Abstract and Keywords
This chapter summarizes and concludes the discussion, and provides some considerations for future research. Working in a cognitive-functional framework, this study has presented an analysis of the impersonal verbs and the impersonal construction in early English that has focused on semantic and discourse-pragmatic factors and on the interaction among verbal, grammatical, and constructional meaning. By focusing on the semantic and discourse-pragmatic properties of the impersonal construction and on the interaction among verbal, grammatical, and constructional meaning, the reasons for the loss of the impersonal construction could be reassessed, pointing to the semantic and discourse-pragmatic motivations for its loss as a consequence of the drastic grammatical changes that affected the expression of transitive relations in the course of Middle English.
8.1 Summary and Results
Working in a cognitive-functional framework, this study has presented an analysis of the impersonal verbs and the impersonal construction in early English that has focused on semantic and discourse-pragmatic factors and on the interaction among verbal, grammatical, and constructional meaning. The investigation was undertaken with the following aims:
• Determining the place, function, and use of the impersonal construction within the grammatical system of Old English (OE)
• Revealing the different factors that influenced the impersonal and nonimpersonal syntactic uses of individual verbs in Old and Middle English (ME)
• Understanding the reasons for the loss of the impersonal construction as a consequence of grammatical changes that affected the expression of transitive relations in the course of Middle English
After determining the function and use of the OE impersonal construction and its place within the OE grammatical system, the investigation was extended to the Middle and Early Modern English (EModE) periods with the aims of (1) outlining diachronic tendencies in the further syntactic development of impersonal verbs, and (2) revealing functional aspects that may have motivated the ultimate loss of the impersonal construction towards the late 15th century. This diachronic investigation was based on the analysis of data from two electronic dictionaries: the Middle English Dictionary Online (MED) and the Oxford English Dictionary Online (OED).
The semantic analysis of the set of verbs that were capable of impersonal use in Old English showed that these verbs were semantically more heterogeneous than has been acknowledged in some of the previous literature. They belong to the conceptual domains of PHYSICAL SENSATION, EMOTION, COGNITION, EXISTENTIAL EXPERIENCE, MOTION, OWNERSHIP/APPROPRIATENESS, (NON)AVAILABILITY, and BENEFACTION. All of these verbs profile two participants in their semantic frames. The most important difference between these verbs lies in the fact that some express inherently dynamic processes (i.e., the verbs of physical sensation, emotion, cognition, existential experience, and motion), while others express nondynamic relational situations (i.e., the verbs of ownership/appropriateness, (non)availability, and benefaction). Only the dynamic verbs of physical sensation, emotion, cognition, and existential experience can be said to profile the Experiencer role—or a more concrete instantiation of the same—in their semantic frames (cf. Fischer & van der Leek 1983, Anderson 1986). This group of verbs, and particularly the verbs of emotion, shows lexically inherent middle semantics (Kemmer 1993). The group of verbs denoting nondynamic relational situations tends to express highly subjectively construed possessive relations, especially when the verbs are found in impersonal use.
In terms of their type and token frequencies in impersonal patterns, verbs denoting emotional events are most prototypical of impersonal use, followed by verbs of physical sensation, cognition (especially (ge)þyncean), existential experience, and appropriateness. Verbs of (non)availability, benefaction, and motion are only marginally found in impersonal uses. Emotional event verbs thus seem to be most central for the impersonal construction in early English.
Despite their semantic differences, some common properties can be identified in the semantics of the OE impersonal verbs that facilitated their use in the impersonal construction. They are all low in transitivity; that is, the two profiled participants are not maximally opposed to each other. Although all of the verbs profile a human first participant (who might potentially be capable of having the full range of proto-agent properties), none of the profiled participants show the proto-agent properties of control and volition/intention. The proto-agent property of causation (responsibility) can be identified with only a few of the profiled participants. The SoA denoted by these verbs are thus conceived of as being only weakly initiated or even noninitated.
As especially the comparison of the impersonal and (in)transitive syntactic uses of the verbs showed, verbal meaning closely interacted with grammatical and constructional meaning in Old English. Many of the morphosyntactic idiosyncrasies shown by the OE impersonal verbs, particularly with respect to their variable argument structures, (p.231) were motivated by their lexically inherent meaning in combination with the semantic and syntactic properties of the OE nominal cases (see Plank 1981, 1983). The various options in the morphosyntactic realization of an argument depended on the conceived relationship between the participants in a concrete SoA, on the properties associated with the participant referents, particularly on the degrees of control/intention and causation exhibited by any of them over the SoA, and on the perspective taken on the SoA by the speaker. This also explains why some of the OE impersonal emotion verbs (e.g., OE (ā/ge/for)sceamian, (ā)twēonian) can show different morphosyntactic realizations for the Emoter. Depending on whether the Emoter is conceived of as being more or less intentionally involved and more or less responsible for causing the emotional SoA, it can be realized either as a nominative subject or as an acc/dat object (see Croft 1991: 219; see also Allen 1995: 146).
The interaction between lexical and grammatical or constructional meaning is not the only source of the morphosyntactic idiosyncrasies shown by impersonal verbs in the OE records. On closer examination, the textual sources of the individual syntactic instances found for these verbs reveal that many of their morphosyntactic idiosyncrasies are the result of Latin interference. Several OE verbs owe either their impersonal or their (in)transitive uses to Latin (e.g., the impersonal uses of (ge)hrēowsian or geweorðan for ‘to come to an agreement’ or the (in)transitive uses of (ge)hyngrian, þyrstan, (ge)nihtsumian, and sometimes also (ā/ge/for)sceamian). On the whole, there seems to be little systematicity in the direction of influence exhibited by a Latin translation model on the argument structure of an OE verb, but both directions, from impersonal to personal (in)transitive use as well as vice versa, are possible. This probably indicates that both syntactic uses were well established in Old English and could be used productively. It may also point to the fact that possible semantic differences between a nominative and an acc/dat argument of person were beginning to weaken in the course of the OE period, particularly with respect to the verbs that profiled an Emoter as their first participant.
The semantic and morphosyntactic analysis of the OE verbs capable of impersonal use also shows that the impersonal construction was far from being a fossilized relic in the language. In late Old English, the impersonal construction was productively extended to verbs that deviated from the semantic prototype, such as the motion verbs (ge)missan and (ge)nēalǣcan—which acquired a metaphorical meaning in impersonal use—or the ownership/motion verb becuman. In the extension of the impersonal construction to these verbs no Latin interference is evident.
The core function of the IMPacc/dat construction was to syntactically express a shift of perspective on an inherently transitive SoA by (1) suppressing, or backgrounding, the nominative subject representing the entity that controlled the SoA, and (2) foregrounding the Goal-endpoint of the SoA as the primary ‘locus of action.’ As a consequence, dynamic processes of physical sensation, emotion, cognition, and existential experience could be viewed from the perspective of the (more or less affected) human endpoint and conceptualized as being noninitiated, that is, as occurring spontaneously and unintentionally (see also McCawley 1976). The properties of foregrounding the endpoint of an inherently dynamic event and backgrounding the initiator is typical of middle-marking patterns cross-linguistically (Kemmer 1993). The OE IMPacc/dat construction thus syntactically enforced middle events.
(p.232) IMPzero patterns lacked the argument of person that denoted the human endpoint of the dynamic process and represented the sentence topic. These patterns express sentence-focus structures in that they present “all-new” information (Lambrecht 1994: 222). They expressed dynamic, spontaneously occurring events (e.g., natural phenomena or existential events of HAPPENING), which did not affect a particular, individuated Goal-endpoint. Especially in combination with verbs of existential experience, which are commonly used in narrative contexts, IMPzero patterns occur when the Goal-endpoint of the experience is retrievable from the context of the narrative.
From late Old English on, the impersonal construction (both IMPacc/dat and IMPzero patterns) acquired an extended function. In being expanded to verbs expressing nondynamic possessive relations (i.e., verbs of ownership/appropriateness and (non)availability), the impersonal construction expressed a shift of perspective from an objective to a subjectively construed possessive relation that was controlled from outside, (i.e., by the speaker). The speaker expressed his or her opinion that the Possessor (Claimee, Needer) should have (or show) the Possessed. This frequently led to the development of verb senses that expressed obligation (e.g., OE behōfian ‘to need’ 〉 ME behōven ‘to befit’). Verbs of appropriateness and (non)availability often occur in IMPzero patterns in late Old English. The argument representing the Possessor (Claimee, Needer) is omitted in these sentences, because the claim is made on behalf of speaker and addressee alike and may be retrieved from the discourse context. The textual sources suggest that the extension of the impersonal construction to verbs expressing nondynamic relational situations may have been influenced by Latin to some degree.
The comparison of impersonal with alternative syntactic uses of the OE and ME verbs showed that some constructions were formally and functionally closely related to the impersonal construction. The external(–internal) possessor construction (E(I)PC) was functionally similar to the IMPacc/dat construction, because it also foregrounded a dative argument that referred to a human participant who was conceptualized as the primarily affected entity in a change of state concerning one of its body parts. The infinitive of obligation showed some functional overlap with impersonal uses of verbs of appropriateness in the expression of obligation. Both personal and impersonal variants of the E(I)PC and the infinitive of obligation existed in Old English. They did, however, not compete with the impersonal constructions, because they were semantically too specialized. Neither the E(I)PC nor the infinitive of obligation survived long after the OE period.
The middle-reflexive construction shared the function of syntactically enforcing middle events with the IMPacc/dat construction. Middle-reflexive uses, however, appear very infrequently with the OE impersonal verbs. This syntactic alternation is expanded in Middle English, where it seems to be motivated by the model of the French middle-reflexive construction. There is, however, no evidence to confirm the assumption that the middle-reflexive construction influenced the use and functionality of the impersonal construction in Old or Middle English (cf. Seefranz-Montag 1984: 525).
Both the personal and the impersonal passive constructions showed some functional overlap with the impersonal construction in that they backgrounded the initiator (i.e., the Effector) of a transitive SoA and foregrounded the Affected or Goal-endpoint. But despite the similarities in expressing a shift of perspective, only a few OE and ME (p.233) impersonal verbs turn up in a passive construction. If they do, either the initiator becomes more salient in the background (compared with the impersonal sentences with the same verb), or the SoA receives a stative and punctual interpretation. There is no evidence in Old or Middle English that might warrant the conclusion that the passive construction influenced the use and functionality of the impersonal construction (cf. Seefranz-Montag 1984: 525). The passive construction developed as a syntactic option largely for verbs that were high in transitivity, that is, verbs that expressed SoA with two maximally opposed participants. Impersonal verbs, in contrast, are characteristically low in transitivity.
The use of the impersonal construction was clearly influenced by the rise of the hit-extraposition construction. Both IMPzero and IMPacc/dat patterns with a second clausal complement began to be replaced by personal patterns with a formal hit-subject in Old English (see also Seefranz-Montag 1984: 525, Kim 1996, 1999). The distribution of hit-extraposition in the data on OE impersonal verbs shows a diachronic as well as verb-specific picture. In early Old English, only IMPzero patterns with a clausal complement, in which the position of the acc/dat argument of person (i.e., the topic position) is empty, are replaced by hit-extraposition (see also Seefranz-Montag 1984: 528). This type of hit-extraposition is particularly frequent with verbs of existential experience, and especially with their main representative gelimpan (‘to happen’). The pattern is gradually extended to verbs of appropriateness and to isolated verbs from the other conceptual domains in the course of Old English. Hit-insertion in IMPacc/dat patterns with clausal complements (i.e., impersonal patterns where the position of the topic is filled by the acc/dat argument of person) appears only in the late OE sources. It is particularly found with the cognition verb (ge)þyncean and rarely occurs with the verbs of appropriateness and (non)availability, but it is not attested with verbs from the other conceptual domains. With these verbs the argument of person is typically a Goal-endpoint to which another entity ‘moves’ (often metaphorically) or a reference point at which another entity is located. It has none of the properties of an initiator. The replacement of these IMPacc/dat patterns by hit-extraposition may thus have been motivated by the need to syntactically disambiguate the acc/dat argument of person as the semantic endpoint of the process by moving it into the prototypical object position. From the 14th century onward, hit-extraposition seems to have become increasingly productive at the cost of impersonal patterns with clausal complements (see also Kim 1999: 340). IMPzero patterns and IMPacc/dat patterns with clausal complements thus seem to have petered out in the ME period through the competition with the hit-extraposition construction.
The investigation of the impersonal and alternative syntactic uses of the OE impersonal verbs from a semantic and functional perspective shows that, by and large, the OE system of nominal cases was functionally intact throughout the period. The OE nominal cases still showed considerable semantic functions in that they specified the relationship between the participants in an SoA and the roles they took in it. They also had prototypical syntactic functions in aligning with the grammatical relations of subject and object. It was asserted in this investigation that the distinction between the accusative and the dative case was beginning to break down in late Old English (see Allen 1995: 211–213). In early Old English the choice between accusative and dative case for the first argument in IMPacc/dat patterns was still to some degree semantically motivated, depending on (p.234) the degree of affectedness of the human participant (the accusative case being preferred over the dative particularly for the Feelers of physical sensations and the Emoters of emotional events). The data, however, show that this subtle semantic distinction was given up in late Old English, when the dative became the conventional case form for the argument of person in IMPacc/dat patterns.
The data indicate that the process of reanalysis of the dative argument of person as, or its substitution by, a nominative subject already began in the late OE period, when several verbs already appeared with nominative arguments of person. The analysis also showed that several of these nominative arguments of person were modeled on Latin, for instance with the verbs (ge)hyngrian, þyrstan, (ge)nihtsumian, and (ā/ge/for)sceamian. The interference of Latin complicates the interpretation of the data with respect to the question, when ‘reanalysis’ really started. It is however, clear from the data that the alternation between nominative and acc/dat arguments of person was restricted to a small group of OE impersonal verbs that belonged to the conceptual domains of physical sensation, emotion and cognition, and the nonavailability verb ME nēden. These verbs also for the most part retained their impersonal uses as syntactic alternatives until the late 15th century. On the basis of the data that were investigated for Middle and Early Modern English it is not possible to make any statements about the relative frequency of the two alternative constructions with these verbs, which would be necessary to estimate the rate of change from impersonal to personal (in)transitive use with these verbs.
Between 1200 and 1450 the impersonal construction, and particularly IMPacc/dat patterns, were productively expanded to about sixty-three verbs that had not been used impersonally in Old English or that were newly coined or borrowed. The ME impersonal construction was particularly productive with verbs of emotion; that is, it was extended especially in its function of syntactically enforcing middle events.
The analysis presented here suggests that the loss of the IMPacc/dat construction was a very long and gradual process in which several grammatical changes accumulated that affected the expression of transitive relations and the relationship among subject, object, and topic in early English. The impersonal construction became nonfunctional and unproductive in the late 15th century. Its fate ultimately seems to have been sealed by the loss of verb-second and the word-order changes triggered by it, most notably the loss of object-fronted word orders for topicalization. The acc/dat argument of person in IMPacc/dat patterns had properties of both prototypical subjects and objects. Its semantic endpoint or reference point properties aligned it more with the category of object, as did its morphological properties of dative, and earlier accusative, case marking. It was, however, also typically the sentence topic, a discourse-pragmatic property that aligned it more with prototypical nominative subjects and that made it appear in a syntactic position that was typical of the nominative subjects of discourse-pragmatically neutral sentence structures. When between about 1400 and 1500 the first position in the sentence was established as the syntactically fixed position of the subject, topical objects, or object-like arguments such as the impersonal argument of person, came to be disfavored. This would explain why the IMPacc/dat construction, whose functionality relied on the acceptability of object case-marked topics in verb-front position, was lost in the course of the 15th century.
(p.235) The different strategies that individual verbs followed in compensating for the demise and final loss of the IMPacc/dat construction essentially depended on their inherent semantics. Substitution of the acc/dat argument of person by a nominative subject was a path particularly followed by verbs of physical sensation, emotion, cognition, and nonavailability, presumably in analogy to semantically related verbs that had always had nominative Experiencers or Possessors by rule (e.g., OE lufian 〉 ME loven ‘to love,’ OE (ge)þencan 〉 ME thinken ‘to think, ponder,’ OE nīed habban ‘have need’). A transitive construction in which the argument of person was encoded as the direct object became the preferred syntactic choice with verbs that had rather salient Stimuli, such as OE acan 〉 ME achen ‘to ache,’ OE (ge)eglian 〉 ME eilen ‘to ail,’ ME plēsen ‘to please,’ and ME wratthen ‘to make angry.’
Particularly verbs that predominantly occurred with clausal second arguments (e.g., OE (ge)þyncean, ME sēmen ‘to seem,’ OE gelimpan, ME happen(en), OE (ge)byrian, ME behōven) developed a hit-extraposition construction. The reason the acc/dat argument of person was not reanalyzed as a nominative subject with these verbs seems to lie in the very salient endpoint or reference point properties of the participant represented by it—or rather in the absence of any initiative properties, such as intention or causation/responsibility.
By focusing on the semantic and discourse-pragmatic properties of the impersonal construction and on the interaction among verbal, grammatical, and constructional meaning, the reasons for the loss of the impersonal construction could be reassessed, pointing to the semantic and discourse-pragmatic motivations for its loss as a consequence of the drastic grammatical changes that affected the expression of transitive relations in the course of Middle English. It was further possible to show that the diachronic development of the early English impersonal verbs was not idiosyncratic but followed systematic paths of change that were motivated by the inherent semantics of the verbs in question (see also Fischer & van der Leek 1983: 365–366, Seefranz-Montag 1984: 526).
8.2 Research Agenda
The diachronic outlook taken on the further development of the impersonal construction in Middle English and the hypotheses raised with respect to the causes of its initial expansion and later loss showed that the semantic and discourse-pragmatic analysis of impersonal expressions might fruitfully be extended to the period of 1150 to 1650 in a corpus-based investigation. A corpus-based investigation would offer a larger data basis than that provided by the MED and the OED, and it could account for the different frequencies of impersonal as against alternative syntactic uses. Determining the frequencies with which impersonal uses occur with individual verbs in Middle English (i.e., expanding the type analysis offered in chapter seven into a type-and-token-analysis) might as well lead to a reevaluation of some of the hypotheses raised with respect to the diachronic development of individual verbs, the centrality or marginality of impersonal uses in their syntactic spectrum, and the overall productivity of the impersonal construction in Middle English.
(p.236) The emergence of the hit-extraposition construction as well as the process of reanalysis of the argument of person, or rather its substitution by nominative subjects, began in late Old English and proceeded throughout Middle English, as it seems gradually spreading over individual verb lexemes and semantic groups. However, most verbs that developed these syntactic uses are also recorded in impersonal patterns until the late 15th century. It may be that the impersonal construction increasingly came to be stylistically marked, for instance as literary, poetic, or archaic, and that these stylistic differences account for the syntactic alternation shown by these verbs (see, e.g., Ogura 1990: 45). A corpus-based analysis of the ME and EModE impersonal construction could systematically account for its use in different text types, consider its special textual functions and correlate this with the semantic and discourse-pragmatic properties and the perspectival function of the impersonal construction. In this context it might also be worthwhile to look in more detail into individual verb histories, which might open up the possibility to take a closer look at the textual sources for the individual instances found of a verb as was possible in the present analysis.
Regional variation could not be identified as a factor influencing the syntactic use of individual verbs as either impersonal or not in Old English. This is, however, mainly due to the nature of the OE textual sources, which allow only very limited conclusions about their regional provenance. Regional variation can be much better accounted for in the ME records, and it may be that particularly the gradualness by which the impersonal construction was lost turns out to be influenced by dialectal differences. The grammatical changes affecting the early English inflectional system and syntactic structure began in the Northern varieties and only gradually spread southward (e.g., Allen 1995, 2000, Trips 2002, Los 2009), so it is possible that the impersonal construction was also lost earlier in the North than in the South. This hypothesis, too, might be affirmed or disproven in a corpus-based analysis of the ME and EModE verbs capable of impersonal use, for instance by comparing case-impoverished texts like the 12th-century Ormulum with more southerly texts from the same period.1
(1) . I would like to thank C. Allen for pointing this out to me.