The Suffix -ation in English Pius ten Hacken and Renáta Panocová 1 This paper analyses the role of the suffix -ation in English, based on data from the OED. The suffix forms deverbal nouns. Its origin can be traced back to Latin, but French had an important part in it, too. For the analysis, nouns in -ation and their corresponding verbs were retrieved from the OED. The purpose of the analysis was to characterize the position of -ation in the mental lexicon of current speakers of English and to assess the role of the word formation rule in the history of the formation of nouns in -ation. For the analysis, Marchand’s classification of nouns in -ation is used, which distinguishes nouns with verbs ending in -ify, -ize, -ate, and other verbs. This classification is combined with the information about first attestation dates provided by the OED. Cases where more than one verb can underlie a particular noun are studied in detail. Particular attention is paid to the issue of backformation, i.e. cases where the noun has been attested earlier than the verb. In the interpretation of these cases, the perspective of individual speakers is highlighted. 1. The Suffix -ation In English, the suffix -ation can be used to derive a noun from a verb. It is an example of what Bauer et al. (2013: 196) call “non-native affixation”. They call it “the only one of the non-native nominalizing affixes that displays real productivity in modern English” (2013: 201). The English suffix correlates with suffixes in a range of other languages, as illustrated for organization in (1). 1 This work was supported by the Scientific Grant Agency of the Ministry of Education, Science, Research and Sport of the Slovak Republic and Slovak Academy of Sciences VEGA under the project No VEGA 1/ 0130/ 21. AAA - Arbeiten aus Anglistik und Amerikanistik Band 47 · Heft 1 Gunter Narr Verlag Tübingen DOI 10.24053/ AAA-2022-0002 Pius ten Hacken and Renáta Panocová 30 (1) a. organization EN b. organisation FR, DE, DK c. organización ES d. organizzazione IT e. organizaç-o PT f. organisatie NL g. organizácia SK h. organizacja PL i. организация RU In (1b-i), equivalents in three Germanic, four Romance and three Slavic languages are represented orthographically. In (1b), the same spelling corresponds to different pronunciations in three languages. Words such as the ones in (1) are often called internationalisms. Many words with the suffix -ation fall into this category. The origin of the suffix can be traced back to Latin. However, what appears as -ation in English was the result of a number of separate morphological processes in Latin. The forms in (2) can be used to explain this for the verb laudare (‘praise’). (2) a. laudare infinitive b. laudatus passive participle c. laudatio noun, nominative singular d. laudationem noun, accusative singular The infinitive in (2a) shows the thematic vowel -aof the first conjugation. The -tis added in (2b) as a marker of what Aronoff (1994) calls the “third stem”, which is traditionally indicated with the passive participle. The noun-forming suffix in Latin is -ion. It is attached to the third stem and reduced to -io in the nominative singular (2c). In oblique cases, such as the accusative in (2d), the full form is visible. In Romance languages, nouns were generally adopted in the accusative form, which is reflected in the spelling of (1b-d). The suffix -ation in English and in the other languages illustrated in (1) does not reflect the Latin analysis in (2). It is a combination of elements that in Latin express both inflectional and derivational information. In particular, the thematic vowel marks the first conjugation, which was the largest in Latin, but by no means the only one from which nouns of this type were formed. Some of these other nouns were also borrowed in English, as illustrated in (3). (3) a. definīre definitio ‘delimit’ b. deducĕre deductio ‘deduce’ c. defendĕre defensio ‘defend’ d. complēre completio ‘fulfil’ The Suffix -ation in English 31 In (3), the Latin infinitive is followed by the nominalization in -io and the English translation of the verb. In (3a), we have an example of a verb from the second-largest conjugation class, traditionally the fourth conjugation, which has a thematic vowel -i-. The third conjugation does not have a thematic vowel. Two examples are (3b-c). The consonant marking the third stem, written -tin (2c) and (3a-b), was generally pronounced as an affricative before -io. In (3c), it is reduced to a sibilant. Similar phonological processes can also be observed for some of the forms in (1). The much smaller second conjugation class, which has the thematic vowel -e-, is illustrated in (3d). Forms such as the nouns in (3) were representatives of the same sequence of morphological processes as illustrated in (2) and they were also borrowed by many of the languages in (1), but in smaller numbers. Bauer et al. (2013: 201) recognize “a number of different variants” of the suffix on this basis, but state that the string -ation “is common to all productive variants”. This implies that they assume variants such as -ition in definition are not productive. Semantically, the Latin suffix -io does not necessarily affect the base it attaches to. In English, we can observe the same, as illustrated in (4). (4) a. It took several months to organize the event. b. Organizing the event took several months. c. The organization of the event took several months. The three sentences in (4) express the same meaning. In (4a), an infinitive is used and it is more natural with an extraposition of the infinitival clause. The gerund in (4b) can be recognized as a verb form because it has a direct object and cannot have an article. The nominalization in (4c) requires a preposition and an article. The process forming organization in (4c) is a transposition in the sense of Dokulil (1982), because it changes the syntactic category without changing the meaning. As illustrated in (5), it is also possible for the meaning to shift. (5) a. The temple was at the very centre of the organization of the Minoan economy. b. Each organization or agency has its own lending criteria. The examples in (5) are modified from BNC (2007). In (5a), it is not the process of organizing the economy, but the result of this process that is addressed. This process-result alternation is quite regular with nouns in -ation. A further derived meaning is illustrated in (5b). Here the noun refers to the body that constitutes the result of organizing. Such further derived meanings are not at all uncommon, but slightly less regular than the result reading of the type in (5a). Ten Hacken (2015) proposes a rule-based account of the relationships between these reading types. In this paper we will not be concerned by these semantic alternations. Pius ten Hacken and Renáta Panocová 32 Our purpose in this paper is to explore the position of -ation in the English morphological system. We are interested on one hand in its place in the contemporary language, i.e. in the language system of current speakers of English, on the other in its historical development. The data we took as a basis for both aspects of our exploration are taken from the OED. We used a list of nouns in -ation collected and analysed by Thomas (2013) as our starting point. She collected nouns from OED (2000-2012), using as a criterion that the noun and the corresponding verb are both in current use. This excludes cases such as comparation and obstipation, where the noun or the verb are obsolete. She also excluded complex words such as cell degradation and counter-accusation, where there is no full verb cell degrade or counter-accuse in the dictionary. This produced a list of 2941 nouns in -ation. Some further operations were carried out on this list with OED (2000-2020) as a source of information, which reduced the list to 2927 items. 2 In the study of the current place of -ation in the language, we make use of Marchand’s (1969) division into four classes (section 2). Then we turn to the history of -ation on the basis of first attestation dates in OED (section 3). In section 4, we address a number of questions that arise in the interpretation of these data. On this basis, section 5 brings together the insights gained from the analysis Marchand’s classes and first attestation dates. Section 6 summarizes the conclusions. 2. Marchand’s Four Classes Marchand’s (1969: 259-261) discussion of -ation organizes the data into four classes, illustrated in Table 1. Class Noun Verb I modification modify II neutralization neutralize III translation translate IV expectation expect Table 1. Four classes of nouns in -ation distinguished by Marchand (1969: 259) What we present here as classes of nouns in -ation is introduced by Marchand by the phrase “Formally we can distinguish four groups” (1969: 259). The use of group indicates that the division may be intended more as 2 We would like to thank Claire Thomas for allowing us to use her data and Irene Jiménez Alonso for recording further information from OED about the nouns and the verbs. The Suffix -ation in English 33 a presentational device than as a theoretical classification, but we will argue that the distinction is of enough morphological interest to deserve further study. The use of formally may suggest that it is the form of the noun that determines the class, but in fact it is the form of the verb that is crucial. The distinction between Classes III and IV is not visible in the noun. The criterion for the distinction between the classes in Table 1 is the ending of the verb. We use ending rather than suffix for reasons that become obvious when we consider the examples in Table 1 more closely. In neutralize, -ize is a suffix, because it attaches to neutral. In modify, however, -ify cannot be analysed as an English suffix, because it attaches to a base that is not a lexical item in English. Historically, modify and modification are both borrowings from French. In fact, according to OED neutralize is also a borrowing from French, but in this case the existence of neutral as an adjective in English makes a reanalysis possible. Whether neutralize is a borrowing or a verb formation based on neutral depends on the individual speaker. The significance of the classes in Table 1 derives in part from their Latin origin, which determines their role in the etymology of the words in -ation. This etymological basis means that the same classes can be found in different languages. Thus, Panocová (2017) uses them in her analysis of Slovak nouns in -ácia. In Class I, the Latin base of the verb ends in -ficare, which is an intensive form of facĕre (‘make’). Italian has the verb modificare, which preserves the stop. In German, it is modifizieren, where the stop has become an affricative. The form of the English verb shows the influence of French, where the stop was elided in modifier. In Class II, the suffix -ize is of Greek origin, which is also the reason why OED writes it with a <z>. This suffix appears in many languages. For English, Marchand (1969: 318-321) observes that, after a surge in formations between 1580 and 1700, it can be considered an English suffix in the sense that new English verbs in -ize do not depend on borrowing. Durkin (2014: 310-311) identifies the 17 th century as the peak of Latin borrowing. The suffix attaches to adjectives and nouns and has a range of related meanings that can be described as ‘make X, convert into X, adapt to X’. Although there is a sense that the base tends to be [+learned], Marchand (1969: 319) gives several examples with [-learned] bases, e.g. standardize, winterize. The ending -ate in Class III is often not a genuine suffix. Its origin is the Latin passive participle illustrated in (2b). Durkin (2014: 320-323) observes that the pattern of these verbs developed from borrowing of what he calls the Latin past participle. Marchand calls the use of -ate in cases such as translate “adaptational” (1969: 256) and mentions backformation as an important source of these verbs (1969: 256-258). In the case of translate, the actual Latin verb is transferre (‘carry across’), with a prefix attached to ferre (‘carry’), which has an irregular passive participle latus. Although transferre has given rise to English transfer, the link between transfer and Pius ten Hacken and Renáta Panocová 34 translate is not generally made by speakers of English. Indeed, the emergence of translate suggests that the speakers involved did not know the Latin verb transferre or did not relate it to the noun translation. From a Latin perspective, Class IV is the unmarked class. The Latin verb corresponding to expect is expectare. When the 2927 pairs of nouns and verbs both of which are in current use are divided up according to the classes in Table 1, we find the distribution in Figure 1. Figure 1. Distribution of Marchand’s (1969) classes of -ation nouns In Figure 1, the absolute number of types is given for each class. From the shape of the noun, it is generally possible to count Classes I and II and distinguish them from the other two classes. We still have to check the actual verb, because the ending -ization is no proof that there is a corresponding verb in -ize. Thus, for solmization, the verb is not *solmize, but solmizate. However, this only concerns a small number of nouns. For the distinction between Classes III and IV, the form of the noun does not provide any guidance. As Figure 1 shows, there are a number of nouns for which OED gives both a verb with and one without -ate. For these nouns, the question arises which class they actually belong to. We will come back to this question in section 4.1. The Suffix -ation in English 35 3. The History of -ation One of the points on which OED generally puts a lot of emphasis is recording the first date of attestation. As described by Gilliver (2016: 271-272), antedatings were regarded as particularly urgent in the revision process, even at the stage of proofreading. Although Durkin (2014: 336-340) emphasizes that the history of a word is also determined by its increase in frequency, which may occur only much later than its first attestation, the first date of attestation of words remains an anchorpoint in their lexicographic description. In interpreting any analysis based on such information, it is important to be aware of what exactly a date of first attestation means. Let us take as an example nationalization. OED (2000-2020 [2003]: nationalization) 3 gives three senses, the second of which is split into two subsenses, as described in (6). (6) a. The action of making national in character, or of making distinctively national. b. The action of forming into a nation or nations; the process of becoming a nation. c. Inclusion or absorption into a nation. d. The action of bringing land, property, an industry, etc., under state ownership or control. Sense 1, described in (6a), is marked obsolete. Sense 2, with the subsenses in (6b-c) is marked rare. Sense 3, defined as in (6d) is unmarked. For each of the senses, a number of dated examples are given. The first example for (6a) is from 1801, the other senses have examples starting from 1813, 1885 and 1847, respectively. The question is, then, what we can say about the date of the formation of nationalization. A first observation is that we can be sure that the word has been used at least from 1801. Whether it was used earlier is not known. Not all occurrences of a word are archived in a way that they can be retrieved. A second point to be made, however, concerns the relationship between the use of a word and the existence of the word. As English does not refer to an empirical entity, there is no purely empirical way to verify the existence of a word in English. 4 Therefore, we must admit that the conclusion that nationalization has existed as a word of English since 1801 is the result of a chain of assumptions. 3 Where relevant, we give in square brackets the year of the last full update of an OED entry. 4 The truth of the statement that nationalization exists as a word of English depends on a range of decisions concerning what constitutes English. When we find a word in a text, we first have to decide whether the text is English, then whether the word is not an error. Even in an English text, not all words are English words. An English text can, for instance, contain a French or Latin quotation. This does not make the Pius ten Hacken and Renáta Panocová 36 As we are interested in the date of the formation of the word, we have to make a further step, connecting the existence of nationalization to its formation. As discussed by ten Hacken and Panocová (2011), the formation of a word is something only individual speakers can do. Texts do not record the meaning of a word, so that the existence of a word in a text is only partial. The interpretation depends on a speaker attaching a meaning to the word. In principle, each speaker who knows nationalization has their own meaning attached to it, realized in their mental lexicon. When they come across the word for the first time, they have to form the word as an entry in their mental lexicon. What we usually mean by “the formation of nationalization”, however, is not the extension of one’s mental lexicon by a word on the basis of receptive understanding, but rather setting up the word for use in spoken or written language without hearing or reading it first. This is an action that is much rarer than the reception-based extension of the lexicon, but it must have occurred at least once. It can have occurred multiple times if speakers were not aware of the word having been used by others before. Taking these considerations into account, any analysis of the history of a set of words on the basis of first attestation dates in OED is based on a series of abstractions. Nevertheless, the information about first attestation gives evidence on the degree of prominence of a word formation process among speakers in a particular historical speech community. Bauer et al. (2013: 195-206) discuss a range of competing word formation processes corresponding semantically to -ation, e.g. -al and -ment. Which one is actually used by speakers depends not only on the relative prominence of these word formation processes, which may vary from one century to the next, but also on factors such as which other words had been formed recently or were established in the speaker’s vocabulary. In our calculations, we have assumed that first attestation dates as given in OED are significant. We have taken all senses of a word together and adopted the oldest date as the relevant one for the word. In the case of nationalization, we have used 1801, even though the sense attested in that year, given in (6a), is now obsolete. We have also normalized years. OED uses abbreviations where the year of a text is unsure. As Marsden (2013) describes, for Middle English there is a system using brackets to distinguish date of composition and date of manuscript, combined with abbreviations c for circa and a for ante and question marks where dating is unsure. The use of this system is not restricted to Middle English, as shown by the first attestation date of “? a1793” for recompilation. In a similar vein, sometimes a range of years is given, e.g. 1687-1700 for hibernation, or a decade or century, e.g. “138.” for cavillation, “16..” for disseveration. For our calculations, we have taken words in the quotation English words. The need for such decisions implies that, as Uriagereka (1998: 27) put it, “English does not really exist”. The Suffix -ation in English 37 away the abbreviations and question marks and taken the middle year for any range, decade or century, rounding down if necessary. This means we have adopted 1793 for recompilation, 1693 for hibernation, 1385 for cavillation and 1650 for disseveration. 5 As a result of these operations, we can represent the historical development of the use of -ation as in Figure 2. Figure 2. Distribution of first attestation dates for nouns in -ation The data we used for Figure 2 are not the same set as the one we used for Figure 1. Figure 1 takes a synchronic outlook, including only those verbnoun pairs where both are in current use. This is reasonable if one is interested in the mental lexicon of current speakers, but is problematic if the purpose is to outline the historical development of the word formation process. Therefore, we extended the data set to include those cases where one of the two is now obsolete. We did not include nouns in -ation for which no corresponding verb is recorded in OED. The result is a set of 3947 nouns. As shown in Figure 2, the earliest attested cases of -ation nouns are from the 13 th century, but it is only in the 14 th century that we find a substantial number of them. This is the period when the renewed use of English in legal and administrative domains, previously dominated by French and Latin, led to many loanwords from these languages. As Durkin (2014: 236- 5 In cases where we had to translate “OE” into a year, we have adopted 900. This does not apply to nouns in -ation, but occasionally to verbs from which such nouns have been formed, e.g. starve for starvation. 0 200 400 600 800 1000 1200 Pius ten Hacken and Renáta Panocová 38 240) shows, it is often difficult to distinguish between French and Latin as the source of a loanword. The increase in new nouns in -ation continues rapidly until the 17 th century, then drops suddenly in the 18 th century. In the 19 th century, the formation of new words resumes and for the 20 th century, fewer new words are recorded. An important question is to what extent these fluctuations are due to the way the OED was compiled and to what extent they reflect genuine developments in English usage. In his study of borrowings from French and Latin, Durkin (2014: 310-315) finds similar fluctuations for the period up to the 20 th century as we do here and he argues that they are not due to the OED compilation process. He attributes the rise in the 17 th century and the fall in the 18 th century to a change in writing style. The renewed rise in the 19 th century concerns mainly technical and scientific vocabulary and is caused by significant advances in these domains. 6 The lower number for the 20 th century can be explained by the history of work on OED. As described by Gilliver (2016), the first edition was compiled in the period 1858-1928, with a supplement to update the earliest parts completed in 1933. After the completion of the first edition and the supplement, work on updates was tuned down to such an extent that Gilliver (2016: 453) characterizes Robert Burchfield’s task as a new editor in 1957 as “Learning to swim (again)”. In order to gain an impression of how special the development of -ation as depicted in Figure 3 is, one might compare it to the numbers for -ee in Mühleisen’s (2010) study. Table 2 compares the two suffixes. 6 A different position on this issue is taken by Brewer (2012). She argues that the 18 th century is underrepresented in OED because of “systematic under-reading” (2012: 87), so that first attestations of words would be placed in the 19 th century even though more extensive reading of 18 th century sources would have given earlier attestation dates. In our argumentation we follow Durkin (2014). The Suffix -ation in English 39 Period -ation -ee Proportion 1200 - 1299 10 1300 - 1399 176 4 44.00 1400 - 1499 315 12 26.25 1500 - 1599 527 21 25.10 1600 - 1699 1020 47 21.70 1700 - 1799 300 40 7.50 1800 - 1899 961 152 6.32 1900 - 1999 638 68 9.38 Table 2. Comparison of the number of formations with -ation and -ee Mühleisen’s (2010: 107) figures are used for the third column in Table 2, but for the 20 th century we give the number of OED entries she mentions in the footnote in order to make the data comparable. The last column gives the number of times the -ation nouns exceed the -ee nouns. Obviously, -ation is much more common. However, the reduction of new cases in the 18 th century is far stronger for -ation and the proportion remains similar afterwards. The relative strength of -ation can also be assessed by comparing the number of new formations in -ation with the number of new nouns recorded by OED. Figure 3 gives the proportion as a percentage. Figure 3. Number of new words in -ation as a percentage of new nouns in OED 0 0,5 1 1,5 2 2,5 3 3,5 4 Pius ten Hacken and Renáta Panocová 40 In line with the data in Table 2, Figure 3 highlights the reduction of the number of new entries with -ation after the 17 th century. This may be connected to Durkin’s (2014: 307-308) observation that the ornate writing style of the 17 th century gave way to a more restrained one in the 18 th century if we assume that nouns in -ation are more typical of this ornate writing style than other nouns. We also see that the drop in numbers for the 20 th century in Figure 2 is a result of the history of the dictionary. In relative numbers, Figure 3 actually shows a slight increase in this period. 4. Questions of Priority One of our main objectives can be formulated as relating the distribution of Figure 1 to the historical development in Figure 2 and Figure 3. This will provide a basis for the evaluation of claims such as Bauer et al.’s (2013: 196) that the productivity of -ation in contemporary English exists only “by virtue of prior affixation”. As Class IV does not involve affixation in the formation of the verb, this would predict that this class is no longer productive. Furthermore, it raises the question of the position of -ate in Class III. If the verb is a backformation, it can hardly count as a case of prior affixation. The questions of priority that arise for nouns in -ation are of two types. One type concerns the case when we have two competing verbs. This is particularly relevant for Classes III and IV. The other type concerns the priority relationship between the verb and the noun. The question is then whether the noun is actually derived from the verb or the verb is a backformation based on the noun. The extensive information about first attestation dates in OED provides a rich source of information for studying these questions, even though the interpretation of this information is subject to the considerations in section 2. On the basis of this information, we will first address the distinction between Classes III and IV in section 4.1, then turn to the issue of backformation in section 4.2. 4.1. The Distinction between Classes III and IV As indicated in Figure 1, we have 247 nouns in our collection for which OED gives both a corresponding verb with -ate and one without. This is the set of nouns for which both the noun and at least one of the verbs are in current use. An example is devaluation, which has both devaluate and devalue as possible verbs. One can then ask the question which class such nouns actually belong to. The central cause of the difficulty in answering such a question is that asking the question of actual class membership implies that words, classes and the language of English are empirical entities for which properties can The Suffix -ation in English 41 be determined observationally. However, as mentioned in section 3, this is not the case. Empirical entities are a speaker’s competence and a speaker’s performance. In the performance, words can be realized, but classes are not. Classes are part of a system. Systems of this type are not realized in performance, but only in a speaker’s competence. Competence is realized in individual speakers and it is not the same for different speakers. Of course speakers are in contact with each other and influence each other’s competence through this contact, but the nature of this influence is not absolute and automatic. Let us assume that Charles and Deborah are both speakers of English. They both know the word devaluation. Charles links it to the verb devaluate, but Deborah to the verb devalue. It is perfectly possible that Charles and Deborah never meet and never read each other’s writings so that they do not come across each other’s use of different verbs corresponding to devaluation. Even if they communicate, it may well be that neither uses either of the verbs. In this case they do not influence each other. For the classification of devaluation in English, this means that both Class III and Class IV are correct, but for different speakers. If Charles and Deborah come across each other’s verbs corresponding to devaluation, they will perceive the other speaker’s verb as unusual. How this perception is subsequently processed depends on their attitudes to language and their relative social position. If Charles is Deborah’s teacher, he may correct her. If they are friends, they may simply add each other’s verb to their mental lexicon as an entry for passive or even active use. This means that devaluation may belong to both classes even for a single speaker. Although it is possible that a word belongs to both classes, many speakers perceive such a situation as undesirable. In language, there is a significant community pressure towards a standard. This means that when a word is less common than a synonym, especially if the two are formally very similar, the less common word will tend to fall out of use. In our data, we found evidence of this phenomenon for many nouns in -ation with both Class III and Class IV verbs. A first example we want to consider is elicitation. OED gives both elicitate and elicit as verbs. For elicitate, there is one example only, dated 1642, and the verb is marked as obsolete. In such a case, we can assume that elicitation belongs to Class IV for present-day speakers of English. A second example is excavation. Again, OED gives both excavate and excave. The latter is a cognate of French excaver, but it is marked as “rare” by OED (2000-2020 [1894]). In view of the compilation date of 1894, the example from 1884 clearly excludes a categorization as obsolete. Nevertheless, we can safely assume that there are few speakers of English for whom excavation will not be Class III. When we extend this type of reasoning to all 247 nouns in this class, we get the distribution in Figure 4. Pius ten Hacken and Renáta Panocová 42 Figure 4. Current use of verbs for nouns in -ation for which OED gives verbs according to both Class III and Class IV with absolute numbers Although the class of nouns in -ation for which both the verb with -ate and the one without are in common use is the largest in Figure 4, more than half of the cases can be attributed to either Class III or Class IV, because the other verb is marked as rare or obsolete. In a few individual cases, one of the verbs is regionally marked (mostly Scottish) or both verbs are marked as rare or obsolete. For the nouns that have two corresponding verbs in common use, one of the ways to determine whether they are primarily in Class III or in Class IV is to consider which verb is older. When we consider, for instance, permutation, both permutate and permute have an entry in OED and they are fairly similar in frequency. 7 However, permutate has a first attestation date of 1598 and for permute the oldest example is dated c1400 (OED 2000- 2020 [2005]: permutate and permute). The noun permutation has a first attestation date of a1398. This makes it unlikely that permutation was formed on the basis of permutate. It is important to realize that we are now making a rather different point about class membership than in the classification in Figure 4. In Figure 4, classes were considered properties of current speakers of English. When we consider first attestation dates, we turn our attention to the point where 7 OED assigns permute to Frequency Band 4 (“most words remain recognizable to English-speakers, and are likely be used unproblematically in fiction or journalism”) and permutate to Frequency Band 3 (“These words are not commonly found in general text types like novels and newspapers, but at the same they are not overly opaque or obscure”). Class III only; 58 Both in use; 100 Class IV only; 81 Regional; 5 Neither; 3 The Suffix -ation in English 43 permutation was coined. The question is then whether speakers who did not hear permutation from others used permutate or permute as a basis. 8 The difference of about 200 years in first attestation date makes it likely that the base verb was permute. In order to elaborate this point, we compared first attestation dates for all verbs corresponding to nouns in -ation that have two verbs in current use. The result is represented in Figure 5. Figure 5. Comparison of first attestion dates of verbs for Class III and Class IV Figure 5 uses classes based on the number of years one verb has an earlier first attestation date than the other and gives absolute numbers for each class. In calculating the data for Figure 5, we prepared the OED information about first attestations in the same way as we did for nouns in section 3. As there are exactly 100 nouns in the class covered in Figure 5, the absolute numbers are identical to percentages. Figure 5 shows a clear tendency for the verb without -ate to be earlier than the verb with -ate. For more than 80% of the nouns, the verb making the noun a member of 8 It should be noted here that permutation can also be a borrowing, so that it is not necessary that either verb is the base for the formation of the noun. We return to this issue in section 4.2. IV earlier 101+; 61 IV earlier 31- 100; 16 IV earlier 11- 30; 4 Almost equal; 5 III earlier 11- 30; 5 III earlier 31-100; 6 III earlier 101+; 3 Pius ten Hacken and Renáta Panocová 44 Class IV is attested more than ten years earlier than the verb in -ate. Moreover, within this set of nouns, the largest group has a difference of more than a century. The data in Figure 4 and those in Figure 5 are of a different kind, so that it is not legitimate to combine them directly. Whereas Figure 4 represents information about the synchronic state of language in modern speakers, Figure 5 targets the historical situation at the point when words were formed. However, both point to a stronger position of the verb which makes the noun a member of Class IV. In relation to Figure 1, we can therefore say that most of the nouns in -ation that are ambiguous between Classes III and IV should rather be attributed to the latter. This does not change the fact that Class III is the largest class, but it is clearly less than half of the total. 4.2. Backformation A word formation rule derives a word from a base. In affixation, the derived form is the result of adding an affix to the base. In backformation, the direction of the formation process is reversed. We still have one form with an affix and one form without, but now the form with the affix is the starting point and the form without the result of the process. A well-known example discussed by Marchand (1969: 391-392) is the case of the verb peddle and the noun peddler. As Marchand notes, there is a difference between the synchronic and the diachronic perspective. Synchronically, for a modern speaker, peddler refers to a type of person and peddle to the corresponding action so that the relation between them is similar to the one between driver and drive. In both cases, the suffix derives the agent noun from the verb. Diachronically, it is unlikely that such a derivation actually took place. OED (2000-2020 [2005]: peddler and peddle) gives 1650 as the first attestation of the verb, 1307 as the first attestation of the noun. 9 As it is unlikely that the verb was used for such a long time without appearing in any of the material covered by OED, we can assume that in an earlier stage only the noun existed. A question raised by these observations is how the diachronic relationship can be accounted for. Marchand (1969: 391) only mentions that peddle is “historically an extraction from peddler”. Štekauer (2015: 342) ascribes to the earlier edition of Marchand’s work the idea that it “is arrived at through zero-derivation and the subsequent dropping of the (pseudo)morpheme”. In our view, a more plausible analysis of the process is that peddler was at some point reanalysed as involving the suffix -er. In fact, Marchand (1969: 392) already refers to “the change from moneme to syntagma”. 9 Marchand (1969: 391) gives different dates, based on an earlier edition of OED. Marchand refers to the process as backderivation. The Suffix -ation in English 45 Kiparsky (1982) discusses this reanalysis process for a number of items. Bauer (2001: 93) links this kind of reanalysis to analogy. In the case of -ation, word formation interacts with borrowing. This means that there are in principle always two possible origins for a noun in -ation. It can be a borrowing from Latin or French or it can be the output of the word formation rule taking a verb as its base and adding the suffix -ation. This constellation increases the potential for backformation. Thus, Marchand (1969: 260) notes that for nouns in -ation “which go with verbs in -ate”, i.e. Class III, the noun is “as a rule, older than the verb”. This implies that there are many verb-noun pairs in which the verb is a backformation. On the basis of the first attestation dates from OED, it is possible to analyse how general the phenomenon of backformation is from a diachronic perspective. Figure 6 divides all -ation nouns we collected according to the difference between the first attestation date for the noun and the verb. Figure 6. Comparison of first attestation dates of nouns in -ation and corresponding verbs The classes in Figure 6 are analogous to the ones in Figure 5, but we divided the class marked Almost equal into three classes, because here the larger number of pairs justifies such a division. In calculating the values Verb > 100 earlier; 815 Verb 31-100 earlier; 699 Verb 11-30 earlier; 411 Verb 1-10 earlier; 334 Verb and noun equal; 387 Noun 1-10 earlier; 287 Noun 11-30 earlier; 285 Noun 31- 100 earlier; 388 Noun > 100 earlier; 345 Pius ten Hacken and Renáta Panocová 46 for Figure 6, we took the verb with the earliest attestation date if more than one verb occurs, in particular in the case of ambiguity between Classes III and IV. Because of the historical perspective, we included cases with an obsolete noun or verb. As in the case of Figure 5, the greater the divergence of the dates of first attestation, the more confident we can be that the order of the formation of the words is represented correctly. Figure 6 shows a clear bias towards the verb being earlier, because more than half of the pairs (57.2%) have an earlier first attestation date for the verb, but there is still almost a third of the pairs (33.0%) where the noun is earlier. For a more precise discussion of the interpretation of these data, we will use the examples in (7) and (8), giving dates of first attestation from OED. (7) a. transplant c1440 b. transplantation 1601 (8) a. deviate a1634 b. deviation 1603 In (7) we have an example where the verb is more than 100 years earlier, in (8) one where the noun is in the range of 31-100 years earlier. Both nouns appeared in the first decade of the 17 th century. The words in (7) and (8) are all common words of English, generally known by fluent speakers of the language. OED (2000-2020) assigns them to frequency band 5, which is described as “literate vocabulary associated with educated discourse”, except for (8b), which is in frequency band 6 (i.e. more frequent). Data from COCA (2008-2020) and BNC (2007) indicate that in both pairs the noun is significantly more frequent than the verb, a factor 4-5 in BNC. Let us now consider a child, Helen, growing up in the 21 st century and at an age where a child’s vocabulary is typically expanded with words such as (7) and (8). If Helen comes across the verb in (7a) she will not know what the corresponding noun is until she encounters it. For some other verbs, there is no noun in -ation, e.g. *transferation. Conversely, if she encounters the noun in (7b) first, which is more likely in view of the frequency data, she will not know whether the verb is transplant or *transplantate until she comes across an example. Thus, although she can connect the two as soon as she knows them, she will have to wait for them to appear in her environment or risk producing a word that nobody else uses. In the case of (8), the situation is slightly different. On the basis of the verb in (8a), Helen can predict the noun in (8b), because the verb has -ate in the ending. The years of first attestation in (7) and (8) have nothing to say about the position of these words in the mental lexicon of current speakers. They give information about the considerations of the speakers who first used these words. Let us therefore consider the situation of Isaac, a well-educated English speaker of the early 17 th century who is involved in scientific The Suffix -ation in English 47 activity and for this reason often has to speak and write about concepts for which he does not know a word yet. For such a speaker, it will not be implausible that he comes up with the nouns in (7b) and (8b) at some point. As shown in Figure 2, the 17 th century was a time when many new nouns in -ation were first attested, so that the word formation rule will have been active for many speakers. The question is, then, on what basis Isaac comes up with transplantation and deviation as new words. The first-attestation dates in (7) indicate that it is highly unlikely that Isaac did not know the verb in (7a) when coming up with the noun in (7b) as a new word. In the case of (8), a more plausible scenario is that Isaac used deviation as a borrowing from French or Latin. For (8a), there is no direct parallel in these languages, so that it is most likely a backformation. This presupposes that Isaac already knew enough Class III pairs to create a pattern. When we speak about backformation and use first-attestation dates as evidence, it is the point at which a particular speaker uses a word without having heard or read it first that we address. In (7), we can be confident that there was no backformation at this point. In (8), the difference of 31 years should make us more cautious compared to the much longer distance in (7), but backformation is a plausible hypothesis. The discussion of (7) and (8) suggests that the issue of backformation may also depend on the class from Table 1 a pair belongs to. In order to test this hypothesis, we calculated the proportions for each of the classes separately. The resulting proportions are represented in Figure 7. Figure 7. Proportions of anteriority for each Class 67 175 184 402 46 209 299 183 33 132 159 107 25 123 143 60 29 123 190 58 24 90 151 44 16 46 201 59 25 331 86 10 404 72 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Class I Class II Class III Class IV Pius ten Hacken and Renáta Panocová 48 In Figure 7, the colours have the same meaning as in Figure 6. The numbers are the absolute number of words (types) in the relevant class. In cases where two verbs correspond to one noun, both have been counted in the relevant class. For the same reason as in Figure 6, the data set in the calculation includes the cases where one member of the verb-noun pair is obsolete. In the interpretation of Figure 7, the position of the yellow blocks is a central component. In Classes I, II and IV, the 70% mark is in the yellow block. This means that 70% of the pairs have the verb attested earlier than or at the same time as the noun and 30% of the pairs the verb later than or at the same time as the noun. Class III is an exception. Here the yellow block includes the 40% mark. As is clear from the absolute numbers, Class III is also the largest class. The contrast between Class III and the other classes suggests very strongly that backformation is frequent in the former, much less so in the latter. The reason why we do not make the claim that Figure 7 demonstrates frequent backformation in Class III is that there is always another interpretation of the data about the first attestation dates. Ten Hacken (2019: 81) discusses the Dutch example of ontroostbaar (‘broken-hearted’). The adjective is derived from the verb troosten (‘comfort’) by two processes, one attaching the suffix -baar, the other the prefix on-. The suffix -baar attaches to verbs and is semantically comparable to English -able, the prefix onattaches to adjectives and is semantically comparable to English un-. Because of the categorial restrictions on their application, -baar must have been attached first, resulting in troostbaar. This adjective has a transparent meaning, but it does not occur in Dutch dictionaries nor in the Corpus Hedendaags Nederlands (CHN 2013). The reason for this non-occurrence seems to be that it is not deemed necessary by Dutch speakers for their communication needs. A web search shows some occurrences. It may therefore become established in future. However, troostbaar is not a backformation, but an intermediate result of the derivation of ontroostbaar. The difference of this case to the phenomenon discussed for (7) and (8) can be illustrated by the example in (9). (9) a. casual c1374 b. casualize 1950 c. casualization 1920 The noun in (9c) is the result of two formation steps, starting from the adjective in (9a). One can interpret the appearance of the verb in (9b) as a backformation from (9c). However, one can also see it as the realization of an intermediate stage in the formation of (9c) from (9a). In that case, (9b) is less of a new word than if (9c) had not existed. The appearance of (9b) can be seen as the realization of a word implied by the derivation of (9c). Such an analysis is only possible if the verb is perceived as the result of suffixation. This option is not available for (7) and (8). In transplant, there The Suffix -ation in English 49 is no suffix and in deviate, the ending -ate cannot be analysed as a suffix attached to an attested base. In fact, the pattern in (9) occurs mainly in Class I and Class II. However, as Figure 7 shows, in these classes the verb is usually attested earlier than the noun, so that the number of actual cases is rather limited. In the discussion of backformation, we noted that it is a category that only appears in a historical account of the lexicon. It applies to the context where a speaker adds a word to their mental lexicon without hearing or reading it first. In the case of the V-N pairs we are looking at here, if the V is earlier, the N is the result of regular word formation, but if the N is earlier, the V is the result of backformation. Backformation involves the reanalysis of the N as a complex word. In the case of -ation, this usually means that the N was originally a borrowing from French or Latin. As illustrated also by several of the studies in ten Hacken and Panocová (2020), such a reanalysis is a necessary step in the emergence of a new word formation rule on the basis of another language. The distinction between diachronic and synchronic perspectives is important both in backformation and in the distinction between Class III and Class IV, but not in the same way. In the case of backformation, the significance of the phenomenon depends on the diachronic perspective. Only when we consider the situation of a speaker producing a new word without hearing or reading it first, is backformation distinguished from word formation. For this situation, first attestation dates provide valuable information, because they allow us to assess the likelihood that a speaker would have encountered a particular word before coming up with the one we are studying. In the case of the distinction between Class III and Class IV, this information is crucial. In general, the first attestation date is not a property of the word as a part of the language. Instead, it gives information about the likely input individual speakers have when they start using a word that is new to them. 5. Marchand’s Classes in the History of -ation As a final question, we will now consider how Marchand’s four classes of nouns in -ation, listed in Table 1, developed over the centuries. A crucial type of data here is the date of first attestation of nouns in -ation. We can take Figure 2 as a starting point, but, as we are interested in the individual classes, we should divide the columns. A problem in dividing the set of nouns in -ation according to Marchand’s classes is that there is a non-negligeable degree of ambiguity. The ambiguity between Class III and Class IV was discussed in section 4.1. This ambiguity can be expected because it is not possible to determine from the form of, for instance, devaluation whether the verb should be devalue or devaluate. However, it is in principle possible to construct a Pius ten Hacken and Renáta Panocová 50 verb in -ate for every noun in -ation, also for those ending in -ification and in -ization. In the case of nouns in -ification, it is clearly the preferred option to have a corresponding verb in -ify. In Figure 1, we represented these nouns as belonging to Class I. However, of the 240 nouns in this class, 29 (12%) have a corresponding verb in -ate. In two cases, listed with first attestation dates in (10), OED only gives a verb in -ate. (10) a. despecification 1872 despecificate 1873 b. nostrification 1885 nostrificate 1889 Despite the form of the nouns, the cases in (10) belong to Class III. For both pairs, the verb is slightly later than the noun, but the difference is so small that we cannot draw any definite conclusions from it. When we look at the other 27 cases, however, we note that in all but two cases, the verb in -ate has a later first attestation date than the noun in -ation. In many cases there is a significant difference. This suggests that these verbs are generally backformations. Most of the verbs in -ate are obsolete or rare, but there are also cases such as pontificate, which is much more frequent than the expected verb pontify. In the case of nouns in -ization, the situation is somewhat different, because -ize is a productive suffix. Although Marchand (1969: 300-301) treats -ify as a suffix, most of the nouns ending in -ification do not have a corresponding verb where -ify can be analysed as a reflection of a suffixation rule in English. Triples such as (9) are frequent for nouns in -ization, but not for nouns in -ification. This means that there is a stronger pressure to treat any noun in -ization as a Class II noun. For only ten such nouns (1.1%) OED gives a corresponding verb in -izate and in all but two cases, the verb in -izate is obsolete. In section 2, we already mentioned solmization, in which -ize is not a structural component. The other case is more interesting and involves the words in (11). (11) a. recognize 1388-9 b. recognition c1430 c. recognizate 1799 d. recognization 1560 The noun-verb pair (11a-b) is well-established and does not involve -ation. In Latin, it corresponds to recognoscere, an irregular verb of the third conjugation with the passive participle recognitus. OED gives (11b) as a borrowing from French and Latin.For the verb (11a), OED gives an etymology based on French reconnaître with an alignment on -ize to adapt it to English. The Suffix -ation in English 51 The early attestation date is misleading, because it refers to an obsolete meaning restricted to Scottish law. Other senses are attested from 1509 only. (11d) is the regular nominalization based on this verb. OED (2000- 2020 [2009]: recognization) gives it as semantically equivalent to (11b) and qualifies it as “Now rare (chiefly U.S.)”. The verb in (11c) is clearly a backformation based on (11d). OED labels it “rare”. This example shows that backformation is always possible, but the frequency of verbs in -ize makes its use in the case of nouns in -ization vanishingly rare. In view of the discussion in section 4, we argue that it would be inappropriate to divide the full set of -ation nouns into the four classes in Table 1. Instead, we consider each corresponding verb as yielding a pair that can be assigned to one of the classes. Nouns such as devaluation and recognization occur in two such pairs. In Figure 8, we represent the nouns by century and by class. Figure 8. First attestation dates of nouns in -ation sorted by Marchand’s classes When we consider Figure 8, it is obvious that Class III and Class IV were at the origin of the suffix -ation in English. Class III predominates and accounts for roughly twice as many pairs as Class IV. However, for both classes, the peak of their activity lies in the past. About half of the formations are from the 16 th and 17 th centuries, 53.4% for Class III and 49.3% for Class IV. This can be connected with Durkin’s (2014: 307-308) observation about the ornate writing style in this period. 0 100 200 300 400 500 600 700 800 Class I Class II Class III Class IV Pius ten Hacken and Renáta Panocová 52 The situation for Class I and Class II is quite different. Class I is not large, but it can be observed that more than half of the cases are from the 19 th and 20 th century (53.8%). For Class II, this effect is much stronger. In fact, 85.4% of the formations are from the last two centuries. The relative importance of Class II started growing in the 18 th century already. As Figure 8 shows, whereas all other classes show a sharp decline in that century, Class II has more cases in the 18 th than in the 17 th century. The different fate of the individual classes can also be seen when we consider the proportion of cases that have become obsolete among the overall cases. The development over time is visualized in Figure 9. Figure 9. Proportion of pairs of which the noun or the verb has become obsolete In Figure 9, three of the classes show a peak in the 17 th century. Of the large numbers of Class III and Class IV for this century, more than half of the pairs did not survive as a pair. This suggests that the frequent use of -ation in these classes was driven by a fashion that passed. Class II shows a different pattern. The rise in the 15 th century is caused by the appearance of the class. Before that, there were no pairs that could become obsolete. From the 15 th century, there is a regular and gradual drop. This is what we can expect if we assume a constant rate of obsolescence. Older pairs are more likely to become obsolete. We can summarize the history of -ation as follows. After a number of borrowings of nouns in -ation in the 13 th century, there were enough cases of such nouns with corresponding verbs for a reanalysis of the nouns as the result of a word formation rule in the 14 th century. Many of the verbs were based on the Latin passive participle, either as a borrowing with reanalysis or as a backformation on the basis of a noun in -ation. In the 16 th century, a fashion for an ornate style with synonyms came up, which greatly increased the need for new verbs and nouns with a Latinate appearance. This 0 10 20 30 40 50 60 70 Class I Class II Class III Class IV The Suffix -ation in English 53 fashion favoured the spread of the -ation rule and peaked in the 17 th century. The change of fashion in the 18 th century hit the -ation rule harder than other word formation rules. The number of new formations dropped dramatically and many earlier formations fell out of use. When naming needs increased again in the 19 th century, it was especially to verbs in -ize and, to a lesser extent, -ify that the revival of the -ation rule applied. 6. Conclusion As set out in the introduction, this paper proposes an analysis of the English suffix -ation on the basis of data from OED (2000-2020). The use of a historical dictionary makes it possible to study the emergence and obsolescence of words as documented by lexicographers using corpora. Given the history of the compilation of OED, as described by Gilliver (2016), the data we collected are less homogeneous than one would have desired. Although at each stage, OED has been based on corpus use, work on the first edition drew on citation slips resulting from a reading programme rather than the modern computerized access to the corpus. Nevertheless, we decided to use all data provided in the current state of OED rather than restricting ourselves to the sections updated for the third edition, as Durkin (2014) does. Both approaches are viable, but each has its caveats. Whereas Durkin has to extrapolate from a smaller proportion of the lexicon, we have to make some allowance for the partial coverage of 20 th century data. In our study of -ation, we considered two fundamentally different questions, formulated in (12). (12) a. How is -ation represented in the mental lexicon of current speakers of English? b. What were the considerations of speakers of English when they formed a new noun in -ation? Whereas (12a) is essentially synchronic, (12b) has a diachronic perspective. In the study of these questions, we used different data sets. For (12a), we used the set of noun-verb pairs in which neither the noun nor the verb is marked as obsolete. In this set, there are 2927 nouns. For (12b), we included also those pairs in which the noun or the verb (or both) are now obsolete. For this set, we added 1029 nouns. In some cases, more than one verb is connected to a noun, so that the number of pairs we considered is higher. In our analysis, we used the four classes distinguished by Marchand (1969) and exemplified in Table 1 as a starting point. For (12b), we also used the dates of first attestation of nouns and verbs as recorded by OED. Pius ten Hacken and Renáta Panocová 54 The distribution of the set of 2927 nouns over the classes in Table 1 is recorded in Figure 1. In this diagram, there is a significant segment representing nouns that have both a corresponding verb in -ate (Class III) and one without any special marking (Class IV). As we argued in section 4.1, it is not legitimate to divide these nouns into those that are ‘actually’ in Class III and those that are ‘actually’ in Class IV, because this is not a property that can be assigned to nouns as abstractions. It is only the mental lexicon of individual speakers that may assign them to a single class, but in cases such as devaluation with the verbs devaluate and devalue, even individual speakers may know both verbs. Also the set of 2927 nouns we worked with is an abstraction. It is only in the mental lexicon of individual speakers that nouns and corresponding verbs may be realized. As shown in Figure 1, Class III is the largest, accounting for around 40% of the nouns, followed by Class II with 30%. Class I is the smallest, with just over 8%. In the discussion of (12b), we considered three more detailed questions. One question was the historical prominence of -ation over the centuries. For this question, we evaluated the number of newly attested nouns in -ation per century (Figure 2) and the proportion of these nouns as a percentage of all newly attested nouns (Figure 3). The peak in new -ation nouns in the 17 th century is visible both in absolute and in relative numbers, whereas the subsequent development of -ation follows the general rate of new nouns quite closely. A second question concerned the extent to which new nouns in -ation were motivated as deverbal formations or as borrowings. We only considered those nouns in -ation for which a corresponding verb existed at some point, so that if the noun is a borrowing the verb is a backformation. Of course, first attestation dates give only indirect evidence about the probability of deverbal formation, but Figure 6 shows that overall there is a preponderance of verbs appearing before the corresponding nouns. We found a marked difference between the classes in Table 1, with backformation much more frequent in Class III than in the other classes. The third question is about the development of the individual classes distinguished by Marchand (1969) and listed in Table 1. As depicted in Figure 8, Class III shows a quite spectacular rate of new formations in the 17 th century, which was never again attained afterwards. This may be explained by the fashion of doublets and triplets described by Durkin (2014). Class IV follows Class III in its development, but at a markedly lower general rate. As Figure 9 shows, for many 17 th century formations of these classes the noun or the verb is now obsolete. It is rather a different story for Class II, which emerges in the 18 th and peaks in the 19 th century. Given the partial coverage of the 20 th century in the current state of OED, there is no reason to assume that formations of this class have passed their peak. Class I is so small that the effects are more difficult to evaluate, but there is a clear peak in the 19 th century. Here, a corpus-based analysis of the developments in the 20 th century would be of particular interest. The Suffix -ation in English 55 As a final note, we would like to return briefly to the question of the formal variants of -ation. Bauer et al. (2013: 201) give no less than eight variants, including -ition, -tion, and -ion. As we explained in section 1, -ation includes the thematic vowel -aof the Latin first conjugation. For Latin, we gave examples of formations from other conjugation classes in (3). In (13) we give the English verbs and nouns corresponding these Latin examples. (13) a. define definition b. deduce deduction c. defend defense d. complete completion Obviously, English speakers link the nouns in (13) to the corresponding verbs. In the case of (13c), only speakers with etymological knowledge will see the connection to -ation, but for the others, it is clearly visible. However, all nouns and verbs were borrowed from Latin or French. The difference to pairs including -ation is that whereas for -ation, there is a word formation rule that leads to new pairs, for -ition in (13a) and -tion in (13b, d), the only use of the generalization is to connect independently borrowed nouns and verbs. The strong position of the first conjugation in Latin has thus led to a reanalysis of pairs with a noun in -ation to form an English word formation rule which has -ation as a suffix. As there are no conjugation classes marked by thematic vowels in English, the function of the -aas a thematic vowel was lost. There was therefore no reason to use -ition and -tion for nouns corresponding to a particular class of verbs. They were only used in borrowings. The large number of nouns in -ation increased the barrier to reanalysis of nouns such as definition to produce a rule for -ition. The pattern with -ition was less frequent and the rule would compete with the one for -ation. Bauer et al. (2013: 201) give many examples of -ition and -tion as attached to “bound bases”, e.g. diction. In our view, the bound base in diction does not exist in English, but only in Latin. In English, diction is a simple noun. English has a word formation rule marked by -ation. It cannot have been borrowed from Latin, because Latin does not have a corresponding word formation rule. In Latin, there is a suffix -ion, which attaches to the verb stem marked by -t-. The English rule emerged as a consequence of reanalysis. This reanalysis did not take into account the Latin structure of words, but assigned sense to the forms that had been borrowed. This resulted in a rule for -ation, leaving pairs such as the ones in (13) stranded. Word formation rules such as the suffixation with -ation have two functions. One is the formation of new words as additions to the language of the speech community, the other to assign new words a place in a speaker’s mental lexicon. In our study, we showed how both functions can be studied on the basis of the data from OED, provided they are interpreted properly. Pius ten Hacken and Renáta Panocová 56 References Aronoff, Mark H. (1994). Morphology by Itself: Stems and Inflectional Classes. Cambridge (Mass.): MIT Press. Bauer, Laurie (2001). Morphological Productivity. Cambridge: Cambridge University Press. Bauer, Laurie, Rochelle Lieber & Ingo Plag (2013). The Oxford Reference Guide to English Morphology. Oxford: Oxford University Press. BNC (2007). British National Corpus. XML edition, University of Oxford. [online] http: / / www.natcorp.ox.ac.uk/ [26 November 2021]. Brewer, Charlotte (2012). “‘Happy Copiousness’? OED’s Recording of Female Authors of the Eighteenth Century”. The Review of English Studies 63 (258). 86- 117. CHN (2013). Corpus Hedendaags Nederlands. Leiden: Instituut voor Nederlandse Lexicografie. [online] https: / / portal.clarin.inl.nl/ search/ page/ search [27 August 2021]. COCA (2008-2020). The Corpus of Contemporary American English. Mark Davies (Ed.). [online] http: / / corpus.byu.edu/ coca/ [26 November 2021]. Dokulil, Miloš (1982). “K otázce slovnědruhových převodů a přechodů, zvl. transpozice”. Slovo a slovesnost 43. 257-271. Durkin, Philip (2014). Borrowed Words: A History of Loanwords in English. Oxford: Oxford University Press. ten Hacken, Pius (2015). “Transposition and the Limits of Word Formation”. In: Laurie Bauer, Livia Körtvélyessy & Pavol Štekauer (Eds.). Semantics of Complex Words. Cham: Springer. 187-216. ten Hacken, Pius (2019). Word Formation in Parallel Architecture. Berlin: Springer. ten Hacken, Pius & Renáta Panocová (2011). “Individual and Social Aspects of Word Formation”. Kwartalnik Neofilologiczny 58. 283-300. ten Hacken, Pius & Renáta Panocová (Eds.) (2020). The Interaction of Borrowing and Word Formation. Edinburgh: Edinburgh University Press. Gilliver, Peter (2016). The Making of the Oxford English Dictionary. Oxford: Oxford University Press. Kiparsky, Paul (1982). “Lexical Morphology and Phonology”. In: Linguistics in the Morning Calm: Selected Papers from SICOL 1981. Linguistic Society of Korea. Seoul: Hanshin. Vol. 1. 3-91. Marchand, Hans (1969). The Categories and Types of Present-Day English Wordformation: A Synchronic-Diachronic Approach. 2 nd edition, München: Beck. Marsden, Lloyd (2013). “Dating Middle English evidence in the OED”. OED blog, 8 March 2013. [online] https: / / public.oed.com/ blog/ dating-middle-english-evidence-in-the-oed/ [9 April 2020]. Mühleisen, Susanne (2010). Heterogeneity in Word-Formation Patterns: A corpusbased analysis of suffixation with -ee and its productivity in English. Amsterdam: Benjamins. OED (2000-2020). Oxford English Dictionary. Third edition, edited by John Simpson and Michael Proffitt. Oxford: Oxford University Press. [online] www.oed.com [26 November 2021]. Panocová, Renáta (2017). “Internationalisms with the Suffix -ácia and their Adaptation in Slovak”. In: Eleonora Litta & Marco Passarotti (Eds.). Proceedings of the Workshop on Resources and Tools for Derivational Morphology (DeriMo), 5-6 October 2017, Milano, Italy. Milano: EDUCatt. 61-72. The Suffix -ation in English 57 Štekauer, Pavol (2015). “Backformation”. In: Peter O. Müller, Ingeborg Ohnheiser, Susan Olsen & Franz Rainer (Eds.). Word-Formation: An International Handbook of the Languages of Europe. Berlin: Mouton De Gruyter. 340-352. Thomas, Claire (2013). Characterizing the polysemy of French and English deverbal nominalization suffixes. PhD thesis, Swansea University. Uriagereka, Juan (1998). Rhyme and Reason: An Introduction to Minimalist Syntax. Cambridge (Mass.): MIT Press. Pius ten Hacken Leopold-Franzens-Universität Innsbruck Renáta Panocová Pavol Jozef Šafárik University Košice