Adverb only is classified as a focusing exclusive, which main function in the discourse is to restrict the applicability of what has been mentioned about the focused sentence constituent [1]. The diachronic studies of the English language indicate that it might have originated from two sources. The first hypothesis is based on the findings from Old English (OE) records, where the form aenlic is registered and could be interpreted as Modern English only (1)- (2).
1. Ðú ðe ǽnlic eart Godes bearn (only, single) (ex. from Andrushenko 2017). 2. Ǽnlic (unicus) and ðearfa ic eom (alone, solitary) (ex. from Andrushenko 2017).
According to traditional grammars this OE adverb provided the basis for further formation of Middle English (ME) exclusive only. This approach later turned problematic considering that adverb aenlic eliminated in English records of XII and XIII (1 st half) cen. From this perspective, its origin from a different source, i.e., OE numeral one, is widely discussed in literature [2; 3; 4]. Studies centered on ME ane [5; 6] propose the scheme for only emergence and grammaticalization in ME which is given in (3).
3. num. one>polysemous Adj/Adv one > exclusive focusing onli [5].
The instances from Rissanen [3] illustrate the assumption above (4). 4. Ane þurh þet þu luuest þe God … þu makes wiþ uten oþer swink hire god tin ahne (Ancrene Riwle 151 ex. from Rissanen 1985).
Our previous research on ME exclusives shows that the positional variations of ME adverb only, taking into consideration the word it modifies, correlate with sentence information structure and Foci types [7; 8]. Hence, the data show that a post-modifying position of the adverb dominates, which is particularly evident with NPs. In contrast to this, the collocates search in British National Corpus [9] reveals that only in the position before the NP occurs more frequently (total 21,096) than when the adverb is placed in a post-modifying position (total 10,034). The same seems true for Corpus of Contemporary American English [10] (cf. [only [NOUN]] total token of 43,560 vs. [[NOUN] only] total instances of 10,307). These figures trigger the question in what period this transition in the adverb placement occurs and how it is connected with information structure of the sentence in further periods of English evolution, viz. Early Modern English (EModE).
The current study attempts to investigate the adverb when marking NPs in the corpus of Early English Books Online (EEBO) [ 5)-( 6).
The study of the NP (the only faith) in sentence (5) shows that the placement of article before only evidentiates that it functions as an adjective, taking into account the fact that a focusing adverbs cannot be modified by other adverbs or be pronominalized, thus no article can be placed before them [12, p. 168]. It is clearly seen from sentence (6) that only functions as an adverb.
To avoid the ambiguity in reading presented in examples ( 5)-( 6) a contextual analysis of EEBO Corpus should be done manually giving consideration to the major characteristics of sentence word order and peculiarities of the NP structure.
This part focuses on the major word orders prevailing in EModE, which syntax seems more rigid compared to the previous centuries on account of the gradual loss of inflectional system typical of the earlier phases of English development [13]. Furthermore, it also highlights some aspects of the NP structure at that time which differentiate it from NPs in the Present-day English.
Modern English word order forms a clear pattern already in the fourteenth century, which gets fully established by mid XV cen. Thus, according to traditional grammars the position before the verb in EModE has become "the territory of the subject in declarative actor-action-goal sentences" [14, p. 11 [17, p. 82]. A noun is used obligatory as a head, with adjectives or participles becoming prototypical modifiers of nominal elements [22, p. 47; 23]. Besides, adjectives regularly need a dummy head or a pronominal one. [24]. If the NP contains more than one adjective, the second adjective can frequently follow the noun: a rare younge man and wise [15, p. 102-103]. In this case dummy one can either be present or omitted in the sentence surface structure. The NP can also consist of two (or more) NPs that agree by means of the conjunction and; it is addressed as a single NP because it functions as one syntactic unit. A definite number of complex NPs may also contain embedded NPs. [16, p. 117].
The patterns with adverb only are retrieved from diachronic Corpus of EEBO [11] Instances with only functioning as an adverb are further analyzed with relation to the sentence IS. It presupposes the investigation of those areas of the language that facilitate speaker's considerations concerning the reader's immediate information state and to foster the communication flow [25]. The examples are tested on the basis of two oppositions: given vs. new information with further analysis of sentence Topic and Focus [see among others 26; 27 ; 28].
Given is something that is specified or reencountered (7), while new information has not been highlighted in the current linguistic surrounding (8). Cf.:
7. with the most honorable and honestest persones … men… should not be begotten of euery man, but of the most honestest men only (The lives…, ex. from EEBO).
The NP the most honestest men only in sentence ( 7) represents discourse given information, and simple search reveals that noun men was encountered in the previous sentences, NP the most honestest men also refers to NP the most honorable and honestest persones mentioned above.
8. that their priuate Parlament religion is the publike faith of christ… so in scotland not only reason persuadeth the like (A treatise…, ex. from EEBO).
The NP only reason in (8) renders discourse new information inferable from the previous discourse.
A theory of givenness distinguishes several conditions: a discourse referent being completely new is non-identifiable for the reader on account that it has not been activated in one's memory. However, it can also be interpreted as unrepresented in the memory since there is no allusion to discourse referent for an extended space of time [29].
The investigation of givenness presupposes building a file that contains entries of all the subjects, objects or events previously highlighted in the text termed discourse referents (DRs). To annotate discourse given elements it is critically important to keep track of all DRs [30]. So, the core idea is to check whether the NP initiates the presence of a new DR or not. Under the present study, information can be tagged as given if the antecedents lie within a 13-sentence window [31, p. 36].
To illustrate how givenness is annotated Haugh, Eckhoff & Welo [31] turn to Discourse Representation Theory (DRT), which shows how the reader constructs a certain mental model of the ongoing discourse. Thus, a certain mental representation (termed a discourse representation structure (DRS)) is built up by the reader as the discourse progresses. It involves two aspects: a universe of DRs and a range of DRS criteria that facilitate the information encoding [32]. In DRT the box visualization of the sentence 'A horse neighed' is assigned the manifestation as in (9). 9. The upper area in the box known as the 'universe' is able to keep track of all DRs initiated in the discourse. The lower part is composed of the 'conditions' representing the true statements made in the discourse. If the sentence 'A horse neighed' is a starting point in the discourse, the DRS in ( 9) is constructed forming the background against which the next sentence finds its interpretation. Let's assume that sentence (10) follows sentence (9).
10. It was scared.
We get a representation as in Figure 6 (11). 11. Pronoun it outlined as a DR y correlates with a discourse referent, which is alluded to in the prior context. Thus, during the next stage the merging of DRS in (9) and the ones in (11) takes place. Hence, Figure 7 (12) illustrates the outcomes of such merging operation.
12.
The new DRS comprises a universe as the combination of the universes of two DRSs that have been merged.
The methodology above was applied to annotate givenness in PROIEL corpus, known as a corpus of old Indo-European New Testament translations [30; 34]. Similar annotation methodology was also used when tagging IS categories in Tatian Corpus of Deviating Examples (T-CODEX) [35]. Tags as a result of annotation are given in Table 1.
The informational status of various linguistic expressions can also be identified in the context of topicality and focality. In the current study Topic is defined in pragmatic terms, i.e., it is "the subject of predication, what the sentence is about" [36]. The remaining part of the sentence represents a comment, i.e., something that is predicated about the Topic. Topics in their turn are subdivided into aboutness (13) and contrastive ( 14) ones [37]. The latter represent a combination of a Topic and Focus consisting of an aboutness Topic that has an embedded Focus.
Focus presents the most important or salient information in the sentence [38, p. 143]. It is further subdivided into informational, identificational, emphatic, exhaustive, contrastive and verum Foci.
Informational focus ( 15) is characterized as a part of the sentence with a great level of novelty; following Jackendoff's definition, it is information that is not shared by a speaker and a reader [39].
15. … to be kepte in a pryson … hath ete nothing but brede &; water oneli … (Thystorye of the right noble and worthy knyght…, ex. from EEBO) Contrastive focus in the present investigation is represented by the Focus employed for purely contrastive reasons [37], which means that the common ground comprises a proposition which the sentence can be contrasted against (16). 16. … if then i bring… not only reason, but testimonie also for mine assertion … (The praise of musicke …, ex. from EEBO) Identificational focus (17) expresses identity statement referring to the presence of alternatives highlighted in the previous discourse; thus, it should not necessarily have a contrastive reading [40]. 17. … this life … whether men suffer these things in this life only, or whether such iudgements follow … (Tvvo treatises written against the papistes…, ex. from EEBO). The function of emphatic focus ( 18) is to establish a scale of values, as well as, target the extreme value on the scale, signaling that the predication is either exceptional or surprising [41, p. 554 Verum focus (20) represents the sentence truth value and is characterized by placing an accent on the auxiliary [36].
20. … wythout he be the mooste proudest: only god doth knowe, iudge and open the lowe degre or lowlynesse .. (ex. from EEBO). Table 2 summarizes the tags applied to the analysis the second type of dichotomy, viz. topic/focus.
Tags Assignment in the Extended Annotation Scheme for Topic/Focus
The methodology above allows tagging sentence components in the Corpus in terms of information givenness, as well as, identify a Topic and Focus of the sentence and their variations.
Each pattern retrieved from EEBO [11] was separately analyzed based on the methodology highlighted above. Since the research presupposes both automated and manual analysis the timeframe of the current study has been limited to years of 1470s-1590-s.
] is represented by 2,438 tokens, the most frequent collocates for years 1470s and 1590s are given in Figure 8. As Figure 8 indicates the ratio of tokens changes significantly for these time frames. Despite high frequency of this model, the analysis reveals that only when pre-modifying NPs predominantly functions as an adjective (21). It is normally preceded either by definite or indefinite articles or pronouns and its functioning as an exclusive adverb is limited to 17.6 per cent of instances (22).
] makes the reading ambivalent, especially with the noun faith, since it is not quite clear whether the speaker implies by NP 'only faith' a single conceptual meaning or emphasizes on the role of faith (23). Such instances are also excluded from our investigation.
23. but if only faith doth iustifie vs … (A sermon had at Paulis, ex. from EEBO). The overall analysis of patterns indicates that when adverb only is in a pre-modifying position, the NP it marks represents more frequently emphatic (45.8%) or contrastive (34.7%) Focus. Less common Foci types registered for this model are exhaustive (9.2%), identificational (8.4%) and informational (1.9%) ones.
Investigation of word-orders where [only [NOUN]] is used shows that the most typical patterns are as follows: SV (neg) only->O1 but (neg) O1, S1Vno(t) O but (and) only->S2, it is only->S, only->SV (O), O only->S(v)V, SV only->O, (X) not only ->S1 VO but also S2, not only-> O1SV, but also O2, only-> OSV. All the models fall into two general word-order types: SVO amounting to 92.8 per cent and OSV with 7.2 per cent respectively.
This pattern is further subdivided into 3 groups: only->SVO, SV only->O and it is only->S. The analysis of only->SVO model represented by two patterns, i.e., only->SV (O), (X) not only ->S1 VO but also S2, reveals that when the adverb marks the NP that is the subject of the sentence, the subject in 78.7 per cent is tagged as emphatic Focus and given activated information (24). Topic is tagged as aboutness. The rest of the tokens contain negation not before only, thus the NP represents the subject of the sentence, which discourse function is to render contrastive Focus and discourse given or activated information (25). The NP marked by adverb only represents in 100% percent given-active information and exhaustive Focus, whilst Topic is primarily annotated as aboutness.
The NP with adverb only, which functions as the object of the sentence, is mostly tagged as contrastive Focus representing given activated information (69%). Topic therewith is marked as aboutness (27). An insignificant number of instances represent the NPs that render informational Focus and discourse new information. The rest of NPs with only in a pre-modifying position are split between sentence objects that are tagged as identificational Focus conveying given active information (18.2%) and the ones tagged as identification Focus rendering inferred information (4.8%). E.g. The notable characteristic of [only [NOUN] model is that the adverb preceding a noun is predominantly registered when the NP functions as the sentence subject (68.9%). The examples where it functions as the object of the sentence are more rarely observed due to the fact that only in these instances is frequently interpreted as an adjective in NPs.
The OVS word-order though being registered with 5.96 per cent of instances is rather controversial in terms of the analysis. Thus, the inverted word order implies that the NP with only is emphasized, therefore the noun can be tagged as emphatic Focus. However, the presence of NEG not … but relates this structure to contrastive Focus (29). Comparing the information structure of the sentences with contrastive Focus that have different word orders, viz. SVO and OVS, it was notices that with SVO order V renders discourse new information, whilst in OVS pattern it presents discourse given and activated information, which can explain the inverted word order from the information-structural perspective.
] is represented by 409 tokens, the most frequent collocates are given in Figure 10. Pattern SV neg only->O(X) is the most frequent in our data amounting to 46.6%. In all the instances only marks an NP is discourse given or situationally accessible and presents contrastive Focus (31). Less frequent Focus types registered with this pattern are identificational (given or inferred information with topic marked as aboutness) running to 3.9 per cent (33), identificational (given information with topic marked as contrastive) amounting to 0.6% (34), and verum (given information) found in 1.1 percent of examples analyzed (35).
Figure 11 shows the frequency of various Focus types and their informational status in the discourse.
Instances of [[NOUN] only] in EEBO Corpus amount to 1860 tokens. Figure 12 illustrates the most frequent collocates with the adverb for years 1470s and 1590s.
When marking a sentence subject adverb only highlights NPs that render given-active information and emphatic Focus in 73.7 per cent of instances (37); examples of given-active information, contrastive Focus are represented by 18.1 per cent (38). The most uncommon instances are the ones where NPs highlight given-active information and verum Focus (39) accounting to 3.1 per cent.
The most common NPs with the adverb in a post-modifying position represent given-active information and either identificational Focus (41) or contrastive Focus (42), registered in 55.7% and 28.2% of instances respectively. 41. … ad be by special vocatio ministers … , which ministratio ad dispesatio is to be foud in the church only… (A reioindre to M. Iewels…, ex. from EEBO).
To less frequent Foci types with NPs marked by the adverb belong situationally accessible information, identificational Focus (43) or inferable, informational Focus (44), acc-sit, emphatic Focus (45), as well as new information and contrastive Focus (46).
The study of less frequent word order patters where post-modifying only is registered as a part of NPs, i.e., OSV, SOV and O(X)VS, indicates that the NPs regularly represent given inactive information and mostly contrastive Focus (47) or emphatic one (48).
The investigation of tokens with exclusive only as a part of NPs shows that its placing after the noun is more preferable, particularly in the instances where the NP functions as the sentence object. We may assume that this deliberate positioning allows the addressee to avoid ambiguous reading of only as an adverb or an adjective. Still 31.1 per cent of instances with only preceding the noun-object indicate that it demonstrates the tendency to pre-modify the noun in EModE, which syntax is more rigid in terms of word-order. It is also notable that with the NP functioning as the subject of the sentence, only is more likely to be found when it pre-modifies the noun. We assume that this pattern might have triggered the dominant placement of the adverb before the noun functioning as the object of the sentence in further centuries.
Analyzing different word-orders registered with the adverb as a part of the NPs, it was observed that different Foci types correlate with various sentence components as far as the most widespread SVO pattern is concerned. Thus, it is more preferable to allocate NPs with only in the subject position while marking emphatic and contrastive Foci. NPs, representing a sentence object, are more characteristic when they highlight identificational Focus for models [[NOUN] only] and informational Focus for models [only [PREP] [NOUN]]. The investigation of the minor word-orders, viz. SOV, OSV, OVS, shows that the verb regularly represents given information, while NPs with only, though conveying given-activated information, mark either contrastive or emphatic Focus and facilitate an inverted wordorder in the sentence for the purpose of emphasis.