An Early Stage of the Development of Complementizer Agreement: Evidence from Wisconsin Heritage...

33
For use by the Author only | © 2015 Koninklijke Brill NV Moribund Germanic Heritage Languages in North America Theoretical Perspectives and Empirical Findings Edited by B. Richard Page and Michael T. Putnam LEIDEN | BOSTON

Transcript of An Early Stage of the Development of Complementizer Agreement: Evidence from Wisconsin Heritage...

For use by the Author only | © 2015 Koninklijke Brill NV

Moribund Germanic Heritage Languages in North America

Theoretical Perspectives and Empirical Findings

Edited by

B. Richard Page and Michael T. Putnam

LEIDEN | BOSTON

For use by the Author only | © 2015 Koninklijke Brill NV

Contents

Editor’s Preface  vii

1 Researching Moribund Germanic Heritage Languages: Theoretical and Empirical Challenges and Rewards  1

B. Richard Page and Michael T. Putnam

2 A Syntactic Model for the Analysis of Language Mixing Phenomena: American Norwegian and Beyond  12

Tor A. Åfarli

3 An Early Stage of the Historical Development of Complementizer Agreement: Evidence from Wisconsin Heritage German  34

Joshua Bousquette

4 Verb Second and Finiteness Morphology in Norwegian Heritage Language of the American Midwest  64

Kristin Melum Eide and Arnstein Hjelde

5 Where Discourse Structure and a Heritage Language Meet: Oral History Interviews of Swedish Americans  102

Angela Falk

6 Noun Phrase Case Shift in Volga German Varieties on the Great Plains of Kansas  133

William D. Keel

7 Incomplete Acquisition and Verb Placement in Heritage Scandinavian  153

Ida Larsson and Janne Bondi Johannessen

8 Language Shift, Religious Identity, and Phonological Traces of Pennsylvania German in Pennsylvania English: The Laxing of Unstressed /i/ among Pennsylvania German Anabaptists  190

B. Richard Page

vi contents

For use by the Author only | © 2015 Koninklijke Brill NV

9 Minimizing (Interface) Domains: The Loss of Long-Distance Binding in North American Icelandic  203

Michael T. Putnam and Birna Arnbjörnsdóttir

10 Sociolinguistic and Syntactic Variation in Wisconsin German Narratives  224

Alyson Sewell

Index of Authors Cited  251

© koninklijke brill nv, leiden, 2015 | doi 10.1163/9789004290211_004For use by the Author only | © 2015 Koninklijke Brill NV

Chapter 3

An Early Stage of the Historical Development of Complementizer Agreement: Evidence from Wisconsin Heritage German*

Joshua Bousquette

1 Introduction

This chapter examines the status of inflectional afffijixes in Complementizer Agreement (C-agr) contexts in heritage varieties of German spoken in Wisconsin, here referred to collectively as Wisconsin Heritage German1 (whg). Bousquette (2013) in his application of van Gelderen’s (2011) Linguistic Cycle accounts for the rise and development of C-agr in modern varieties of West Germanic as a process of grammaticalization and reanalysis of both subject pronouns and phonetically-derived hiatus efffects as inflectional afffijixes. This work focuses specifijically on the latter process, where C-agr develops through the reanalysis of surface forms. Data are drawn from interviews conducted with four fluent speakers recorded in 2011 and 2012, and will be compared to data from interviews with 11 speakers of Bavarian (mostly Oberpfälzisch) and 7 speakers of East Franconian recorded in May and June, 2012. Data from these interviews were fijirst analyzed for morphological distribution of C-agr, which shows that inflectional afffijixes derived from reanalyzed hiatus efffects are restricted to second person singular in whg, regardless of phonetic envi-ronment. Second, data were analyzed also for phonetic assimilation patterns,

* I would like to acknowledge Joe Salmons, Michael Putnam, Alyson Sewell, Ben Frey and Dan Nützel, both as fellow fijield researchers gathering data in Eastern Wisconsin, and for their feedback that was enormously benefijicial to developing this project. I would again like to thank Dan Nützel and also Johann Schmuck for going above and beyond to help in fijinding consultants in Germany. Lastly, thanks are due to the conference organizers and participants from the Third Workshop on Immigrant Languages in America at Penn State University on September 27–29, 2012 for their feedback. This work is based in large part on a presentation given at that conference.

1  whg is a blanket term for a set of separate varieties spoken in Eastern Wisconsin, rather than a homogeneous variety. No Wisconsin German koiné is recognized as having devel-oped, though speakers certainly accommodated between disparate varieties to facilitate communication.

35Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

which following de Haan (2010) provide a diagnostic for diffferentiating the surface level form (clitic) from the lexical form (afffijix), showing that subject pronouns in whg maintain their pronominal status, and have not been reana-lyzed or grammaticalized as afffijixes. This contrasts with attested modern variet-ies of West Germanic that exhibit C-agr, in which the C-agr inflectional afffijixes contain a pronominal component. As such, the comparative study of these heritage and continental varieties provides evidence of diffferent stages of the Linguistic Cycle specifijic to C-agr: the relatively more advanced stage seen in continental varieties contrasts with the earlier stage shown in whg, which maintains the C-agr structures inherited from the (West) Central Franconian varieties spoken by 19th century immigrants who settled in eastern Wisconsin (Bousquette 2014, cf. Nützel and Salmons 2011). This stage of the C-agr cycle shown by whg is unique among modern varieties of German, and provides invaluable insight into the process of language change that all C-agr varieties have arguably already undergone.

While the pronominal origin of the C-agr paradigm is visible in many vari-eties, including West Flemish (Haegeman 1992), still other varieties like West Frisian, Franconian and Bavarian exhibit inflectional afffijixes that include reanalyzed hiatus efffects in addition to the pronominal component (Weise 1907, Fuß 2004: 60–63). Innovation in the C-agr inflectional paradigm there-fore involves not only a reanalysis of subject pronouns, but also reanalysis of phonetic material that was not original to the lexicon or present in the syntac-tic derivation. Being that both the reanalyzed hiatus efffects and pronominal forms co-occur in modern continental varieties of West Germanic, it might appear that hiatus efffects and pronominal forms were reanalyzed at the same time. However, acoustic analysis of phonetic assimilation patterns in whg provides evidence that that subject pronouns have not been reanalyzed as inflectional afffijixes; C-agr inflection is instead derived from reanalyzed hiatus efffects. These data demonstrate a process which may give rise to C-agr irre-spective of the degree of grammaticalization of subject pronouns, and may additionally initiate the C-agr Cycle through the reanalysis of surface forms (Bousquette 2013: §1.2, §2.1).

Building on Bousquette’s model, it is here argued that C-agr in whg dem-onstrates an early stage in the development of the phenomenon, which in continental varieties is comparatively more progressed. It will be argued here that C-agr exhibited in present-day whg is consistent with earlier stages of the development of C-agr in the late 19th and early 20th centuries, as evidenced by surprisingly similar data from Schwäbl (1903), Böttger (1906) and Weise (1907). These parallels between whg and earlier continental varieties relate to the two main lines of argument considered here: fijirst, that C-agr in the heritage variety

36 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

has historical analogues from a chronologically earlier period of German, and represents an earlier stage in the development of C-agr than other modern varieties; and second, that acoustic analysis of whg C-agr consistent with the orthographic representation of C-agr in 19th century C-agr varieties of German shows that reanalysis of surface forms may initiate the C-agr Cycle. whg there-fore sheds light on our understanding of the historical development of C-agr, and on the mechanisms of syntactic change operating especially in the mark-ing of the subject agreement.

This paper begins with an overview of C-agr, its historical development and a brief account of C-agr as participating in a linguistic cycle, in §1. In §2 the role of phonetic efffects in language change will be discussed, with special refer-ence to how surface forms reflect deeper aspects of the grammar. This section also positions the reanalysis of phonetic forms as a parallel development to the grammaticalization model provided in §1. The method of elicitation and analysis of data for this particular study will be discussed in §3, with the pre-sentation of the data following in §4. Finally, in §5, the fijindings of this study are summarized, followed by a discussion of their implications for the study of C-agr, and language change in general.

2 C-agr and The Linguistic Cycle

C-agr is attested in multiple varieties of modern West Germanic, including West Flemish, Luxemburgish, and West Frisian, and non-standard varieties of German and Dutch. It characteristically involves the addition of an inflectional afffijix on a complementizer, as in the contrastive example here, of a pre-C-agr variety in (1a) (Standard German) and Bayer’s (1984) account of Bavarian (1b).

(1) a. wenn du kommst (pre-C-agr variety)If come‘If you come’

(1) b. wenn-st kummst (Bavarian; Bayer 1984: 233)if-2sg [pro] come‘If you come’

C-agr is attested most commonly for second person singular in West Germanic C-agr varieties, with some varieties licensing C-agr also in second person plu-ral, fijirst person plural, and in rare cases across the entire paradigm.

37Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

Additionally, C-agr inflection may appear on not just complementizers, but also on other topicalized elements in subordinate clauses such as temporal and locative wh-elements, indirect objects, topicalized prepositional phrases, and comparatives. An example is given below (2) where the dative indirect object wen inflects also for second person singular (Bousquette 2014: 566; see also Reis 1985, Kathol 2001).

(2) Weisst du, wen-st das Geld gegeben hast?

know you whom-2sg the money given have (East Franconian, Speaker ab) ‘Do you know to whom you gave the money?’

The formal synchronic explanation for the licensing of C-agr is that afffijixation is base-generated in the lexicon and becomes overtly realized on C through a goal-probe relationship with T/TP (cf. Carstens 2002; Chomsky 2002; van Koppen 2005; Putnam and van Koppen 2009). I adopt this explanation here (Figure 3.1).

The historical development of C-agr is therefore the reanalysis of either sub-ject pronouns or post-syntactic, phonetically-derived hiatus efffects as afffijixes generated in the lexicon and licensed in the syntactic derivation. This model “allows us to see cross-linguistic variation as located in lexical items, not in the computation” (van Gelderen 2007: 41). This account of the way language change occurs treats the reinterpretation of lexical items as fundamentally

figure 3.1 Feature checking in C-agr contexts

cp

x

tp

tvp

dp

d

pro<i-Phi>

[-st]<u-Phi>

wenn-st

x

x

c

38 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

changing their function and syntactic features, and not just their use in a given context.

There are two primary mechanisms of syntactic change by which C-agr may develop. Following Bousquette (2013: 63), the fijirst is “the language-internal grammaticalization of subject pronouns as afffijixes.” The grammaticalization of subject pronouns is formalized by van Gelderen (2011: 41) as the progressive loss of interpretable phi-features (i-phi) as speakers reanalyze them as afffijixes:

Changes connected to the subject agreement cycle occur when the inter-pretable person (and gender) features of a full pronoun are reanalyzed, i.e. selected from the lexicon, as uninterpretable when they become agreement. Topic/emphatic pronouns have semantic phi features that can be reanalyzed as interpretable and subsequently as uninterpretable. The reanalysis means that the phi features are reanalyzed from interpre-table on the (pro)noun to uninterpretable on T as part of the agreement.

These i-phi features are therefore inherently present or absent from lexical items; the progressive loss of a lexical item’s interpretable phi features over time is a formalization of grammaticalization. This change is not language-specifijic to Germanic, but rather is typologically supported across many lan-guages, in which lexical items are commonly selected from the lexicon with fewer and fewer interpretable features, efffectively restricting the lexical item to a progressively more grammatical function. This defijinition of grammatical-ization based on the reanalysis of syntactic features is essentially a syntactic formalization of Kuryłowicz (1965: 52): “Grammaticalization consists in the increase of the range of a morpheme advancing from a lexical to a grammati-cal or from a less grammatical to a more grammatical status.”

The second mechanism of syntactic change that may give rise to C-agr is the reanalysis of phonetic material as inflectional afffijixes. This serves as an alternate—but complementary—mechanism for the reanalysis of subject pronouns as inflectional afffijixes in addition to typological grammaticaliza-tion, but also applies to the reanalysis of other phonetic material as afffijixes, including the reanalysis of hiatus efffects. German is a prosodic language where “well-formed prosodic feet are either quantity-insensitive dactyls (x'xx) or tro-chees (x'x)” (Hanna 2009: 201). This means that pronouns, which as a lexical category do not bear main stress and only infrequently bear sentence stress, regularly exhibit phonetic reduction. This is particularly regular when sub-ject pronouns appear right-adjacent to the syntactic node C in subordinate clauses and subject inversion contexts in V2 languages like modern varieties

39Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

of West Germanic. Somers Wicka (2007, 2009) argues that pronominal forms appearing in that syntactic position in Otfrid’s Evangelienbuch do not con-stitute their own phonological word, but rather necessarily form a ‘prosodic word’ (pword) with the verb. Somers Wicka (2007: 9) argues that the forma-tion of pwords allows for post-lexical phonological process of umlaut—a pro-cess which typically does not occur across word boundaries. This argument fijirstly provides evidence that phonetic reduction of pronouns can result in the redrawing of word boundaries; and secondly, provides a second mechanism besides grammaticalization where pronouns may be interpreted as part of another lexeme.

Following Lightfoot’s (1979) ‘Transparency Principle’, which requires deriva-tions to be minimally complex, these phonetically-reduced pronominal forms become reanalyzed not as a phonetic variant of the same lexeme, but rather as a set of unstressed pronouns that contrasts in use and distribution with the original set of pronouns from which they were derived. Similar also to the syn-tactic principle of ‘simplifijication’ in which the “derived form is interpreted as the basic” (Kiparsky 1996: 146), the reanalysis of phonetically reduced pronouns as a separate set of unstressed pronouns eliminates the process of deriving two diffferent phonetic realizations for the same phonological form. And in fact, many of the modern varieties of West Germanic that exhibit C-agr also have dedicated sets of stressed and unstressed pronouns, including West Flemish (Haegeman 1992: 60), Bavarian (Fuß 2004: 60), and even standard languages like Dutch (Oosterhofff 2009: 62–63; Donaldson 2008: 67–69). The result of such systematic phonetic reduction is the creation of a set of pronominal forms that is already diffferentiated from full pronouns, not only in terms of phonetic real-ization, but also in terms of distribution, since stressed and unstressed pro-nominal forms cannot co-occur (Haegeman 1992: 60). It is this dedicated set of phonetically reduced pronouns that regularly occurs in syntactic positions directly adjacent to complementizers in subordinate clauses that regularly become reanalyzed as part of the same pword with the complementizer. The shift of the word boundary from lexical to prosodic facilitates the reanalysis of the unstressed pronominal variant as an afffijix.

The development of C-agr then also initiates the Linguistic Cycle specifijic to C-agr. Both mechanisms of syntactic change—pronominal grammatical-ization, and reanalysis of surface forms—give rise to C-agr and initiate the Linguistic Cycle. Building on the eponymous framework of Jespersen (1917), the Linguistic Cycle framework has been greatly refijined by van Gelderen (2007, 2011), as a cyclical, unidirectional evolution of a part of speech, motivated by principles of feature economy. Working within the holistic system of grammar

40 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

and not restricted to changes occurring independently in the phonology or morphology, the loss of interpretable phi features through either grammatical-ization or reanalysis requires the compensatory addition of interpretable phi features through the introduction of an innovative, reinforcing element. The basic, three-phase linguistic cycle is outlined in Figure 3.2, where the ‘weak-ening’ of the original marker in phase I necessitates reinforcement, and both the original and reinforcing markers thus co-occur in phase ii. As the cycle progresses, the reinforcing element supplants the original marker by phase iii.

Phase I: original marker > ‘weakens’ Phase II: original marker + reinforcing element Phase III: reinforcing element supplants original markerfigure 3.2 The three-phase Linguistic Cycle

C-agr involves the subject-agreement cycle, where subject pronouns are reana-lyzed as afffijixes. Shown in Figure 3.3 below (cf. van Gelderen 2011: 41) is the grammaticalization of subject pronouns, progressing from left to right, with each stage marking a change in syntactic position, lexical category, and +/- presence of phi features. As earlier mentioned, the process of grammaticaliza-tion is a reanalysis that occurs in the lexicon, such that the reanalysis of lexical category corresponds with a change in feature marking.

Adjunct Specifijier Head afffijixemphatic > full pronoun > head pronoun > agreement[semantic] [i-phi] [u-1/2][i-3][u-#]2 u-phifigure 3.3 Feature economy

The grammaticalization of subject pronouns as afffijixes locates the afffijix and therefore its u-phi features at C, the position of the complementizer in sub-ordinate clauses, under van Gelderen’s Late Merge Principle (lmp, 2007: 284). These u-phi features must then probe for i-phi features within a maximally local domain (cf. Figure 3.1; van Koppen 2005). Because the subject pronoun has been reanalyzed as an afffijix—a process in which i-phi features are lost—

2  Van Gelderen argues that fijirst and second person tend to participate earlier in processes of grammaticalization. The reanalysis is from i-Phi to u-Phi features therefore occurs earlier in fijirst and second than in third person.

41Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

a compensatory pronoun with i-phi features is inserted to repair the subject-agreement marking of the utterance. This initiates the progression of the C-agr Cycle into phase ii. This sort of change can be thought of as a push-chain change, since the process of grammaticalization results in the subject pronoun vacating its earlier syntactic position.

The development of C-agr through the reanalysis of surface forms—whether pronominal forms or hiatus efffects—acts instead as a pull-chain. The reanal-ysis of phonetic material in the surface form as inflectional afffijixes similarly locates u-phi features at C, albeit without the grammaticalization of subject pronouns. The resultant syntactic structure for subject agreement is thus the same as in fijigure 1, with u-phi features on C probing for a maximally local i-phi feature bundle, which is in this case still the subject pronoun (since it has not been reanalyzed). C-agr developed through reanalysis of surface forms there-fore results in the presence of u-phi features on C, but does not at its earliest stages initiate the C-agr cycle involving the reanalysis or grammaticalization of subject pronouns. However, the presence of u-phi features on C necessitates feature checking against the subject position, with the person and number features of the subject being necessary for the licensing of C-agr. This is the pull-chain that draws pronominal forms towards afffijixal status. Additionally, the development of C-agr prior to—and independent of—the grammatical-ization of subject pronouns has already laid the ground work for parameter resetting, efffectively providing a path for grammaticalization of subject pro-nouns within existing syntactic structures. Such structures follow the principle that extension is rule-governed and based analogically on extant elements of the grammar, that “the process of extension is systematic. . . . Observed exten-sions generalize to a natural class based on categories already relevant to the sphere in which the rule applied before it was extended” (Campbell and Harris 1995: 101). C-agr as exhibited by whg is crucially at this pivotal diachronic stage of the development of the cycle, and is our point of departure going forward.

2.0 Phonetic Efffects and Their Role in Language Change

This section discusses in greater detail how changes in surface structure are reanalyzed and internalized as parts of the (competence) grammar. As a starting point, consider grammar as a system that seeks to fijit aspects of the surface structure into aspects of the competence grammar along logical, sys-tematic or analogical lines. This reanalysis tends to prefer the most economi-cal form, which returns to the discussion of Lightfoot’s Transparency Principle (1979, 1991) for child language acquisition, and Kiparsky (1996) for simplifijica-tion in syntactic change through reanalysis. A similar processes of leveling is observed in heavy contact situations between multiple varieties observed in ‘new towns’ such as Milton Keynes, where variation across varieties is arguably

42 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

dis- preferred in favor of a single, more frequent form; or “majority forms in the mix, rather than minority forms, win out . . . marked regional forms are disfa-vored” (Kerswill and Williams 2000: 84).

Nevertheless there are counter-examples where reanalysis results in increased complexity, for example, where “high-contact, long-term contact situations involving childhood language contact are likely to lead to complexi-fijication through the addition of features from other languages” (Trudgill 2010: 23). This complexifijication through analogical extension has also been observed in Fortson (2003) in ‘metonymic’ and ‘metaphoric’ types of extension, which expand the semantic range of a given lexeme based on analogical similarities between referents. Such a diachronic process has also been observed in Hock (2003) for ‘morphophonemic’ extension, where the British English ‘intrusive /r/’ is the insertion of a phoneme not historically present in a given lexeme, inserted based on an earlier phonetically conditioned rule of r-deletion post-vocalically; the /r/ phoneme is inserted in phonetic environments where the /r/ would historically be retained in lexemes with a coda /r/, i.e. “the idea is” > “the idear is” (Hock 2003: 442).

An example more closely related to C-agr might be Mithun (2003), whose study of Yup’ik “shows that the evolution of the past contemporative is part of a larger constellation of similar processes, all involving the extension of deri-vational sufffijixes to new functions as inflectional mood markers” (2003: 561). Mithun’s account of Yup’ik provides additional evidence for the reanalysis (or grammaticalization) of derivational morphology as more-grammatical inflec-tional morphology, similar to the process of grammaticalization observed for C-agr. Furthermore, she provides supporting evidence that the underlying structure of feature-marking may be altered through analogical processes. Most relevant to this section is that the accounts of Fortson (2003), Hock (2003) and Mithun (2003) all rely heavily on the extant structure in the grammar to deter-mine in large part the process of reanalysis and extension in language change.

Restricting the discussion now to examples of syntactic change in subject agreement, we see that innovative inflectional and pronominal elements—afffijixes and pronouns—are widely attested across varieties of West Germanic, and moreover are frequently drawn from surface level phonetic efffects, and incorporated into aspects of the competence grammar. De Vogelaer (2010) charts two possible paths of reanalysis—pronominal and inflectional—for the reanalysis and analogical extension of phonetically derived, innovative inflec-tional forms. The locus for reanalysis is the subject position right-adjacent to the syntactic node C, in subject-verb inversion contexts for main clauses, and also subordinate clauses in V2 languages like Dutch and German. In example (3) below, Standard-like Dutch gaan we “go we” exhibits phonetic assimilation,

43Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

where the labial point of articulation in the pronoun we assimilates to the nasalization of the preceding /n/ to yield a phonetic realization of the pronoun as me (De Vogelaer 2010: 15).

(3) Ga=me naar Brussel? (Flemish, Brabantic, Zeelandic)Go=3pl to Brussels‘Are we going to Brussels?’

Due to common inheritance, varieties of Dutch and also German all share this phonetic environment, since infijinitival verbal forms and many plural conjuga-tions end in –n, and the 1pl pronoun begins with a labial (or labial-dental). This assimilation is not surprisingly attested across many varieties of Dutch and German, such that many southern German dialects have 1pl pronoun mir derived from wir (see also Fertig 2000: 47).

The next phase in De Vogelaer’s account is the analogical extension of the innovation into one of two ‘paths’—inflectional (4) or pronominal (5) (2010: 15).

(4) . . . da=me naar Brussel gaan. (Flemish, Brabantic, Zeelandic). . . that=3pl to Brussels go‘. . . that we are going to Brussels.’

(5) Me gaan naar Brussel. (Flemish, Brabantic, Zeelandic)We go to Brussels‘We are going to Brussels.’

While the me form in (3) is phonetically-conditioned, neither the complemen-tizer da or dat in (4) nor the topical position in (5) have the same phonetic environment to yield the me form through a phonetic process. Rather, the pho-netically derived pronoun in (3) was reanalyzed as the basic form, and was thereafter extended to use in other syntactic positions and functions. Key to this analysis is the ambiguity of the me form in (3), which appears in a posi-tion right-adjacent to C, which is occupied either by inflectional afffijixes or sub-jects. The innovative form, being interpreted as one of two extant categories as either inflectional or pronominal, is extended based on the existing syntactic structures and parameters of the language.

In looking at the historical source for inflectional morphology in C-agr contexts, Fuß (2004) provides data from Bavarian that parallels the inflec-tional path of reanalysis and extension, as shown in (3) and (4). As in the development of 2sg inflection demonstrated in (1a) and (1b), the innovative

44 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

–st in (1b) originates in a position right-adjacent to C. The t in the –st inflection is derived from the reduced 2sg pronoun du, and the s is a hiatus efffect derived to bridge points of articulation between the alveolar /n/ and the dental /d/ (see also Weise 1907: 201–202 for similar arguments). Fuß also proposes a similar analysis for the development for Bavarian 2pl –ts inflection in C-agr contexts, where the s is similarly derived from a dialectal 2pl pronoun ës, with the t appearing as a hiatus efffect (Fuß 2004: 60–61). The example of –ts as a parallel development to the development –st, which in addition to adding weight to his historical account, provides evidence that 2sg –st is derived through a his-torical process of reanalysis, and not through a simple analogous extension of the verbal –st to C-agr contexts. In the historical development of –st and –ts, it can be seen that both pronominal and hiatus efffects are reanalyzed as afffijixes, and that the reanalysis follows a predictable path as dictated by the extant grammatical structures, consistent with De Vogelaer (2010).

Consistent with the data from Bavarian, Flemish, Brabantic and Zeelandic presented in this section, whg provides a similar example of innovative inflec-tional morphology, where innovative inflectional afffijixes are derived from hia-tus efffects and internalized in the grammar. This line of argument provides an empirical basis for the methodology and presentation of the data in §3 and §4. Difffering from the data presented by De Vogelaer and Fuß, however, whg data exhibit the reanalysis of hiatus efffects without the co-occurring reanalysis of pronouns. This discrepancy will be discussed further in §5.

2.1 Phonetic Analysis as a Diagnostic of Reanalysis

As noted in §1 and §2, the development of C-agr is derived through the reanal-ysis of subject pronouns and/or phonetically derived hiatus efffects. Evidence of ‘clitic doubling’ (cd) has been commonly used as a diagnostic for the completeness of reanalysis of the original pronoun (Schweizer 2008: 691, my translation):

Charakteristisch für den Stand der zimbrischen Entwicklung ist de Tatsache, dass häufijig neben den enklitischen Formen die Vollformen der-selben Pronomina im gleichen Satz auftreten, ohne dass sie besonders betont wären. Dies zeigt wohl deutlich, dass die Enklitika schon völlig zum Verbum gerechnet werden.

Characteristic for the state of Cimbrian development is the accepted fact that the full form of the same pronoun appears with the enclitic form, in the same clause, without it being particularly emphatic. This shows

45Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

entirely clearly, that the enclitic has already completely been counted as part of the verb.

However, this reanalysis of the original pronoun is often not apparent until already into phase ii of the C-agr Cycle, in which the reanalyzed pronoun is necessarily reinforced by an innovative marker (see Figure 3.2). Thus, while attestation of cd is certainly a reliable diagnostic for the reanalysis of the origi-nal pronoun, it still tells us more about the compensatory process occurring in the middle of the cycle, and not as much about the initiation of the cycle. For that, we turn to acoustic analysis.

In his study on C-agr in West Frisian, de Haan (2010) conducted a percep-tion study, alternating the directionality of phonetic assimilation with respect to the pronoun, in C-agr and subject-inversion contexts. In examples (6a) and (7a), the [voiceless] feature of the /s/ exhibited progressive assimilation into the 2sg pronoun do, efffectively resulting in a devoiced surface form –to, which was enclitic to the element in the clause to which the pronoun was right-adjacent. Alternatively, examples (6b) and (7b) exhibited regressive assimila-tion of the [voiced] feature of the 2sg pronoun do, such that the preceding /s/ was efffectively voiced (de Haan 2010: 227).3

(6) a. Miskien [moas-to] Pyt helpe

Perhaps must-2sg Pyt helpb. *Miskien [moaz=do] Pyt helpe

Perhaps must=2sg Pyt help‘Perhaps you should help Pete.’

(7) a. [da-sto] Pyt helpe moatst

that-2sg Pyt help mustb. *[da=zdo] Pyt helpe moatst

that=2sg Pyt help must‘that you should help Pete’

In both C-agr and subject-inversion contexts, consultants preferred instances of progressive assimilation, evaluating (6a) and (7a) as grammatical, and conversely (6b) and (7b) as ungrammatical. De Haan concludes that the

3  Morphological parsing is here added to De Haan (2010) to illustrate the diffference between the afffijixes in (6a) and (7a) and the subject clitics in (6b) and (7b). The complementizer dat, ‘that’, loses its fijinal –t in the phonetic environment in (7a) and (7b).

46 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

preference for progressive assimilation among his consultants supports the claim that subject pronouns have been reanalyzed as inflectional afffijixes, argu-ing that (2010: 227):

If syntactic incorporation gives rise to an adjunction structure which is interpreted in the phonological component as one phonological word, then the occurrence of progressive assimilation is explained, since pro-gressive assimilation only applies within phonological words, and not over word boundaries. The unacceptability judgments [sic] with regres-sive assimilation . . . demonstrate that syntactic incorporation is obliga-tory in the case of the order V/C-do.

This line of argument is similar to Somers Wicka (2009), who shows that reanal-ysis of word boundaries from lexical to phonological allows for phonological processes to occur within pwords. However, whereas Somers Wicka shows that regressive phonetic efffects (i.e. umlaut) signal a post-lexical process, de Haan argues that progressive voicing assimilation evident in the surface form is indicative of processes in the lexicon. Both limit phonological efffects to within the word boundary, but the evidence for progressive assimilation in de Haan (2010) provides evidence that the word boundary is established at an earlier stage of derivation—or even prior to derivation—in the lexicon. Similar to van Gelderen (2011), this locates reanalysis in the lexicon, as a reanalysis of syntac-tic features inherent on the lexeme, and not merely as a product of syntactic derivation.

Applying the fijindings of de Haan (2010) to varieties of German, acoustic analysis of progressive versus regressive voicing assimilation in produced speech provides a diagnostic for the reanalysis of subject pronouns as inflec-tional afffijixes in C-agr contexts. Attested varieties of German C-agr in 2sg—including Bavarian, East Franconian and whg—exhibit an –s hiatus efffect in addition to a dental component derived from the 2sg pronoun du. Progressive voicing assimilation signifying the reanalysis of the subject pronoun should result in the devoicing of the /d/ in du to a /t/, yielding an –st inflection, as in (1b) reproduced here as (8).

(8) Wenn-st kummst (Bavarian; Bayer 1984: 233)if-2sg [pro] come‘If you come’

This orthographic representation of progressive voicing assimilation sug-gests that the subject pronoun has been reanalyzed as part of the inflec-tional afffijix, since the progressive assimilation signals a phonological process

47Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

within a single lexeme. Furthermore, this phonetic diagnostic identifijies the reanalysis of the subject pronoun without cd, thereby identifying the reanaly-sis of subject pronouns occurring in phase I of the C-agr cycle, rather than in phase ii.

However, regressive assimilation would signal a post-lexical phonological process akin to Somers Wicka (2007, 2009), in which word boundaries were not lexical but rather prosodic. Alternatively, the lack of progressive assimila-tion would signal the presence of a word boundary between the hiatus efffect and the subject pronoun, which blocks phonological processes from occurring. Just such an example of the latter is provided by Weise (1907: 200, cf. also Kisch 1905: 59) from Mosel Franconian and Nösnisch (so-called ‘Transylvania Saxon’, a Franconian language island in Transylvania) shown in (9).

(9) Wenn-s de willst (Nösnisch)if-2sg you will‘If you wish’

Contrasting with (8), the orthographic representation in (9) shows not only a separate parsing of the 2sg pronoun de from the inflected complementizer wenns, but also shows maintenance of the underlyingly voiced phonological form. As contrasted with (8) in which the subject pronoun has been reanalyzed as part of the inflectional afffijix –st, the lack of progressive voicing assimilation in (9) shows that 2sg de maintains its pronominal status. Additionally, (9) pro-vides at least orthographic evidence from multiple German dialects that hiatus efffects may be reanalyzed as inflectional afffijixes in C-agr contexts independent of the reanalysis of subject pronouns.

Measuring voicing distinction by voice onset time (vot) for Modern Bavarian, East Franconian and whg provides a more reliable, empirical basis for the directionality of voicing assimilation than is possible with orthographic records or impressionistic evaluations, though the theoretical basis remains the same. Voicing distinction in German is a privative system in which the binary distinction between the features [voiced] and [voiceless] is signaled by the presence or absence of the mechanical feature [spread glottis] (Avery and Idsardi 2001, Iverson and Salmons 2006). Known as a ‘glottal width’ (gw) sys-tem, the spread glottis mechanic used in German results in a negative vot—that is, a delay between closure and release of a voiceless stop—which serves as a perceptual cue to the listener for [voiceless]. The vot value for a voiceless /t/ averages roughly 75ms, and is often accompanied with aspiration, or a pufff of air resultant from the build-up of pressure during the closure time (Lisker and Abramson 1964). This contrasts with the neutral gesture, which has little or no delay such that vot values for /d/ hover around 0ms. In the absence

48 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

of the negative vot, there is no perceptual cue to interpret the consonant as voiceless, and the default in a gw system is [voiced].

If the du pronoun has been reanalyzed as an afffijix, as in (8), then this should be confijirmed by vot values for /t/ consistent with a gw language, ranging from 25–100ms, with the average value around 75ms (Lisker and Abramson 1964: 401). However, if the subject pronoun maintains its pronominal status, then vot values should be below the 20–25ms range. Approximately 20ms is the threshold of perception for auditory cues, and any vot values below that value would not register with listeners.

3 Methods

3.1 Elicitation

Data were gathered during separate interviews with 4 profijicient speakers of whg in 2011 and 2012, in a region of Eastern Wisconsin encompassing Manitowoc, Calumet, Fond du Lac and Sheboygan counties. Interviews were conducted in English, Standard German, and whg, and all interviews were digitally recorded. All consultants were male, and in their 70’s or 80’s. All are Wisconsin-born, ranging from third to fijifth generation German-American, with consultants’ ancestors emigrating to Wisconsin in the middle or late 19th century. Following Rothman’s defijinition of a heritage lan-guage, these speakers acquired a variety of German as children, and spoke the heritage variety predominantly at home or in limited social domains, within a culture where another language was dominant (2009: 156). In addi-tion to whg, these consultants were early bilinguals, acquiring English no later than fijirst grade with formal instruction in the public school system. Exposure to multiple varieties of German was also common in the com-munity due to the immigration from multiple parts of Germany (and prior to 1871, the German Empire). Exposure to Standard German through formal instruction, print media or continued social ties to Europe is also common, though the spoken variety was whg.

Data were elicited through English-to-German translation tasks. Consultants were asked to translate up to 52 sentences, of which 8 were relevant to this study; the rest were either relevant to other studies, or were distractors. The 8 sentences provided below in Table 2.1 were designed to test for C-agr in the most commonly attested person and number combinations (1pl, 2sg, 2pl), and also to test for rarely attested C-agr in third person. Translation prompts also tested for C-agr in various contexts on complementizers (1–7), causational conjunctions (8), and indirect objects (9–10).

49Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

Data were also elicited from interviews with 11 speakers of Bavarian and 7 speak-ers of East Franconian in May and June, 2012. Because C-agr is characteristic of both varieties, consultants were selected to provide a sampling of socio-linguistic factors.4 Consultants in this data set ranged in age from late 20’s to late 80’s, and ranged in education from farmers with a basic education, to an engineering student and PhDs. Consultants ranged in language ability from monolingual dialect speakers to educated polyglots. Additionally, speakers were recorded in both urban and rural settings, with Bavarian speakers being recorded in Regensburg and Hof am Regen, and East Franconian speakers interviewed in Bayreuth and Harsdorf.

Data were elicited from both Bavarian and East Franconian speakers using a handful of tasks, including: acceptability judgments on a 5 point Likert-like scale, with 1 being “something I’ve heard or would say” and 5 being “unnatural;

4  No discernable diffferences were observed in the distribution or realization of C-agr based on sociolinguistic factors. One possible reason for this lack of variation across generations, education level and location (urban/rural) is that all speakers reported using dialect daily, and in multiple contexts. In a stable linguistic community where the language is frequently spoken, variation is not expected to be frequent.

table 3.1 Data elicitation in WHG

Person & Number Elicitation

1 If you go to the store, buy me some chocolates.

2sg Complementizer

2 If you guys are able, can you help me pick vegetables?

2pl Complementizer

3 She said to call if Fred gets sick again. 3sg Complementizer4 If we have time, then we’ll stay for

dinner.1pl Complementizer

5 Do you know whether we have any milk? 1pl Complementizer6 Do you know whether she’s married? 3sg Complementizer7 Ask him whether the neighbors are

staying for dinner.3pl (full np) Complementizer

8 We ate because we were hungry. 1pl Causative9 Look, there’s the man I talked to

yesterday.1sg Indirect object

10 No, not the man you gave the money to! 2sg Indirect Object

50 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

something I would not say or hear in my community”; a narrative, picture elicitation task; and a directed task designed to elicit subordinate clause struc-tures. For the purposes of this study, data is drawn from the acceptability judg-ment task and commentary thereupon—in addition to providing a numerical evaluation, consultants were encouraged to provide their own version of the sentence, which they found most natural. Examples suggested by consultants were recorded, and entered as a 1 on the acceptability scale, since it was identi-fijied as most natural by the native speaker consultant. This methodology, also employed in Bousquette (2013), provides not only the necessary ceiling efffects for designating what is most natural to native speakers, but also provides valu-able data in addition to the structured interview protocol. Acceptability judg-ments from speakers of non-standard varieties of German have been used by Sewell and Salmons (2014), who similarly employ a sliding scale rather than a binary grammatical/ungrammatical distinction, in a study on parasitic gap-ping in Wisconsin. Use of a Likert like rather than binary scale allows for a more nuanced evaluation of the data, recording variation that is inherent in any speech community, but imperative for comparative study of related lin-guistic varieties, especially those that have similar—but not identical—mani-festations of a given phenomenon, as is the case for the varieties with C-agr considered here. Such subtle variation between two related varieties—or between speakers within a community—is often indicative of ongoing lin-guistic change; this is crucially not apparent when using an overly simplifijied, binary scale of grammaticality. Sentences presented for evaluation are pre-sented below in Table 3.2, with the person and number combinations focused on attestations of C-agr in both varieties. These sentences tested for C-agr in a variety of diffferent ‘stems’ to which C-agr inflectional afffijixes, e.g. complemen-tizers (1–2), temporal and locative wh-elements (3–5), comparatives (6), and indirect objects (7–8). Additionally, consultants’ own volunteered versions are provided, as well (9–12). The total number of sentences to be evaluated, includ-ing consultant generated variants, totaled 31, though 12 will be discussed here.

table 3.2 Data eliciation in East Franconian and Bavarian

Elicitation sentence Person & Number

Elicitation

1 Sag’s mir, wennst nach München kommst. 2sg complementizer2 Sag’s mir, wennst du nach München kommst. 2sg complementizer

(cd)

51Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

3.2 Analysis

Data were fijirst analyzed for morphological distribution of C-agr. In whg this was measured by the presence of inflectional afffijixes in morphologically restricted domains (e.g. 2sg) in subordinate clauses during the translation task. For data collected during interviews with Bavarian and East Franconian speakers, morphological distribution was evaluated based on values given dur-ing the acceptability judgment task. Using a 5 point Likert-like scale, average evaluations ranging from 1 to 2.5 show a relatively high degree of acceptability, with little variation across speakers. Such was the case for most instances of C-agr in 2sg (cf. Bousquette 2013: 112). Example sentences with a low degree of acceptability contrast starkly: among all Bavarian and East Franconian speakers interviewed, clitic doubling of the sort Mia gem-ma hoam ‘We are going home’ was given an average value of 1.81, while instances of clitic tri-pling in Mia gem-ma mir hoam ‘We are going home’ was given an average value of 4.92 (Bousquette 2013: 126). Valuations are therefore not absolute or binary

Elicitation sentence Person & Number

Elicitation

3 Kannst du mir sagen, wannst ankommst? 2sg temporal wh-element

4 Kannst du mir sagen, wost wohnst? 2sg locative wh-element5 Könnt ës mir sagen, wots wohnts? 2pl locative wh-element

(Bavarian)6 De Hans is gresser als wiast du bist. 2sg comparative7 Weisst du, wemst das Geld gegeben hast? 2sg indirect object8 Weisst du, wemma das Geld gegeben haben? 1pl indirect object

Consultant generated9 Is des de Mann, mit demst gret hast? 2sg indirect object

(East Franconian)10 Is des de Frau, mit derst gret hast? 2sg indirect object

(East Franconian)11 Is des de Mann, mit dem woistu gret hast? 2sg indirect object

(Bavarian)12 Is des de Frau, mit derre wost gret hast? 2sg indirect object

(Bavarian)

52 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

(e.g. grammatical or ungrammatical), though the aggregate data do provide evi-dence for what examples are largely accepted by native speakers, versus those that speakers categorically reject. Data from continental varieties were then compared to data from Wisconsin, with respect to the distribution of C-agr phenomena, both in terms of which person-number combinations exhib-ited C-agr, as well as which lexical stems were able to host C-agr inflectional morphology. Analysis of modern Bavarian and East Franconian data provide not only an independent assessment of C-agr that should confijirm recent lit-erature on the topic, but also provide responses with largely correlative data sets against which the contemporary whg may be evaluated.

Once the morphological distribution of C-agr was determined for each variety, acoustic analysis of vot was used to test for progressive voicing assimilation, which evaluates whether reanalysis of the subject pronoun as an inflectional afffijix has occurred. Data from East Franconian and Bavarian were spot-checked to confijirm the presence of progressive voicing assimilation represented orthographically in the literature of these current varieties (e.g. Bayer 1984), but whg data was completely analyzed for all instances of C-agr. Additionally, since C-agr was found to occur only in 2sg, comparative values from East Franconian and Bavarian were restricted to this part of the paradigm.

4 Data

4.1 Morphological Distribution

Consistent with recent literature, data show that East Franconian licenses C-agr for 2sg, while Bavarian licenses C-agr also for 2pl, and also 1pl for some speakers. As noted in §3.2, whg like East Franconian restricts C-agr to 2sg,

table 3.3 Average acceptability ratings in C-agr contexts

Valuation

Elicitation East Franconian Bavarian

1 Sag’s mir, wennst nach München kommst. 2.5 1.52 Sag’s mir, wennst du nach München kommst. 2.17 2.273 Kannst du mir sagen, wannst ankommst? 1 1.754 Kannst du mir sagen, wost wohnst? 1.7 1.445 Weisst du, wemst das Geld gegeben hast? 1.3 1.44

53Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

so for the sake of equal comparison, the rest of the discussion will be focused on 2sg. Shown in Table 3.3 below are numerical values for acceptability judg-ments of C-agr for 2sg on complementizers (1–2), temporal wh-elements (2), locative wh-elements (3) and indirect objects (4), comparing average values of East Franconian and Bavarian responses.

Average values for C-agr in 2sg are between 1 and 2 in both East Franconian and Bavarian for almost all stems. The largest diffference in valuation between the two varieties is the valuation on the complementizer wennst by East Franconian speakers in (1), which was valued on average as 2.5. This can be accounted for in light of the valuation of cd instances in (2) which is similar to (1) from Table 2.3, but involves the co-occurrence of C-agr inflection and a rein-forcing subject pronoun, du. On average, East Franconians rated this sentence at 2.17, whereas Bavarian speakers rated it 2.27; East Franconian speakers pre-ferred the cd version, whereas Bavarian speakers preferred the non-cd vari-ant. As argued in Bousquette (2013: 108), this discrepancy may reflect a further progression through the C-agr Cycle in East Franconian than in Bavarian. In general, though, valuations on a scale of 1 to 5 are all on the more ‘natural’ end of the scale, and based on this data are assumed acceptable in these varieties.

Production data from whg similarly exhibit C-agr on the complementizer wenn “if, when”, as in (10)–(12). Additionally, C-agr appears on the whg comple-mentizer af “if, whether” (cf. Standard German ob, Standard Dutch of ), as in (13).

(10) Wenn-s du in de Kaufhaus gehst, dann kauf dich . . .  if-2sg you to the store go then buy you . . . 

kauf mich Milich (whg, Speaker O) buy me milk.

‘If you go to the store, then buy me milk.’

(11) Wann-s du nach . . .  na-m Store gehst, if-2sg you to . . .  to-the store go

d’n kauf mi Milich. (whg, Speaker R) then buy me milk. ‘If you go to the store, then buy me milk.’

(12) Wenn-s du noar Store komms . . .  (whg, Speaker V) if-2sg you to store come . . .  ‘If you go to the store . . .’

54 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

(13) Kanns du mir sage af-s du morge komms?

Can you me.dat say if-2sg you tommorow come (whg, Speaker R) ‘Can you tell me if you are coming tomorrow?’

In addition to exhibiting C-agr on complementizers, some speakers addition-ally inflected indirect objects that had been topicalized in subordinate clauses, similar to sentences (7)–(12) in Table 2.2, as shown in (14).

(14) Net de Mann wo-s du ‘s Geld geben hast!

Not the man whom-2sg you the money given have (whg, Speaker R) ‘Not the man you gave the money to!’

Difffering from continental varieties of German discussed here, however, (14) shows the use of wo as an object pronoun which does not mark case, number or gender (though may mark for animacy in opposition to was “what” or das/

dat “that”). Nevertheless, wo exhibits C-agr inflection consistent with the sub-ject agreement marking in (10)–(13), and in the same syntactic position and on the same category of stem as East Franconian and Bavarian, as in sentence (5) from Table 3.3. Although whg consultants were not tested for C-agr in wh-elements or comparatives, the production data exemplifijied in (10)–(14) show morphological distribution of C-agr in 2sg that is consistent with the realiza-tion of C-agr in 2sg in continental East Franconian and Bavarian. It is there-fore concluded based on these data that these four speakers of whg exhibit C-agr, and that the inflectional afffijix is an –s.

In addition to providing examples of where C-agr does occur in whg, it is also benefijicial to provide examples of where C-agr does not occur. Particularly because it has been argued in §2 that the –s component of C-agr inflection for 2sg is originally derived through a reanalysis of a hiatus efffect, it must be made clear that the –s in (10)–(14) is morphologically restricted to 2sg, and not that the –s actively and regularly occurs as a hiatus efffect in a given phonetic envi-ronment. Exhibited below in (15a) and (15b) are two translations provided by the same speaker, in which the –s inflectional afffijix appears in second person singular (15a), but not second person plural (15b), despite the same phonetic environment between the coda /n/ of the complementizer wenn and the /d/ onset of the pronouns du (2sg) and dir (2pl).

55Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

(15) a. Wenn-s du in de Kaufhaus gehst . . .  (whg, Speaker R)if-2sg you in the store go‘If you go to the store . . . ’

b. Wenn dir zwei Zeit habm . . .  (whg, Speaker R)if you.pl two time have‘If you two have time . . . ’

The same phonetic environment occurs due to the fact that whg Speaker R employs the Franconian dialectal pronoun dir for 2pl, which has been attested not only in Franconian-speaking regions of modern Germany (Weise 1907), but has also been attested as continuing in Franconian-speaking language islands outside of Germany (Böttger 1904). Speaker ab, an East Franconian speaker recorded in Regensburg, also exhibits this dialectal dir pronoun, and likewise restricts the marking of C-agr to 2sg, as exhibited in the diffference between (16a) and (16b).

(16) a. Kannst du mir sagen wo-st wohnst?

can you me.dat say where-2sg live (East Franconian, Speaker ab) ‘Can you tell me where you live?’

(16) b. Könnt (d)hir mir sagen wo dir wohnt?5 Can you.pl me.dat say where you.pl live (East Franconian, Speaker ab)

‘Can you two tell me where you live?’

Consistent with the pattern in whg in (15a) and (15b), the East Franconian speaker ab restricts marking of C-agr to 2sg; despite the fact that the same loc-ative wh-element stem wo “where” appears in both (16a) and (16b), and despite the fact that the phonetic environment between the coda vowel in wo and the /d/ onset of the pronouns du and dir, C-agr is restricted to 2sg.

The evidence presented in (15a–b) provides evidence that C-agr in whg is morphologically restricted to 2sg, with the original hiatus –s having been reanalyzed as inflectional morphology, as exhibited in (10)–(14). Additionally,

5  Speaker ab provided two variants of the 2pl pronoun, one dir and one ihr in the matrix clause; the dir pronoun in unstressed position may appear as de.

56 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

the comparison of (15a–b) to (16a–b) shows that the morphological distri-bution of C-agr exhibited in whg has a possible contemporary analogue in Franconian.

4.2 Phonetic Analysis of vot

Following the theoretical assumptions and methods outlined in §2.1, vot val-ues were measured on the initial dental of the 2sg pronoun du in C-agr con-texts. Data from East Franconian and Bavarian were used fijirstly to test whether orthographic representations of progressive assimilation as in (8) were accu-rate. If orthographic and recorded data were consistent, they would secondly together provide a baseline against which to measure whg. For this purpose, two speakers were selected: ab (East Franconian) and hds (Bavarian).

Speaker ab was selected because his speech exhibits many shared charac-teristics with whg speakers, especially Speaker R (e.g. pronominal use, cf. (15) and (16)). Speaker hds was selected because his speech includes the use of wo as a relative pronoun as a host for C-agr inflection in restricted contexts (cf. (11) in Table 2), similar to the use of wo in whg by Speaker R in (14). vot data from 2sg C-agr for both speakers is given in Table 3.4.

Though based on a small number of tokens specifijic to the given part of the paradigm, all tokens from both the East Franconian and Bavarian speaker fall within the 25–100ms vot range expected for /t/. The average vot values addi-tionally fall within close proximity to the 75ms average, with some variation allowed between speakers, and depending on rate of speech. These data show consistent devoicing of the /d/ in pronominal du to a /t/, resulting from pro-gressive assimilation of the [voiceless] feature of the preceding hiatus –s. Such phonological efffects are only possible within word boundaries, and following de Haan (2010), the presence of progressive assimilation signals a lexical rather than post-lexical process. Available acoustic data for East Franconian and

table 3.4 Voice Onset Time: East Franconian and Bavarian

Speaker Location of Recording

Variety # of Tokens vot Range vot Average

ab Regensburg East Franconian

5 46–62ms 53.4ms

hds Regensburg Bavarian 6 34–93ms 69.7ms

57Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

Bavarian provide evidence for reanalysis of subject pronouns as inflectional afffijixes in C-agr contexts, and additionally confijirm the orthographic represen-tation in (8).

Analysis of vot data from 2sg C-agr in whg is provided in Table 3.5, with data presented for vot range and average vot value for all four speakers.

Contrasting data from the continental East Franconian and Bavarian speak-ers, speakers of whg exhibited much lower vot values for the initial [d] in pronominal du. The range of vot values were much lower than for speakers ab or hds, with each whg speaker producing at least one token that was below the 20ms threshold of perception. Even the average vot values for whg speakers hovered around the minimum value of perception, being far below the 75ms average for the /t/ phoneme. Based on these data for 2sg C-agr in whg, the pronoun du has not been reanalyzed as an afffijix. The absence of any phonological process of assimilation provides evidence that a word bound-ary exists. The absence specifijically of progressive assimilation shows that the subject pronoun has not been reanalyzed as an inflectional afffijix, and therefore maintains its status as a separate lexeme and pronominal subject.

5 Summary and Discussion

In §1, I presented the argument that C-agr may develop historically through either the grammaticalization of subject pronouns as inflectional afffijixes, or through the reanalysis of phonetic material—including phonetically derived hiatus efffects—as inflectional afffijixes. Whether due to grammaticalization or reanalysis of surface forms, all reanalysis occurs in the lexicon, and is tied to the selection of diffferent features out of the lexicon. Furthermore, both the grammaticalization of subject pronouns and the reanalysis of surface forms initiate a linguistic cycle specifijic to C-agr, known as the C-agr cycle, with

table 3.5 Voice Onset Time WHG

Speaker Location of Recording

# of Tokens vot Range vot Average

J Sheboygan 3 15–27ms 19.7msO New Holstein 2 17–40ms 28.5msR Marytown 2 0–31ms 15.5msV Elkhart Lake 2 0–13ms  6.5ms

58 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

the former initiating the cycle through a push-chain process, and the latter through a pull-chain.

Focusing on the reanalysis of surface forms in §2, the discussion turned to the restriction of phonological processes within word boundaries. This restriction allows for the use of acoustic analysis as a diagnostic of underly-ing syntactic structures, specifijically in the case of the reanalysis of subject pronouns as afffijixes: since progressive assimilation can only occur within lexical word boundaries, the evidence of progressive voicing assimilation on subject pronouns in C-agr contexts shows that the subject pronoun has been reanalyzed as an inflectional afffijix. With this assumption in mind, §3 describes the methodology for conducting fijieldwork and data collection in Eastern Wisconsin and among East Franconian and Bavarian-speaking populations in Germany. In light of the theoretical assumptions outlined in §2, data is pre-sented in §4, showing that whg difffers from the continental varieties in that whg exhibits C-agr without the reanalysis of subject pronouns; C-agr inflec-tion is rather derived entirely on the reanalysis of phonetically derived hia-tus efffects, that is, the –s inflection restricted to 2sg. To the knowledge of the author, whg is the only contemporary variety of West Germanic that exhibits C-agr without the concomitant reanalysis of subject pronouns. This singular-ity bears on the historical development of C-agr in modern varieties of West Germanic. Additionally, these data provide evidence that grammaticaliza-tion of lexical categories and reanalysis of surface forms can independently initiate C-agr.

In terms of diachrony, whg data are consistent with Weise (1907), in which orthographic representations of C-agr clearly marked not only word bound-aries, but also the maintenance of phonological features such as [voice] that in this study would signify the maintenance of pronominal status and the absence of reanalysis. This is shown in (9), reproduced here as (17), in which word boundaries between the inflected complementizer and pronoun are clearly marked, and the subject pronoun retains the feature [voice] (Weise 1907: 200, cf. also Kisch 1905: 59).

(17) Wenn-s de willst

if-2sg you will‘If you wish.’

In looking at data from other dialect studies undertaken around the same time as Weise (1907), one also fijinds competing orthographic representa-tions of C-agr inflection. Thus, while Weise and Meyer (1898) present data similar to whg in the maintenance of pronominal status, still other varieties

59Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

are recorded that exhibit progressive assimilation denoting reanalysis, as in Schwäbl (1903). Adding to the variation exhibited in dialect studies around this time is Philipp’s (1897: 49) study of the dialect spoken in Leipzig, which shows both reanalysis (wennste “if you”) and maintenance (äärschde “before you”, obsde “whether you”, wäälsde “whether you”) of the subject pronoun within the same dialect. Because the alternation of [t] and [d] is expected in the historical development of C-agr, this variation in the orthography can be taken at face value, as evidence that reanalysis of the subject pronoun as an inflectional afffijix was underway in multiple continental varieties of German in the late 19th and early 20th centuries. More specifijic to the situation in east-ern Wisconsin: analysis of the immigration records of the ancestors of mod-ern speakers of whg with C-agr locates the input varieties in (West) Central Franconian regions, with speakers’ immigrant ancestors being born in in these regions between 1819 and 1854, and a number of non-standard elements con-sistent with those input varieties are present in modern whg (Bousquette, 2014). Consistent with Nützel and Salmons (2011: 710), C-agr in whg reflects maintenance of the structure present in the input variety at time of immi-gration. In light of these data, whg is positioned as one of many varieties that exhibit the sort of pull-chain initiation of C-agr through the reanalysis of surface forms. At the same time, whg does not exhibit the same progres-sion of the C-agr Cycle beyond the initial stage as is exhibited by orthographic variation in 19th and 20th century continental dialect grammars which is now comparatively progressed in recordings of contemporary speakers in the 21st century; while modern continental varieties exhibit a more advanced state of language change with respect to C-agr, whg exhibits an earlier stage of language change that modern continental varieties have already undergone (cf. Silva-Corvalán 1994 for a similar phenomenon in Heritage Spanish com-munities in the United States).

As regards the mechanisms afffecting syntactic change, whg shows that the reanalysis of surface forms—including those that are phonetically-derived—may initiate the C-agr cycle independent of any process of grammaticaliza-tion of subject pronouns. This argument is supported by data from Philipp (1897), Meyer (1898), Schwäbl (1903), and Weise (1907), which fijirstly provide supporting evidence that hiatus efffects may be reanalyzed as C-agr inflection; and secondly, provide evidence especially from Philipp (1897) that pronomi-nal reanalysis may follow the reanalysis of hiatus efffects chronologically. This shows that there are two separate processes that may be at work independently but simultaneously, and the independent workings of the one may contribute to the development of the other. This returns us to the issue of the push-chain mechanism of syntactic change exhibited in the grammaticalization model,

60 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

versus the pull-chain model initiated by the reanalysis of phonetic material. This work has provided evidence that C-agr may be initiated by either, provid-ing empirical data especially for the pull-chain process. While this particular data set is not a longitudinal study of a single variant, the whg data—espe-cially in light of late-19th and early 20th century dialect grammars—makes it clear that reanalysis of surface level efffects such as phonetically derived hia-tus efffects may afffect language change independent of other processes of lan-guage change, e.g. grammaticalization, and we as researchers must therefore cast a wider net in our analysis of historical and dialectal data.

whg exhibits an earlier stage of C-agr than that which is attested else-where among modern varieties of West Germanic. This variety—as well as these data—is valuable for the developmental stage attested, and for the ques-tions this data set raises. For this present undertaking, it has been argued that whg as a variety exhibits an earlier stage in the development of C-agr than attested elsewhere, developed through reanalysis of phonetically-derived sur-face forms as an alternate mechanism of syntactic change. This argument was supported by comparative, empirical data from contemporary spoken whg, East Franconian and Bavarian analyzed both for morphological distribution of C-agr, as well as for the directionality of voicing assimilation as a diagnostic of reanalysis based on evidence of maintenance of lexical word boundaries.

References

Avery, Peter and William J. Idsardi. 2001. Laryngeal Dimensions, Completion and Enhancement. In Distinctive Feature Theory, ed. T. Alan Hall, pp. 40–71. Berlin: Mouton de Gruyter.

Bayer, Josef. 1984. comp in Bavarian Syntax. The Linguistic Review 3: 209–274.Bousquette, Joshua. 2013. Complementizer Agreement in Modern Varieties of West

Germanic: A Model of Reanalysis and Renewal. Madison: PhD Dissertation, University of Wisconsin-Madison.

———. 2014. Complementizer Agreement in Eastern Wisconsin: (Central) Franconian Eeatures in an American Heritage Language Community. Language Typology and

Universals / Sprachtypologie und Universalienforschung (STUF) 67(4): 561–588.Böttger, Oswin. 1906. Zum Satzbau der erzgebirgischen Mundart. Leipzig: PhD disserta-

tion, Universität Leipzig.Campbell, Lyle and Alice C. Harris. 1995. Historical Syntax in Cross-linguistic Perspective

(Cambridge Studies in Linguistics 74). Cambridge: Cambridge University Press.Carstens, Vicki. 2003. Rethinking Complementizer Agreement: Agree with a Case-

checked Goal. Linguistic Inquiry 34: 393–412.

61Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

Chomsky, Noam. 2002. On Nature and Language. Cambridge, uk: Cambridge University Press.

De Vogelaer, Gunther. 2010. Morphological Change in Continental West Germanic: Towards an Analogical Map. Diachronica 27: 1–31.

Donaldson, Bruce. 2008. Dutch: A Comprehensive Grammar, 2nd edn. New York: Routledge.

Fertig, David. 2000. Morphological Change Up Close: Two and a Half Centuries of Verbal

Inflection in Nuremberg (Linguistische Arbeiten 422). Tübingen: Max Niemeyer.Fortson, Benjamin W., iv, 2003. An Approach to Semantic Change. In Handbook

of Historical Linguistics, eds. Brian D. Joseph and Richard D. Janda, pp. 648–666. Malden, ma: Blackwell.

Fuß, Eric. 2004. Diachronic Clues to Pro-drop and Complementizer Agreement in Bavarian. In Diachronic Clues to Syntactic Grammar (Linguistik Aktuell / Linguistics Today 72), eds. Eric Fuß and Carola Trips, pp. 51–100. Amsterdam and Philadelphia: John Benjamins.

Gelderen, Elly van. 2007. Defijiniteness Cycle in Germanic. Journal of Germanic

Linguistics. 19.4: 275–308.———. 2011. The Linguistic Cycle: Language Change and the Language Faculty. Oxford:

Oxford University Press.Haan, Germen de. 2010. Studies in West Frisian Grammar: Selected Papers by Germen J.

de Haan. Jarich Hoekstra, Willem Visser and Gofffe Jensma (eds.). Amsterdam and Philadelphia: John Benjamins.

Haegeman, Lilianne. 1992. Theory and Description in Generative Syntax: A Case Study in

West Flemish. Cambridge: Cambridge University Press.Hanna, Patrizia Noel Aziz. 2009. Jespersen’s Cycle and the Issue of Prosodic ‘Weakness’.

In Advances in Comparative Germanic Syntax (Linguistik Aktuell / Linguistics Today), eds. Artemis Alexiadou, Jorge Hankamer, Thomas McFadden, Justin Nuger and Florian Schäfer, pp. 197–217. Amsterdam and Philadelphia: John Benjamins.

Hock, Hans Henrich. 2003. Analogical Change. In Handbook of Historical Linguistics, eds. Brian D. Joseph and Richard D. Janda, pp. 441–460. Malden, ma: Blackwell.

Iverson, Gregory K. and Joseph Salmons. 2006. On the Typology of Final Laryngeal Neutralization: Evolutionary Phonology and Laryngeal Realism. Theoretical

Linguistics 32: 205–216.Jespersen, Jens Otto. 1917. Negation in English and Other Languages. Copenhagen: Host.Kathol, Andreas. 2001. Positional Efffects in a Monostratal Variety of German. Journal

of Linguistics 37: 35–66.Kerswill, Paul and Ann Williams. 2000. Creating a New Town koiné: Children and

Language Change in Milton Keynes. Language in Society 29: 69–115.Kiparsky, Paul. 1996. The Shift to Head-initial vp in Germanic. In Studies in Comparative

Germanic Syntax, vol. ii. (Studies in Natural Language and Linguistic Theory),

62 Bousquette

For use by the Author only | © 2015 Koninklijke Brill NV

eds. Höskuldur Thráinsson, Samuel David Epstein and Steve Peter, pp. 140–179. Dordrecht: Kluwer.

Kisch, Gustav. 1905. Vergleichendes Wörterbuch der Nösner (siebenbürgischen) und

moselfränkisch-luxemburgischen Mundart. Hermannstadt: W. Kraft.Koppen, Marjo van. 2005. One Probe—Two Goals: Aspects of Agreement in Dutch

Dialects. Leiden: PhD dissertation, University of Leiden.Koppen, Marjo van, and Michael T. Putnam. 2009. C-agreement or Something Close to

It: Some Thoughts on the ‘alls-construction’. In Advances in Comparative Germanic

syntax (Linguistik Aktuell / Linguistics Today 141), eds. Artemis Alexiadou, Jorge Hankamer, Thomas McFadden, Justin Nuger and Florian Schäfer, pp. 41–58. Amsterdam and Philadelphia: John Benjamins.

Kuryłowicz, Jerzy. 1965. Zur Vorgeschichte des germanischen Verbalsystems. In Beiträge

zur Sprachwissenschaft, Volkskunde und Literaturforschung: Wolfgang Steinitz zum

60. Geburtstag, eds. A.V. Isačenko, W. Wissmann and H. Strobach, pp. 242–247. Berlin: Akademie-Verlag.

Lightfoot, David. 1979. Principles of Diachronic Syntax. Cambridge: Cambridge University Press.

———. 1991. How to Set Parameters: Arguments from Language Change. Cambridge, ma: mit Press.

Lisker, Leigh, and Arthur S. Abramson. 1964. A Cross-language Study of Voicing in Initial Stops: Acoustical Measurements. Word 20: 384–422.

Meyer, Elard Hugo. 1898. Deutsche Volkskunde. Strasburg: Karl J. Trübner.Mithun, Marianne. 2003. Functional Perspectives on Syntactic Change. In Handbook

of Historical Linguistics, eds. Brian D. Joseph and Richard D. Janda, pp. 552–572. Malden, ma: Blackwell.

Nützel, Daniel, and Joseph Salmons. 2011. Structural Stability and Change in Language Contact: Evidence from American German. Language and Linguistics Compass 5: 705–717.

Oosterhofff, Jenneke A. 2009. Basic Dutch: A Grammar and Workbook. London and New York: Routledge.

Philipp, Oskar. 1897. Die Zwickauer Mundart. Leipzig: PhD dissertation, Universität Leipzig.

Reis, Marga. 1985. Satzeinleitende Strukturen im Deutschen: Über comp, Haupt, und Nebensätze, w-Bewegung und die Doppelkopfanalyse. In Erklärende Syntax des

Deutschen, ed. Werner Abraham, pp. 271–311. Tübingen: Narr.Rothman, Jason. 2009. Understanding the Nature and Outcomes of Early Bilingualism:

Romance Languages as Heritage Languages. International Journal of Bilingualism 13: 145–155.

Schwäbl, Johann Nepomuk. 1903. Die altbayrische Mundart. Grammatik und

Sprachproben. Munich: Lindauer.

63Evidence from Wisconsin Heritage German

For use by the Author only | © 2015 Koninklijke Brill NV

Schweizer, Bruno. 2008. Zimbrische Gesamtgrammatik: Vergleichende Darstellung der

zimbrischen Dialekte. Herausgegeben von James R. Dow. Stuttgart: Franz Steiner Verlag.

Sewell, Alyson, and Joseph Salmons. 2014. How Far-reaching are the Efffects of Contact? Parasitic Gapping in Wisconsin German and English. In: Questioning Language

Contact: Limits of Contact, Contact at its Limits, ed. Robert Nicolai, pp. 217–251. Leiden: Brill.

Silva-Corvalán, Carmen. 1994. Language Contact and Change: Spanish in Los Angeles. Oxford: Clarendon.

Somers Wicka, Katharina. 2007. On Cliticization in Otfrid’s ‘Evangelienbuch’. Madison: PhD dissertation, University of Wisconsin-Madison.

———. 2009. From Phonology to Syntax: Pronominal Cliticization in Otfrid’s Evan-

gelienbuch. Tübingen: Niemeyer.Trudgill, Peter. 2010. Investigations in Sociohistorical Linguistics. Cambridge: Cambridge

University Press.Weise, Oskar. 1907. Die sogenannte Flexion der Konjunctionen. In Zeitschrift für

Deutsche Mundarten vol. 2, eds. Otto Heilig and Philipp Lenz, pp. 199–205. Stuttgart: Franz Steiner Verlag.