloedt heftig.
  Jan/the wound  bleeds  fiercely
b. Jan/de band drijft op het water.
  Jan/the tire  floats  on the water

Given that we have adopted as our working hypothesis that internal and external arguments only optionally function as, respectively, originators and delimiters, there is no a priori reason for assuming intransitive or unaccusative status for these verbs. If we assume that drijven and bloeden are unaccusative, we have to conclude that selection of the auxiliary zijn'to be' and attributive use of the past participle are sufficient but not necessary conditions for assuming unaccusativity; Subsection B2 will show that there is indeed reason for assuming that auxiliary selection and attributive use of the past participle not only depend on unaccusativity of the verb but are subject to additional aspectual conditions; see Mulder (1992) and Levin & Rappaport Hovav (1995) for similar conclusions.

[+]  B.  Verbs with two arguments

Table 6 distinguishes three types of dyadic verbs: transitive, nom-dat and undative verbs. The following subsections will discuss these three groups.

[+]  1.  Transitive verbs

The examples in (97b-d), repeated here as (102), have already illustrated that prototypical transitive verbs can denote activities, achievements and accomplishments. In fact, this was the original motivation for our claim that internal and external arguments only optionally assume the roles of originator and delimiter; see Table (97) in Subsection I.

102
a. De jongen droeg een kat.
activity
  the boy  carried  a cat
b. De jongen ontdekte een kat.
achievement
  the boy  descried  a cat
c. De jongen verborg een kat.
accomplishment
  the boy  hid  a cat
[+]  2.  Nom-dat verbs

Nom-dat verbs are characterized by the fact that the subject can follow the object, which appears as a dative noun phrase in German in the unmarked case. Given that this also holds for passivized ditransitive verbs, Den Besten (1985) concluded that the subjects of nom-dat verbs are internal theme arguments.

103
a. dat die meisjesnom Peter/hemdat direct opvielen.
  that  those girls  Peter/him  immediately  prt.-struck
  'that Peter/he noticed those girls immediately.'
b. dat Peter/hemdat die meisjesnom direct opvielen.
  that  Peter/him  those girls  immediately  prt.-struck

This analysis immediately accounts for the fact that examples such as (103) are interpreted as achievements: nom-dat verbs are like monadic unaccusative verbs in that they lack external arguments that could function as originators and that their internal arguments may function as delimiters. The Nom-dat verbs we have discussed so far furthermore exhibit all the typical properties of monadic unaccusative verbs: they select the auxiliary zijn, their past participles can be used attributively to modify a head noun that corresponds to the subject of the clause, and they resist passivization.

104
a. dat die meisjes Peter/hem direct zijn/*hebben opgevallen.
  that  those girls  Peter/him  immediately  are/have  prt.-struck
b. de hem direct opgevallen meisjes
  the  him  immediately  prt.-struck  girls
c. * Er werd Peter/hem direct opgevallen.
  there  was  Peter/him  immediately  prt.-struck

The claim that internal arguments only optionally function as delimiters predicts, however, that there are also nom-dat verbs that do not involve some implied endpoint and thus denote simple states. And, in fact, Den Besten (1985) does list a number of nom-dat verbs with this property. One example is the verb smaken'to taste' in (105).

105
a. dat de broodjes Peter/hem smaakten.
  that  the buns  Peter/him  tasted
  'that Peter/he enjoyed his buns.'
b. dat Peter/hem de broodjes smaakten.
  that  Peter/him  the buns tasted

Although the relative order of the object and the subject in (105b) unambiguously shows that the subject de broodjes is an internal argument, it should be noted that verbs like smaken do not exhibit all of the properties that we find in (104). Like all unaccusative verbs, they do not allow impersonal passivization, but they select the auxiliary hebben instead of zijn, and their past participles cannot be used attributively to modify a head noun that corresponds to the subject of the clause.

106
a. dat Peter/hem de broodjes hebben/*zijn gesmaakt.
  that  Peter/him  the buns  have/are  tasted
b. de Peter/hem gesmaakte broodjes
  the Peter/him  tasted  buns
c. * Er werd Peter/hem gesmaakt.
  there  was  Peter/him  tasted

It is interesting to note that the pattern in (106) is like the pattern established for the stative verbs drijven'to float' and bloeden'to bleed' in (100). This supports the suggestion in Subsection A that the verbs drijven and bloeden are also unaccusative verbs and that their mixed behavior with respect to the unaccusativity tests should be accounted for by assuming that auxiliary selection and attributive use of past participles are subject to both syntactic and aspectual conditions.

[+]  3.  Undative verbs

Undative verbs do not have an external argument and we would therefore expect that there is no originator; undative verbs therefore denote either states or achievements depending on whether their internal theme argument functions as a delimiter or not. The examples in (107) show that this prediction is indeed borne out: depending on the verb in question, we are dealing with a state, an achievement, or a special type of state that we may call an anti-achievement.

107
a. Jan heeft het boek.
state
  Jan has  the book
b. Jan krijgt het boek.
achievement
  Jan gets  the book
c. Jan houdt het boek.
anti-achievement
  Jan keeps  the book

The achievement reading in (107b) may be due to the fact that the IO-subject Jan functions as a goal, which, in turn, triggers a delimiter interpretation of the internal theme argument; if so, this would support our suggestion in the introduction to this section that goals function as a terminus (point of termination) in the event.
      This claim that goals function as a terminus may also account for the fact that the IO-subjects of cognition verbs like weten/kennen'to know' in (108a), which we will show in Section 2.1.4 to be part of a second set of undative verbs, must be interpreted as experiencers; the fact that these verbs normally denote states would then be incompatible with a goal/terminus interpretation of the dative phrase. The dyadic verb leren'to learn' in (108b) stands in an anti-causative relationship to the triadic accomplishment verb leren'to teach'; cf. Marie leert Jan de fijne kneepjes van het vak'Marie is teaching Jan the tricks of the trade'. The indirect object of the triadic and the subject of the dyadic verb both act as a goal, which introduces a point of termination in the event; this leads to the achievement reading of (108b).

108
a. Jan kent de fijne kneepjes van het vak.
state
  Jan knows  the detailed tricks of the trade
  'Jan knows the tricks of the trade.'
b. Jan leert de fijne kneepjes van het vak.
achievement
  Jan learns  the detailed tricks of the trade

Given the discussion of the examples in (108), it may be tempting to analyze other ditransitive verbs with experiencer subjects, like the perception verbs horen'to hear' and zien'to see', likewise as undative verbs; we will leave it to future research to investigate whether this might be on the right track.

[+]  C.  Verbs with three arguments

Indirect objects of ditransitive verbs normally function as goals. If goal arguments introduce a terminus, we would expect that (definite) theme arguments would normally function as a delimiter. If so, we would also expect that, depending on whether the subject functions as an originator or not, ditransitive verbs would normally denote achievements or accomplishments. The examples in (109) show that this expectation is indeed borne out.

109
a. Zijn succes gaf Peter een prettig gevoel.
achievement
  his success  gave  Peter a nice feeling
b. Jan stuurde Peter een mooi boek.
accomplishment
  Jan  sent  Peter  a nice book
[+]  D.  Conclusion

It seems that the semantic classification in (94) and the syntactic classification in Table 6 can to a certain extent be linked. At present, we are able to show this only for the more prototypical cases; future research will have to show whether this is also possible with less prototypical cases. We expect such research to reveal certain potential problems for some of the claims adopted in the discussion above. For example, the unaccusative verbs overlijden'to die', arriveren'to arrive' and vertrekken'to leave' in (110) seem to differ in the extent to which the subject is able to control the event. Whereas the subject of overlijden has no control at all, the subject of vertrekken does have control over the event; the subject of arriveren seems to take some intermediate position in this respect.

110
a. Jan overlijdt morgen.
  Jan dies  tomorrow
b. Jan vertrekt/arriveert morgen.
  Jan leaves/arrives  tomorrow

The contrast might be accounted for either by assuming that the internal argument of an unaccusative verb is not only able to function as a delimiter but also as an originator, or by assuming that assignment of the property of control is not linguistic in nature but reflects our knowledge of the world. Given that the former would open many new classification options, we can only determine whether such an approach would be feasible by investigating whether the newly predicted verb classes do indeed exist.

References:
  • Besten, Hans den1985The ergative hypothesis and free word order in Dutch and GermanToman, Jindřich (ed.)Studies in German GrammarDordrecht/CinnaminsonForis Publications23-65
  • Besten, Hans den1985The ergative hypothesis and free word order in Dutch and GermanToman, Jindřich (ed.)Studies in German GrammarDordrecht/CinnaminsonForis Publications23-65
  • Levin, Beth & Rappaport Hovav, Malka1995Unaccusativity at the syntax-lexical semantics interfaceCambridge, MA/LondonMIT Press
  • Levin, Beth & Rappaport Hovav, Malka1995Unaccusativity at the syntax-lexical semantics interfaceCambridge, MA/LondonMIT Press
  • Mulder, René1992The aspectual nature of syntactic complementationLeidenUniversity of LeidenThesis
  • Tenny, Carol L1994Aspectual roles and the syntax-semantics interfaceDordrechtKluwer
  • Voorst, Jan van1988Event structureAmsterdam/PhiladelphiaJohn Benjamins
  • Voorst, Jan van1988Event structureAmsterdam/PhiladelphiaJohn Benjamins
report errorprintcite