TESTING THE SUFFIXOIDIZATION OF GERMAN -MANN ‘-MAN’

TESTING THE SUFFIXOIDIZATION OF GERMAN -MANN ‘-man’ Throughout its history in Old High German (OHG), Middle High German, Early New High German, and Modern German (ModG), the word Mann ‘man’ has exhibited a high degree of productivity and linguistic flexibility. It has ranged morphosyntactically from a full noun (OHG man ‘man’), compounded noun (OHG werolt.man ‘human, mortal’), affixoid (ModG Sports.mann ‘athlete’), indefinite pronoun (ModG man ‘one’), to something approaching zero (undergoing demorphologization in the OHG compound gom.man ‘husband, man’). The affixoid notion and its usefulness in morphology have been controversial in the literature (e.g., Schmidt 1987). Relatively recently, Stevens (2005) proposed criteria for the category affixoid, and a number of standard German grammatical and etymological references (e.g., Duden 1995) have been utilizing this term. This study involves examining modern usage of the form -mann as found in the online Spiegel newsmagazine’s database, as well as the collection of historical data primarily on the forms of German -mann from the robust Titus database in Frankfurt. Stevens’ (2005) criteria are evaluated and used to measure the validity of German -mann’s membership in the category of suffixoid. Lingvistika-2011-01-93 1/5/12 1:33 PM Page 282


INTRODUCTION
Linguists have argued the usefulness of the concept suffixoid for several decades now (Elsen 2009, Leuschner and Wante 2009, Lightfoot 2000, 2005, Stevens 2000, 2010).Recently, Stevens has proposed criteria to classify this phenomenon (2010).If such parameters for classification prove useful, there is a stronger argument in favor of utilizing the linguistic category of suffixoid.This study uses German Mann/-mann 'man/person', relegated to the status of suffixoid/Halbsuffix (Duden 1995, Lightfoot 2006), as an object of investigation to test Stevens' proposed parameters.Discussion then follows on the advantages and challenges related to the given parameters, and future work is suggested.

THE CATEGORY OF SUFFIXOID
The linguistic notion of suffixoid refers to a morphosyntactic category situated developmentally between a full word and a traditional derivational suffix.The literature also sometimes refers to this construct as a semi-suffix or in German as a Halbsuffix.Similar terminology is likewise linked to affixes (affixoid) and prefixes (prefixoid), but not with confixes (e.g., German bio-, which is a bound form) (Stevens 2010: 79), nor with infixes (e.g., Latin rupit 'broke' with its nasal morphology inserted root-interior which yields rumpit 'breaks') (Matthews 1991:132).Most scholarly work on suffixoids and the like is in Germanic linguistics, though Slavic scholars have given attention in this area at least in the 1970s and 1980s (Lightfoot 2000: 56).It deserves mention that at least three of the standard references for the German language, one an etymological resource (Kluge 1999), one a general grammar work (Duden 1995), and one an introduction to German word-formation (Erben 2000) have all embraced the notion of affixoids as accepted terminology. 1  Simply put, as a linguistic item between the categories of root and suffix, a suffixoid is best viewed from an historical perspective as a kind of example of heterosemy.That is, at least two morphosyntactically variant linguistic items are synchronically on hand that stem from the same etymon.And specifically, one variant must be an unbound form, and another variant shares the same form, but is bound, productive, and has a more generalized meaning.A typical example in the literature is German Werk/-werk 'e.g.work, station' / 'e.g.collective, grouping of a whole whose individual parts are not readily or foremost perceived' (Erben 2000: 139, Henzen 1947: 192-193, Lightfoot 2005: 589-590, Nübling et al 2008: 77-79, Stevens 2010: 76-77).During the Middle High German (MHG) period (about 1050-1350 A.D.), for example, we witness an orthographical variant of Werk, namely -werc, where the crux of the semantics lies on the second compound constituent, quite the norm for Germanic root plus root constructions, thus signaling (1) vor-werc 'outerworks or projection from a fortification'.
Not only does the root exist in compounds at this time, but the genetically related simplex form continued to exist as it has throughout the ongoing High German periodizations.
Thus the formally similar, genetically related items are seen synchronically with a free morphosyntactic manifestation (a root), and at a middle stage displaying semantics associated with a root, but bound formally (a root constituent of a compound), and at a stage bound formally and having lost full semantic gravity associated with its root cognate.These are affixoidal characteristics.Remarkably, though the diachronic picture tends to often yield roots continuing to exist alongside compounding and affixizing formally similar but semantically generalizing cognates, the scholarly discussion has been and remains rather challenged to explore and systematically describe this fairly opaque morphosyntactic transition.2

CRITERIA FOR AFFIXOIDS AS PROPOSED BY STEVENS (2010)
Though scholars have remarked upon the characteristics of affixoids with relative vigor in particular from the 1970s and 1980s onward (Lightfoot 2000: 56, Schmidt 1987), Stevens (2000: 58-59) sought to synthesize these notions in diagnostic fashion.Thereafter, interestingly enough in a volume dedicated to grammaticalization phenomena in German (Leuschner et al 2010), Stevens continues calibrations to assist in identifying affixoids, indicating that his criteria are gleaned for the most part from collegial consensus (2010: 72-73).He arrives at the following compilation for affixoids, which is almost immediately followed by an addendum (inserted in c. below in italics) (2010: 74).
Tests for Affixoids: a. Affixoids in spoken languages are not just serial but usually very productive.They are the basis for new formations.b.Affixoids exist alongside a formally identical, and usually free, "parent" morph.This means there are two linguistic items identical in form and one is derived from the other.c.The meaning of the affixoid is more generalized and abstract than the formally identical parent.A generalized meaning includes metaphorical usage when there is no such usage of the free etymon.In contrast with some affixes, the affixoid does not express grammatical relations (like person, case, tense, etc.).d.There has been a shift of meaning in the relationship between the two parts of the word, so that the first, or other, component determines the basic meaning.That is, the syntactic relationship between the two morphemes is not like that of a subordinating compound.The relationship between the parts of the word are also not like other compounds (copulative, exocentric, etc.).Related to this is the inability to paraphrase the formation of the affixoid, e.g.German Riesenkrach '(a) very loud noise' is not equal to *'Krach eines Riesen' '(the) sound of a giant' or *'Krach, wie ihn ein Riese macht' '(the) sound a giant makes'.e.The affixoid must be in competition with or in complementary distribution with affixes.
(slightly adapted from Stevens 2010: 72-74, translations from Stevens 2000: 59) The main additions in the more recent work building on Stevens (2000) are point e., that is, the affixoid is in a relation of sorts with a system of affixes, and the aforementioned inclusion in c. regarding possible metaphorical extension of the bound form.
These tests appear sensible from a theoretical standpoint.As noted above, they are generally consensual.Let us now put them into practice, taking German Mann/-mann as the principle example.Do the above criterion perform well in categorizing real language?

APPLYING THE TESTS TO GERMAN MANN/-MANN
This study utilizes German Mann/-mann 'man, person' as a primary example due to its recognition in a standard reference as a suffixoid (Duden 1995), which is bolstered due to comparative developments in other Germanic languages (e.g.see Booij 2009: 208-209 for Dutch, and consider an example like English sportsmanship, arguably fully "bleached" of the cognate meaning 'man'), and a final reason for selection is the form's high degrees of usage and heterosemy.The diachronic record for -mann is also examined below, as the form evidences a long history of developments and the semantics involved are complicated.
To satisfy (a) above, the bound form -mann must demonstrate a high degree of productivity.As the Duden (1995) pertains to Modern German, it is in this period where certain examples are initially sought.An online lemma search of the mainstream newsmagazine Spiegel's archive (containing searchable issues since 1947) yields approximately 200 usages of Sportsmann 'sportsman', 75 of Forstmann 'forester', and 23 of Weltrekordmann 'world record holder'.These three types were searched because they are the complete set of examples noted in the Duden (1995: 508).The relatively high frequency of tokens in the newsmagazine genre alone attests to a certain threshold of productivity to satisfy condition (a).These three types also appear in Mater's (1967) backwards dictionary of German, and 207 other such types further attest to the robust productivity of -mann synchronically (as opposed to 71 types for -frau).
Criterion (b) is also thus met, as it is clear the very commonly used noun Mann (an unbound root) co-occurs in Modern German along with bound examples such as those noted immediately above (Mater 1967).That the bound form derives from the unbound similar form cannot be disputed, as noted in the standard references and other works such as Erben (2000: 38) and Henzen (1947: 195).
Criterion (c) potentially presents something of an issue, etymologically speaking.The historical record indicates that reconstructed Germanic *mann 'man, person' (Duden 1995, Kluge 1999) was not limited to signaling male human beings, and this would lend little to no room for generalization of meaning from MALE > ANY PER-SON in the affixoid.As in Kluge (1999), though, we see that the main meaning for mann in the various periods leans predominantly toward 'man' (i.e.adult male), so it could be argued that mann historically foremost meant 'male'.Further evidence for such a consideration of maleness associated with mann comes from the construal by modern speakers of the cognate indefinite pronoun man 'one, you', which is supposed to denote anyone, regardless of sex.Speakers can perceive the usage as awkward in utterances such as the following: (5) Man wird Mutter 'One becomes a mother'.or (6) Man geht auf das Damenklo 'One goes to the ladies' W.C.' In some instances, a "female" pronominal variant may serve as a substitute for the prescribed pronoun to ostensibly balance out the perceived linguistic unfairness. 3 simple Google search reveals, however, that online usage is oriented typically around domains that are female-focused, as in ( 5) and ( 6) above.Example (7) comes from a source about women's fashion (mypashion.com), and the normally unwarranted orthographic capitalization of the nascent indefinite pronoun illustrates the novelty and biological gender meaning encroaching on this construction.
(7) [D]as muss Frau einfach lieben.'One (female) has to simply love it/that.' Based on ( 5)-( 7), I argue that since a substantial segment of modern speakers must construe free forms of man as at least sometimes signaling something like 'one (who is a male)', a modern perception of the bound form -mann generalized to mean 'anyone (regardless of gender)' fulfills point (c) on generalization/abstraction of meaning.Such examples are rare in the modern Spiegel corpus.One must search back to the summer of 1989 to find a lemma that is semantically ambiguous in terms of the signified's sex.The context is that of couples who have lived together before getting married, and what the supposed advantages of such previous experience are thought to be.The message in ( 8) is that one is not better assuring a lasting marriage through intimate premarital knowledge (and up close "studying", as one does with one's opponent or prey) of one's partner.The -mann in (8) thus functions more indefinitely with respect to maleness or femaleness, and thereby criterion (c) is (perhaps) met.
For criterion (d), we should find a semantic shift from -mann to the first element in such constructions, and the meaning of the whole construction should not be directly paraphrasable.This proves difficult for Modern German Sportsmann, which can be paraphrased as 'one who is sporting', and it is a similar case for the other examples given in the Duden (1995): Forstmann 'one who works in forestry'; Weltrekordmann 'one who holds a world record'.The basic meaning can still be borne by the second constituent in these constructions, though it is possible to also construe other lexemes with just single roots, such as 'athlete', 'ranger', and 'champion' respectively.
Criterion (e) states that the affixoid is in competition with or complementary distribution with a system of affixes.The Duden (1995: 508-509) considers -mann to be in competition with the suffixes -ler (Sportler and Weltrekordler) and agentive -er (Förster), to follow through with the previous examples.Given the 8 examples of -frau in Sportsfrau in a google search, it may well be that this form is providing further competition.A variety of other personal suffixes exist in German, such as -iker, -ist, and -ator, along with those derived from status labels (-papst 'pope', -baron), animals (-ratte 'rat', -hai 'shark'), personal names (-fritze, -huber), as well as others (see also Leuschner and Wante 2009).
Having measured up the largely recognized Modern German personal suffixoid -mann in accordance with the five proposed tests for affixoids, a number of questions arise, including whether it is necessary for all of the criteria to be fully met in order for a suffixoid to be recognized as such, how to interpret the diachronic perspective (b, c, d), and how to interpret this phenomena in conjunction with spoken language (a, d).We turn now to these and other questions.

DISCUSSION
The application of Stevens' (2010) diagnostics, themselves largely the product of state-of-the-art consensus, yields an interesting outcome when applied to Modern German -mann.Some criteria seem to apply well, others are more questionable with this particular recognized suffixoid.It may well be that the class of suffixoid, similar to other linguistic classes, is not monolithic and homogeneous, but is instead made up of more central and more peripheral members.Is it in fact necessary to have all five features (a-e) be in play for a suffixoid to be recognized as such?Stevens (2010) does not explicitly make this claim, but Elsen (2009: 319) notes it is important for a constellation of factors to be on hand for affixoidization, namely productivity, semantic change, and the unbound coexistence of a cognate form.This covers criteria (a-d), leaving out (e), systemic interaction with a class of affixes.I generally find (e) useful, just as in the case of -mann where it seemingly helps as an identifier, but as with all or most linguistic classes, it is wise to best consider such diagnostic criteria as possibly varying in applicable degree.
Criteria (a) regarding productivity, as point (e) above, appears at first glance to be relatively unproblematic.Productivity in the modern era is established.Leuschner and Wante (2009: 66) even note that at least three constructions have been what have counted in past scholarship to meet a productivity threshold.There are in fact, however, more -mann constructions that have arisen in modern times (and previously), so the potential minimum question of three is moot in this case (recall the over 200 types noted above in the modern language).
Although it is stated above that criterion (b) is fulfilled, i.e., a free "parent" morph coexists from which the suffixoid is derived, there is room for discussion here.The reconstructed and historical record often show a path leading from full unbound word, to compound constituent, and then to affixoid, but I would suggest we do not wish to imply there is a strong claim that that pathway exists in isolation, i.e., out of reach of any other linguistic influences, analogy, and so-called semantic contamination.How can we assume that the development of the indefinite pronoun man did not also exert a semantic influence (part 4 above) and contribute to indefiniteness (in terms of gender or otherwise) and semantic generalization linked to -mann constructions?
A related question centers on diachronic semantics and points (b) and (c).Various -mann constructions have been lexicalized as compounds or suffix-like formations for at least 1,500 years, if not more than two millenia (Kluge 1999, Lightfoot 2006), and they continue to be formed anew.Man as an indefinite pronoun has been recorded since at least the OHG era (Kluge 1999, etc.).How does one decide at which point in time a given "parent" form's semantics ought to be scrutinized in relation to the suffixoid?It is sometimes difficult to judge if certain semantics have been shed over time.This point is particularly salient for -mann formations due to the modern perception of sexism and exclusion in language.An etymological approach might claim that the meaning associated with the reconstructed Germanic or OHG *mann 'man, person' is the reflexive meaning to refer to when considering Duden's (1995) modern example of Forstmann 'ranger, woodsman' (as opposed to 'male adult'), especially since the construction goes back to OHG forstmann 'woodsman' (Henzen 1947).Indeed, it was back in that early Medieval period in which Henzen explains that the compound constituent was transitioning into something more suffixal (1947: 194-195). 4oes this mean the process of suffixoidization has been ongoing for over 1,300 years in light of the Duden claim that -mann in this construction is (still) a suffixoid?This seems less than likely for at least two reasons, which could mean speakers' perceptions of "parent" morphs may be more salient than the actual etymological record when considering point (b).
The reasons pointing away from 1,300 years of suffixoidization include a lack of data showing similar such lengthy developments in other suffixoids, and the speaker-based nature of affixoidization.Two previous studies, examining the historical suffixoidal nature of German -heit '-hood' as in Kindheit 'childhood' (Lightfoot 2005) and the development of the constituent -mann in the formation gomman 'husband, man' (Lightfoot 2006), were carried out using data from exhaustive searches in the Titus database, which contains all of the main OHG works and some of the minor ones.Evidence points toward suffixoidization having taken place for the -heit form within the OHG period, possibly also somewhat prior to it (cf.similar derivational forms such as -schaft '-ship', -tum '-dom', etc.;Erben 2000, Henzen 1947, Nübeling et al 2008).The gomman study (Lightfoot 2006) is mostly limited to one OHG work, Tatian's translation of the Gospels, and it concluded somewhat more authoritatively that development from a transparent compound constituent -man took just one period, namely OHG, to undergo demorphologization with a host which had a similar if not the same meaning.This is not suffixoidization, but it does show the same rough amount of time that this particular form took to complete a substantial morphological change.
The second plausible reason for the relatively quick progression of most personal suffixoids is their basis in colloquial language, which tends to undergo fairly rapid change in relation to written language.Duden (1995: 508-509) and Stevens (2010: 73) point out this colloquial character.High productivity in spoken language tends to accelerate change (cf. the well-known rate of change associated with youth language, for example).This brings us back to the claim by Henzen (1947: 195) of OHG Forstmann being in transition already up to 1,300 years ago.He was presumably referring to there having been a generalization of the semantics of -mann.A definite man was not meant, and the form appeared among a list of other similar constructions.Henzen's discussion frequently mentions second compositional constituents, which are noted above as the second stage in the progression of root word > compound constituent > affixoid > affix.Thus I would surmise Henzen is identifying the second stage as opposed to the third.
This then begs the question of what sets apart a second compositional constituent from a suffixoid.Stevens (2010) does not raise this issue; his discussion emphasizes the attributes of roots, affixoids, and derivational affixes.Does this question need to be raised, or could it be the case that Henzen's concept has some sort of conceptual overlap with the category affixoid?Henzen notes at least the same qualities as (a), (c), and (d) for the second compositional constituent (1947: 194-195), but his proposed category seems to crucially lack the coexistence of an unbound, formally similar cognate.This paper's assessment of (d), the shift in semantic weight from the second element to the first, along with the inability to paraphrase the whole construction by means of its two related root meanings, with respect to the -mann examples is problematic, as pointed out above in part 4. By and large, for Sportsmann, it almost always is a male referred to who is sporting, athletic, or has the positive qualities of a sporting man.Likewise a Weltrekordmann refers to a man who is a champion or has a world record.On the other hand, (d) seems perfectly in order in applying to other sorts of suffixoids such as -werk.In Laubwerk 'foliage' (Laub 'leaves'), for example, foliage is not "a work of leaves", but a complex whole of greenery or a collective entirety of greenery.In such cases the paraphrasing using the two original roots' meanings does not work, and indeed the semantic weight ends up with the first element.
Finally, the challenges with -mann and the prescribed semantic relations, changes, and paraphrasing lead one to consider whether this is a good example of an affixoid at all.It may alternatively be that some affixoids are better members of this category, and some are more peripheral members.This seems more likely than the possibility that Stevens' (2010) tests are simply off the mark.It has already been pointed out in the literature that personal suffixoids have a particular colloquial connection, and we can guess that this influences their usage and development.The form -mann itself actually does not appear in Leuschner and Wante's (2009) study of personal suffixoids, nor does it appear in Elsen's (2009) work, so this may lend weight to the argument that -mann is not an ideal candidate for recognition as a suffixoid, despite its presentation as such in the Duden (1995).

CONCLUSION
This paper examined the tests for affixoids proposed by Stevens (2010) and utilized them to test the German personal suffixoid -mann.The test points themselves are deemed to be workable.The results here lead to a number of questions and thoughts about the tests themselves, including whether all must be satisfied to count as a litmus test for membership.This work also raises issues such as whether differ-ent subclasses of affixoids need to be considered, and questions whether Modern German -mann actually belongs categorized in the Duden (1995: 508) as a suffixoid.
Previous literature on the topic has not emphasized that the colloquial origin (Stevens 2010: 73) of suffixoids is not to be forgotten-this can make looking for them in the written record especially problematic, and is an important methodological consideration, reminding the researcher to be extremely cautious when dealing with historical data.Personal suffixoids in particular (Duden 1995) occur perhaps even moreso in colloquial domains, so written dialogue and personal electronic communication such as blogs are likely fruitful places for their further investigation, given the oral influence presumed in those domains.