John Benjamins Publishing Company

0 downloads 0 Views 188KB Size Report
Fisterran. Galician ir 'go' has analogical past imperfective forms based on the ..... Fisterran Galician, the verb ir “to go” has Imperfect Indicative forms that result.
John Benjamins Publishing Company

This is a contribution from Historical Linguistics 2011. Selected papers from the 20th International Conference on Historical Linguistics, Osaka, 25-30 July 2011. Edited by Ritsuko Kikusawa and Lawrence A. Reid. © 2013. John Benjamins Publishing Company This electronic file may not be altered in any way. The author(s) of this article is/are permitted to use this PDF file to generate printed copies to be used by way of offprints, for their personal use only. Permission is granted by the publishers to post this file on a closed server which is accessible to members (students and staff) only of the author’s/s’ institute, it is not permitted to post this PDF on the open internet. For any other use of this material prior written permission should be obtained from the publishers or through the Copyright Clearance Center (for USA: www.copyright.com). Please contact [email protected] or consult our website: www.benjamins.com Tables of Contents, abstracts and guidelines are available at www.benjamins.com

Analogy as a source of suppletion* Matthew L. Juge

Texas State University-San Marcos

Analogy is often seen as a force that restores morphological patterns disrupted by regular sound change, but analogy also plays a nearly unexamined role in the creation of the most extreme kind of irregularity – suppletion. Fisterran Galician ir ‘go’ has analogical past imperfective forms based on the semantically linked verb vir ‘come’. Regular loss of -n- in certain forms and palatalization in others facilitated a proportional analogy: vir : viña :: ir : X, X = iña, with a nonetymological nasal creating weak suppletion. Analogy not only sometimes fails to promote regularization but in fact runs counter to it. Analogy is a key factor in suppletion and highlights important semantic, psycholinguistic, and neurolinguistic considerations in the development of suppletion.

1. Interactions between sound change and analogy In this paper I challenge the traditional view that analogy creates regularity in paradigms. Cases of both proportional analogy and contamination in non-standard Galician varieties show that analogical change sometimes creates suppletion. The apparent paradox of analogy as a source of both regularity and suppletion is resolved when analogy is seen as a local process of the type described by Joseph (1992). I will focus on Galician and other minority Romance varieties, especially the Rhaeto-Romance languages, in support of my claims.

* Thanks to the conference participants at ICHL 2011, Sharla Nichols Juge, William F. Weigel, and two anonymous reviewers for their input during the development of this paper. Any shortcomings are mine alone.

© 2013. John Benjamins Publishing Company All rights reserved

176 Matthew L. Juge

2. Neogrammarians: Sound change introduces irregularity and analogy restores regularity Although as Morpurgo Davies (1998) points out, the Neogrammarians were not the first to examine analogy, they are commonly associated with the view that analogy is a process that affects paradigms after regular sound changes create ‘irregularities’, as in the Latin noun honor ‘honor’, in which the change of /s/ to /r/ in intervocalic position left the nominative with a final /s/ while the other parts of the paradigm had stem-final /r/, as in Table 1. Such cases of leveling are typically found in coherent paradigms, sometimes with a certain amount of ‘leakage’, that is, a lack of influence on less-closely-related forms, e.g. the derived noun honestās ‘honesty’. Table 1.  Latin honor Nominative Genitive

honōs honōr-is

honor honōr-is

key sound change: s > r / V__V

As Hock (2003) indicates, closer examination reveals a more complicated state of affairs.

3. Suppletion in synchronic/typological perspective While suppletion is discussed mostly in diachronic terms, it has received attention from a synchronic, typological perspective as well. Corbett (2007) typologizes suppletive patterns using a canonical approach. Veselinova (2006) addresses primarily synchronic issues as well. Though there has been debate over how to define suppletion (see Veselinova 2006 for discussion), I will use the following working definition: Two word forms are in a suppletive relationship if their semantic relationship is regular but their morphological relationship is not. Note that this definition does not refer to etymology, which is unavailable to most speakers and in many cases unavailable to linguists as well. Under this view, suppletion is a gradient phenomenon. Strong suppletion involves forms that have no identifiable synchronic phonological relationship, as in Spanish va “s/he goes” and fue “s/he went” (compare regular canta ‘s/he sings’ and cantó “s/he sang”). Weakly suppletive forms, on the other hand, share some phonological material, as in the Present Indicative forms of Latin facere: faciō (active), fiō (passive). My focus here is suppletion in verb stems. © 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 177

4. Sources of suppletion At least five sources of suppletion have been documented, as shown in Table 2. Of these, the most familiar (and the only ones mentioned in Corbett’s (2007: 13) citation of Juge (1999), which also discusses others) are incursion and sound change. Incursion is the entry of forms from one lexeme into another, as in English go, where went replaced the form ēode (which itself was in a relationship of strong suppletion with the other forms of the verb). Two other sources of suppletion involve interactions between lexemes. Coalescence is the creation of a single lexeme out of one or more previously distinct lexemes, as in Latin esse “be”, whose paradigm contains elements of PIE *es- “be” and *bhuH- “become”. Lexical merger is the loss of forms from two or more lexemes with a single resulting lexeme, as in French être “be”, which features forms from Latin esse “be” and stāre “stand”. Juge (1999), however, also includes two types of analogy – proportional analogy and contamination – and Ronneberger-Sibold (1990) adds a third analogical source, leveling. Table 2.  Sources of suppletion Source

Language

Example

Gloss

incursion coalescence lexical merger sound change leveling contamination proportional analogy

English Latin French English English Galician Galician

go~went sum~fuī suis~été am~is house~houses facer~feña ir~iña

go be be be house do go

4.1

Prototypical suppletion – etymologically distinct roots

The most familiar cases of suppletion involve two or more etymologically distinct roots, as in Table 3. As indicated above, there are at least three scenarios in which suppletive paradigms develop with forms from two or more separate roots, all of which are termed incursion in Juge (1999). Table 3.  Suppletion involving two etymologically distinct roots English Spanish Surmeiran

go ir inf eir inf

went voy 1sg.pres.ind vign 1sg.pres.ind

fui 1sg.pret.ind vo 3sg.pres.ind

© 2013. John Benjamins Publishing Company All rights reserved

giagn 1pl.pres.ind

178 Matthew L. Juge

4.1.1 Coalescence After go~went, the verb to be is probably the most frequently cited instance of verbal suppletion. In the Indo-European languages, copulas tend to result from the merger of verbs with meanings like be (*h1es-) and become (*bhuH-). In Latin, for example, the verb esse has forms from each of these roots (e.g. es 2sg Present Indicative, fuistī 2sg Perfect Indicative). In this case, two formerly distinct verbs provide the forms for a single verb. Here the roots differ according to tense and aspect, but Latin also has a case of coalescence with verbs differing in voice: faciō (< *dheh1- “do, put”) “make” and fiō (< *bhuH-) “become” are in a suppletive relationship, with the latter serving as the passive of the former. 4.1.2 Lexical merger Closely related to coalescence is lexical merger, the joining of some forms of two or more verbs with (nearly) complete paradigms. French être, for example, has forms from two Latin verbs, esse “be” and stāre “stand”, that in some other Romance varieties (e.g. Ibero-Romance) remain distinct with full paradigms (see also Section 6). Similarly, many Romance motion verbs involve the lexical merger of two or more etyma. French aller, for example, combines forms of Latin īre “to go”, vādere “to go hastily or rapidly”, and forms of uncertain origin. Among the candidates for the final group of French forms are ambulāre “to walk” and *ambitāre “to walk”, both of which present phonological difficulties, and the unattested root *allāre, which probably meant “to walk”. Depending on historical attestation, the distinction between coalescence and lexical merger may in some cases be difficult or impossible to make. 4.1.3 Incursion Suppletion caused by incursion involves the replacement of forms of one lexeme by forms from another lexeme. Surely the best known example of incursion is the replacement of English ēode by went, the other forms of which (and a regularized past) continue in wend. A particularly notable case of incursion is the replacement of forms of Latin īre “to go” in the Perfectum by corresponding forms of esse “to be” in the Ibero-Romance languages (see Table 4). This instance of incursion also created overlapping suppletion, the situation in which suppletive forms of two or more lexemes are shared (see Juge 1999 for further discussion of overlapping suppletion and Section 6 below for semantic factors). While the history of īre and esse is clear, there may be cases in which the distinction between lexical merger and incursion seems to be a matter of degree, although cases of suppletive imperatives seem to be overwhelmingly instances of incursion (see Veselinova 2006 for more on suppletive imperatives).

© 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 179

Table 4.  Suppletive verb forms in Galician ser “be”

Form

ir “go”

haber “have”

Indicative Subjunctive Indicative Subjunctive Indicative

Subjunctive

1sg pres 2sg pres 3sg pres 1pl pres 2pl pres 3pl pres

son es é somos sodes son

sexa sexas sexa sexamos sexades sexan

vou vas vai imos ides van

vaia vaias vaia vaiamos vaiades vaian

hei has ha habemos habedes han

haxa haxas haxa haxamos haxades haxan

1sg past 2sg past 3sg past 1pl past 2pl past 3pl past

fun fuche(s) foi fomos fostes foron

fose foses fose fósemos fósedes fosen

fun fuche(s) foi fomos fostes foron

fose foses fose fósemos fósedes fosen

houben houbeche(s) houbo houbemos houbestes houberon

houbese houbeses houbese houbésemos houbésedes houbesen

4.2

Sound change as a source of suppletion

Suppletion sometimes results from sound change. When such changes result in forms that share phonological material but cannot be considered to reflect synchronic allophonic patterns, the outcome is weak suppletion. If the changes lead to forms with no phonological material in common, the outcome is strong suppletion. 4.2.1 Weak suppletion Sound changes that induce phonemic split can create weak suppletion, as in some cases of palatalization. Various instances of such suppletion appear in the Romance languages. For example, the Spanish verb decir ‘to say’ shows velar~interdental or velar~alveolar alternations that result from the differential developments of velars before front and non-front vowels (see Table 5). Table 5.  A case of weak suppletion “say”

Latin

1sg 2sg

dīcō dīcis

Spanish [k] /k/ digo [ɣ] /g/ [k] /k/ dices [θ] /θ/* Present Indicative

* In most of the Spanish-speaking world, the phone is [s] and the phoneme is /s/.

© 2013. John Benjamins Publishing Company All rights reserved

180 Matthew L. Juge

4.2.2 Strong suppletion In some cases, sound changes may render related forms completely unconnected phonologically and thus create strong suppletion, as in some reflexes of one of the Proto-Indo-European copula roots (see Table 6). Table 6.  A case of strong suppletion Proto-Indo-European

English

*h1esmi *h1esti

am is

4.3

Analogy

Discussions of suppletion and analogy usually focus on the resistance of existing suppletive paradigms to analogical forces, such as leveling. As Veselinova points out, this resistance is tied to frequency: Since [suppletion] affects high token-frequency items, it is also considered “immune” to analogy because high token-frequency items are learned by rote and are (Veselinova 2006: 23) thus resistant to analogical leveling. 

There are, however, some cases of suppletion that result from analogical developments. I consider here three types of analogy identified by Hock (2003). Since analogy depends on relationships among surface forms, it is perhaps not surprising that all three types of analogy seem to create only weak – not strong – suppletion. 4.3.1 Leveling – Suppletion as residue Ronneberger-Sibold (1990) shows that the resistance of house~houses to leveling of previously allophonic voicing alternations in fricative-final stems renders the singular~plural forms suppletive. As we shall see next, other analogical mechanisms actively create suppletion, while leveling can be seen as a passive mechanism of suppletion creation in that it leaves behind a suppletive residue. 4.3.2 Proportional analogy As mentioned earlier, proportional analogy is traditionally thought of as a regularizing force. Hock (1991: 172–173) provides a typical example with the creation of a new plural for English word. Like its Germanic cognates (e.g. Norwegian ord), this noun had no distinct plural form. Following the relations found in many other nouns, speakers created words analogically: stone : stones :: word : X, X = words. Generally this process extends more general (‘productive’) classes, as with the non-standard English forms brang and brung, which fit a pattern populated by © 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 181

Table 7.  A case of productive proportional analogy Standard English

Non-standard English

bring~brought~brought bring~brang~brung cf. think~thought~thought ring~rang~rung

Table 8.  A case of proportional analogy resulting in suppletion Latin Infinitive 1s Imperfect Indicative

“come” “go” “come” “go”

venīre īre veniam ībam

Galician > > > >

vir ir viña iba~ía

>>

iña

Principal sound changes

Additional sound changes

ī>i i > j / __ V

e > Ø / __# m > Ø / __# w>b

nj > ɲ / __V n > Ø / V__V e > Ø / __i

more lexemes than the pattern of the standard forms (as in Table 7). Maiden has also pointed out the potential for analogy to spread irregularity (1992). In the case of bring, the earlier and the later patterns are both irregulars, but the analogical forms fit a pattern with more members (cf. Aski’s (1995) claims about templates (discussed in Section 5, below)). Other instances, however, show regular verbs turned into irregulars, as in the case of dove as a replacement for the regular past form dived. Note that blocking does not prevent the creation of such forms. Veselinova (2006) provides further discussion of some of the problems with blocking. A less familiar result of analogical processes is the creation of suppletion. In Fisterran Galician, the verb ir “to go” has Imperfect Indicative forms that result from the application of a proportional analogy involving forms of the verb vir “to come” (cf. RAG 2005). As outlined in Table 8 (where a double arrow indicates an analogical rather than a phonological development), a series of phonological changes rendered the verbs vir and ir nearly homophonous in the infinitive, although they were less similar in the Imperfect Indicative. A new, suppletive stem was created in the verb ir via the application of the proportion vir : viña :: ir : X, X = iña, rather than iba or ía. Thus in this case, an irregular verb becomes even more irregular under analogical influence. This development challenges previous views of analogical change. For example, two of the best known treatments of analogical change, Kuryłowicz (1947) and Mańczak (1957–1958), are concerned primarily with the forms within a © 2013. John Benjamins Publishing Company All rights reserved

182 Matthew L. Juge

paradigm and are thus not well-suited for interactions among lexemes. Some subsequent approaches, such as Natural Morphology, have framed analogical change in terms of considerations like transparency and markedness. Veselinova follows Fertig (1998) in concluding that this approach is “unsuitable for fully explaining the occurrence of suppletion” (2006: 28). 4.3.3 Contamination Perhaps somewhat less striking – but nonetheless significant – is the creation of suppletive forms via contamination. Contamination is better known as a means by which items in a list become more similar, as in the case of adjacent numbers in some languages (e.g. Lithuanian devynì “nine”, dešimt “ten” – Balto-Slavic initial /n/ replaced with /d/ under the influence of “ten”; the initial /f/ of English four under the influence of five). As Maschi (2007) and Juge (1999) discuss, certain Galician verbs exhibit suppletive stems affected by contamination. For example, non-standard Present Subjunctive forms of ser “be”, estar “be”, facer “do”, and dar “give” have etymologically unexpected palatal nasals under the influence of ter “have” (and perhaps vir “come”), where the palatal nasal is etymological (Table 9). Table 9.  Suppletion resulting from contamination in non-standard Galician Gloss have be be do

Infinitive ter ser estar facer

1sg/3sg Present Subjunctive Standard

Non-standard

teña sexa estea faga

seña esteña feña

5. Analogy and paradigms Veselinova (2006: 105–115) addresses analogy as a factor in suppletion in Romance and English, based largely on Aski’s (1995) templatic approach. Although Aski’s analysis has been accepted by a number of scholars (e.g. Corbett 2007 and Hippisley et al. 2004), it presents several serious problems. Developments in various Romance languages (e.g. Galician Imperfect Indicative and Present Subjunctive, Surmeiran Present Indicative, Ibero-Romance Preterit and related forms) falsify her claims.

© 2013. John Benjamins Publishing Company All rights reserved



5.1

Analogy as a source of suppletion 183

Templates, analogy, and suppletion

Aski’s approach concerns paradigmatic relationships and the spread of suppletive forms, not their creation. She aims to “demonstrate that language-specific historical processes provide the motivations for the forms of patterns that develop in each language” (1995: 404). She seeks “(i) a motivation for the suppletive replacement of Latin IRE by forms of VADERE ‘to go quickly’ and an a-stem verb [i.e. Italian andare, French aller – author’s note (see below for discussion of the etymology of these forms)] and (ii) a single diachronic process that resulted in the three diverse outcomes” (1995: 404). Before presenting her analysis, Aski rightly criticizes Rudes’s (1980) claim that suppletivizing stems replace all the forms from a given stem. As she points out, Rudes’s statement has two problems: the identification of stems and overlooked data. Establishing stems is not always straightforward, especially since certain verbs have not only irregular stems but also irregular endings. For instance, he considers the singular forms of the Present Indicative of Portuguese ir “to go” to belong to a single stem {va-} (1980: 670), but these forms certainly do not pattern like the corresponding forms of dar ‘to give’ or other -ar verbs (Table 10). The 2sg and 3sg forms vais, vai, in particular, pose a special challenge. The former looks like a 2pl; it could be analyzed as {v-ai-s} or perhaps {vai-s}. Vai does not fit neatly into any pattern. Furthermore, the 1p imos has been replaced, but the 2pl has not, although the earlier 1pl form is clearly from the same stem as ides. Aski also correctly points out that previous analyses attributing the replacement patterns to the monosyllabic nature of certain forms of īre do not account for all the data (1995: 408–409). If ir had a consistent stem in the Present Indicative, it would have the forms vou (allowing still for the irregular ending), vas, va, vamos, vades, vão. Table 10.  Distinct irregular stems in the Portuguese Present Indicative Infinitive 1sg 2sg 3sg 1pl 2pl 3pl

“go”

“give”

“sing”

i-r v-ou vai-s v-ai v-a-mos ide-s v-ão

d-a-r d-ou d-á-s d-á d-a-mos d-a-is d-ão

cant-a-r cant-o cant-a-s cant-a cant-a-mos cant-a-is cant-am*

* The orthography makes the 3pl forms seem more different than they actually are. The final indicates a nasal diphthong (Parkinson 1988: 151). The accents in forms of dar serve only to distinguish verb forms from contracted forms of the preposition de “of ” and the feminine definite article: de + a = da, de + as = das.

© 2013. John Benjamins Publishing Company All rights reserved

184 Matthew L. Juge

Aski does not explain why she seeks a single process to account for the French, Italian, and Spanish data. Examining only the Present tense makes this goal seem more attainable, since the French and Italian forms do not pose the problems raised by the Preterit and related paradigms in Spanish ir (discussed in Section 4.1.3 above and Table 8). To explain why suppletion occurred in these verbs, Aski expands on an idea from Werner’s economy-based approach (1987), namely that “phonological processes reduce elements to a point where the irregular forms, in relation to their token frequency, are no longer optimal” (1995: 409). She proposes that “phonetic erosion destroyed the balance between token frequency and the compressed, irregular elements of the IRE such that even though the forms had high frequency, their reduction was so severe that they could not survive” (1995: 409). She does not specify whether she means here to account only for lexical merger or also for coalescence and incursion, neither of which shows any sign of requiring minimality. The coalescence of Latin faciō and fiō (see Section 4.1.1 above), for example, involves ‘full’ forms, as does the incursion of English ēode and multiple forms of verbs meaning go in Rhaeto-Romance. See Juge (1999) for further arguments against this type of account. Aski claims that the distribution of suppletive stems involves templates that spread by a kind of analogy. She focuses on go verbs in French, Spanish, and Italian, but even within these three languages, the data do not fit neatly with her claims about templatic determination of suppletive patterns. First, the omission of forms other than the present tense allows her to skip over certain difficulties. Even the restricted data from the present tense present serious difficulties. Aski claims that the 1sg Present Indicative and all Present Subjunctive forms of the Spanish verbs ir “to go”, traer “to bring”, and caer “to fall” share a template (subsequently labeled by Maiden 2005 the ‘L-pattern’), as shown by traigo (1sg Present Indicative)~traiga (1sg Present Subjunctive) and caigo (1sg Present Indicative)~caiga (1sg Present Subjunctive). However, as Figure 1 shows, ir does not fit this pattern, since the corresponding forms are voy~vaya, which violates the pattern of Xo~Xa, where X- is the stem shared by the 1sg Present Indicative and the Present Subjunctive. That is, Spanish would need *vayo instead of voy to fit Aski’s scheme. In this regard ir is like dar “give”, estar “be”, haber “have (aux), exist”, saber “know”, and ser “be”. Interestingly, the non-standard Present Subjunctive forms vaiga, vaigas, etc. (ir) and haiga, haigas, etc. (haber) resemble the forms of caer (but only in the Subjunctive). The sociolinguistic stigma of these forms of the Perfect auxiliary illustrates the normative pressure on speakers to retain irregular forms (see Section 5.4). Aski briefly addresses the failure of ir to fit the template and proposes reduction in the form from earlier vādō via *vao, but this proposal does not account for © 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 185

Indicative

Subjunctive

Indicative

Subjunctive

Indicative

Subjunctive

1sg

1sg

caigo

caiga

voy

vaya

2sg

2sg

caes

caigas

vas

vayas

3sg

3sg

cae

caiga

va

vaya

1pl

1pl

caemos

caigamos

vamos

vayamos

2pl

2pl

caéis

caigáis

vais

vayáis

3pl

3pl

caen

caigan

van

vayan

Figure 1.  Putative template for Spanish ir ‘go’ presented by Aski (1995: 424) vs. caer “fall”

the fact that the 1sg Present Indicative form doesn’t share a stem with the Present Subjunctive. She notes the resemblance of voy to the 1sg Present Indicative of dar (doy), estar (estoy), and ser (soy) and rightly points out that they are all hightoken-frequency items. She also claims that they have all undergone stem reduction, but this claim is questionable, since it relies on a connection between vādō and sum, which was already monosyllabic, and unattested disyllabic variants of dō and stō. She does not, moreover, acknowledge the difference in the Infinitive~1s Present Indicative relationship between ir and these other verbs. Aski claims that Italian andare, French aller, and Spanish ir “share a paradigm type or template with other verbs” and that therefore, unlike some copulas, “they do not have the lowest degree of productivity and thus cannot be considered marginal, unnatural elements” (1995: 426). Here she echoes the concerns of Veselinova mentioned above. However, as just discussed, this is not actually true in the case of Spanish (nor in French). Expanding the data set to include other verbs in these languages (as discussed above) or go verbs in other languages reveals further analytical difficulties. Before considering how other Romance data affect Aski’s claims, a brief etymological discussion is in order. The origins of Italian andare, French aller, and some of the Rhaeto-Romance forms discussed below present a number of difficulties. Perhaps the most commonly cited etymon for these forms is ambulāre “to walk”, but this proposal presents phonological problems. Other proposed etyma for one or more of these forms include *ambitāre “to go around”, allātum (perfect passive participle of afferre “to carry to”), and Greek aláomai ‘I wander’ (cf. Alessio 1951–1955: 118, who rejects the proposal that the verb is a derivative of Latin āla “wing”). For the purposes of examining the role of analogy as a source of suppletion, the exact provenance of these forms is not of great importance. For the identification of semantic patterns, however, they may prove quite significant, if definite conclusions can be drawn. That task, though, lies outside the scope of this paper. © 2013. John Benjamins Publishing Company All rights reserved

186 Matthew L. Juge

Table 11.  Suppletive Present tense forms and etyma in Surselvan ir† “to go” Indicative 1sg 2sg 3sg 1pl 2pl 3pl Latin

Subjunctive

mɔn~mondel mɔndi vas‡ mɔndjəs mɔndi va‡ mejn mejən mejs mejəs mɔndiən van‡ meāre/īre†/vādere‡

Returning to the issue of templates, let us consider go verbs in Surselvan and Surmeiran (both Rhaeto-Romance; Table 11). In the Present, Surselvan shows reflexes of vādere “to go” and a second root, probably meāre “to go, pass” (cf. English permeate; not to be confused with meiere [later meiāre] “to urinate”); as Elcock (1959) points out, Decurtins (1958) suggests that they are a blend of vādere and *amnāre [< ambulāre]), with forms of the latter appearing in the 1sg Present Indicative and in the Present Subjunctive – a common pattern, like the one Aski discusses. However, reflexes of meāre also appear in the 1pl/2pl Present Indicative. While it is common for these two slots to differ from the singular and the 3pl, it is unusual for the 1pl/2pl forms to be from the same stem as the 1sg ({m-}) and from a different stem than the 2sg/3sg/3pl forms ({v-}). Hinzelin (2011) also addresses the role of hybrid forms (especially in Francoprovençal), although his restriction of roots to synonyms does not account for the Ibero-Romance incursion of forms of ser “be” into the paradigm of ir “go”. 5.2

Productivity

As mentioned above, Aski claims that changes in verbal alternations are “consistently to more productive templates” (1995: 403). This is incorrect. As Table 12 shows, the first person singular Present Indicative of the Surmeiran verb eir “to go” is the same as the corresponding form of neir “to come”. Historical data make it clear that the form of neir has replaced the previous form of eir (the same replacement has occurred in other Rhaeto-Romance varieties as well; cf. Haiman & Benincà 1992: 235). This results in a Present Indicative with forms from three Latin etyma. While the infinitive eir and the first and second person plural forms are from Latin īre, the infinitive is probably best analyzed as a separate stem synchronically because of the differences among the forms created by sound change.

© 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 187

Table 12.  Overlapping suppletion in the Present Indicative in Surmeiran “go” Infinitive 1sg 2sg 3sg 1pl 2pl 3pl Latin etyma

“come”

eir†

neir vign vign vignst vast‡ vign vo‡ nign giagn† niz gez† vignan von‡ venīre/īre†/vādere‡

Although a thorough discussion of productivity is beyond the scope of this paper (see Bauer 2003 for discussion and references), in this section I address the issue as presented by Aski in the context of suppletion and analogy. 5.3

Split

According to Bybee, “Suppletion requires the splitting of paradigms – went had to separate from wend in order to become the past tense of go” (2007: 171). This claim is incorrect. The first problem with Bybee’s view on this matter concerns her definition of ‘true suppletion’. Like Aski, Bybee views suppletion as dependent on the presence of “paradigms consisting of stems of different etymologies, such as go, went or am, was)” (2007: 171). This limitation is unnecessary and could lead to disregard for important data, especially if only verifiable cases are counted since many languages do not have adequate attestation to determine whether forms are etymologically distinct. If further investigation reveals that suppletion involving distinct etyma is meaningfully different from suppletion caused by other sources, then this distinction will prove worthwhile, but at this point there is no clear evidence in favor of such a view. Secondly and perhaps more significantly, even cases that do involve distinct etyma do not always involve split of any kind, as discussed in Section 4.1. Of the three types of suppletivizing processes that involve distinct etyma, only incursion exhibits split, while in coalescence and lexical merger, the contributing lexemes do not remain separate. 5.4

Retention of suppletive forms

Aski attributes the retention of suppletive forms to three factors: high token frequency, semantic or pragmatic basicness (as proposed by Osthoff 1899 and © 2013. John Benjamins Publishing Company All rights reserved

188 Matthew L. Juge

espoused by Dressler 1985, among others), and “the fact that they adhere to a conjugational pattern that renders them less anomalous” (1995: 427). Only the first of these has any correlation with how suppletive paradigms develop (see Section 8). As Veselinova points out, Osthoff ’s notion of basicness “is rather hard to delimit in any rigorous way” (2006: 26) and therefore cannot be evaluated coherently. Aski’s third point is ambiguous. It is not clear whether she is suggesting that suppletive forms at a given stage are less anomalous than they were before or that they are less anomalous than they might appear initially. In a sense, it does not matter, because neither interpretation fits the facts. The Rhaeto-Romance verbs meaning go were already anomalous before the 1sg Present Indicative of the verb meaning come replaced the forms derived from Latin vādere, but they became more anomalous. Likewise, the analogical change in Galician ir increased its idiosyncratic character. Veselinova addresses an additional claim sometimes made regarding the stability of suppletive paradigms, namely blocking. Blocking is the mechanism whereby a productive process is pre-empted from applying to a given root because an existing form in the lexicon takes the place that the productively created form would occupy. Under this view, English went blocks the analogically created *goed. Veselinova points out several problems with blocking as a factor in the history of suppletive paradigms. Especially important is the fact that “blocking does not explain the emergence of suppletive forms” nor the replacement of one suppletive form by another, as with English went for ēode (2006: 20). Blocking also fails on a more fundamental level in that there are instances of regularization of suppletive verbs, as with goed in Newfoundland English (Clarke 2004: 307). Bybee suggests a social factor in the maintenance of suppletion: “pressure to conform” (2007: 30). This is consistent with Wolfram & Schilling-Estes’s discussion of the stigma against regularization (2006: 51): the only thing separating [analogically-created] verb forms like helped and rang from forms like knowed and brang is social acceptability… From a strictly linguistic standpoint, the designation of regularized forms as “standard” vs. “nonstandard” is completely arbitrary. From a sociological perspective, of course, it is no accident that the forms associated with socially favored groups become established as standard forms while those associated with low-status groups remain nonstandard.

Like many social phenomena, then, suppletion is not amenable to specific predictions, such as when a particular form might be regularized, though certain trends (based, e.g., on patterns of token frequency) may be identifiable. Since “all change involves variability” (Weinreich, Labov & Herzog 1968: 188), we find that,

© 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 189

despite such social pressure, alternative forms may compete for extended periods (cf. common but condemned forms like Spanish haiga and English brang).

6. The semantics of suppletive lexemes and roots Different types of suppletion exhibit different kinds of semantic relationships between roots (Juge 1999). In-depth semantic analysis of semantic relations among suppletive roots (e.g. verbs of motion [go, come, walk, lead, sink], existence [be, exist], posture [sit, stand], etc.) is still in its infancy (Traugott & Dasher’s (2005) treatment of semantic change, for instance, does not even mention suppletion). Analogy-driven suppletion is a key to a complete understanding of how semantics and suppletion interact. It also connects to issues in grammaticalization discussed in Juge (2007). For the purposes of the present discussion, I will focus on the distinction between sound change and other sources of suppletion. Sound change can, in principle, cause suppletion in any lexeme, irrespective of its meaning. There are, however, important relationships between frequency and analogical regularization on the one hand and between frequency and certain meanings on the other (see Section 7 for further discussion). Suppletion caused by processes other than sound change – incursion, coalescence, lexical merger, and various types of analogy – is directly tied to the semantic connections among the lexemes involved (with the possible exception of leveling, though this too is probably tied to frequency and to semantic factors). While the semantic basis for incursion, coalescence, and lexical merger seems intuitively clear, the details of the relationships among lexemes involved in suppletion are still under investigation. Juge (1999) proposes that two major types of suppletion – overlapping and non-overlapping (discussed above, Section 4.1.3) – differ primarily according to the semantic distance between the roots involved. He acknowledges some complications in measuring semantic distance, but the basic idea is supported not only by the distribution of overlapping and non-overlapping suppletion but also by the relationships among lexemes involved in proportional analogy and contamination. Semantic distance is a gradable notion. The more closely two lexemes approach synonymy, the more likely they are to enter into non-overlapping suppletion. For example, the Latin motion verbs īre and vādere were near-synonyms, and their Romance reflexes belong to non-overlapping paradigms, as in Spanish ir (voy 1sg pres ind) and French aller (vais 1sg pres ind, irai 1sg fut ind). The motion verbs īre and venīre ‘come’, however, are distinguished deictically and do not appear in the same paradigm, except on a limited basis, as in such Rhaeto­Romance varieties as Surmeiran and Puter, where the first person singular © 2013. John Benjamins Publishing Company All rights reserved

190 Matthew L. Juge

Table 13.  Correlations among sources of suppletion and other variables Semantics Phonology Strong~weak ± Overlapping incursion/coalescence/lexical merger sound change leveling contamination analogy



✓ ✓

✓ ✓ ✓ ✓

S/W S/W W* W* W*

+/– – – –** –**

* It is unlikely that strong suppletion results from any type of analogy since analogy involves relations among the forms that strong suppletion lacks by definition. ** Overlapping suppletion is unlikely to result from contamination or proportional analogy.

Present Indicative reflex of venīre appears in the paradigm of the reflex of īre, i.e. the two verbs exhibit overlapping suppletion (Table 11). See Juge (1999) for further details. Recall that go and come are the same two meanings involved in the proportional analogy discussed in Section 4.3.2. Similar semantic relations obtain among the Galician verbs showing contamination. The two copulas, ser and estar, are of particular interest in that these verbs’ meanings illustrate what Juge (2010) calls “intermediate zones and variable outcomes” in semantic continua. Romance reflexes of esse and stāre show several distinct patterns: separate lexemes (IberoRomance ser and estar), non-overlapping suppletion (French être), overlapping suppletion (Italian essere and stare with shared participle stato), and optional overlapping suppletion (Catalan ésser and estar, with participles estat (shared) and sigut (ésser only)). Table 13 summarizes the major types of suppletion, their sources, and whether semantics and phonology play significant roles in their development. Maiden (2011c) identifies similar ‘fuzziness’ in the determination of whether alternations are conditioned phonologically or morphologically. Fertig also invokes the concept of semantic distance. He claims (1998: 1077) that German lernen has more semantic content than sein and that therefore, the suppletion in the latter and lack of suppletion in the former fit with Bybee’s claim that “[t]he more closely related two forms are semantically, the more likely they are to be similar morphophonemically” (1988: 130). As mentioned above, there is as yet no established way to measure semantic distance rigorously. While Fertig’s claim may capture a general trend, it needs at least some refinement. While some meanings like be and go are often found in auxiliary constructions and may be associated with so-called semantic bleaching, there are other meanings involved in suppletion, like eat, for which making such a case seems much harder, as Fertig himself points out. Furthermore, a supposed lack of semantic coherence among the forms of verbs meaning go or be fails to account for how suppletive forms are distributed within such verbs. © 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 191

Maiden presents a set of Romance data that, in his view, resist explanation by appeal to the factors that Fertig says “show suppletion as an effect of competition between synonyms or near-synonyms, resolved by the differential integration of the competing lexemes into a single paradigm” (2011b: 711, fn. 61). As I have just argued, however, the semantics of suppletion goes far beyond matters of (near-) synonymy. Interactions like those between go and be and between go and come illustrate that semantic relationships other than synonymy affect not only what roots contribute to a given paradigm but also which parts of those paradigms will be involved.

7. Motivations and mechanisms In Section 5.1 I argued against minimality as a motivation for suppletion. It is far easier to find counterexamples to putative motivating factors than to discover empirical support for them. In this section I leave aside the broader question of why suppletion occurs and address motivations for analogical operations in suppletion. A full account of the role of analogy in the development of suppletive paradigms requires an understanding of two key elements: motivations for change and mechanisms of change. 7.1

Motivations for analogical change

Maschi analyzes analogical changes in Galician in terms of morphomes (see Aronoff 1994 for the term and Maiden 1999 for additional cases of morphomes in diachrony). She frames these developments teleologically: The morphomic structures that undergo analogical processes … are reinforced so as to constitute, despite their markedness, subregularities within the verbal system, and so as to provide some strategies in language learning and production.  (2007: 137, original emphasis)

Aski also views suppletion teleologically. She asserts, “In order to restore the balance between form and frequency, a change toward greater regularity by analogy to other preexisting language patterns or models is expected” (1995: 409). Furthermore, she challenges Lass’s “model of exaptation, [where] templates arise for no purpose in that … they are the product of random historical change” and claims, “The moment other elements realign themselves according to these patterns, the templates are exapted to organize verbal information according to a coherent system” (1995: 426–427). © 2013. John Benjamins Publishing Company All rights reserved

192 Matthew L. Juge

While a thorough discussion of teleology in language change is beyond the scope of this paper, it is worthwhile to address briefly a few points. First, the notion that changes might occur without direction, i.e. non-teleologically, is not the same as claiming that the changes are random. The sound changes that produced weak suppletion in the Spanish verb decir “to say” are neither phonologically nor morphologically directed, but they are in no way random. Instead, they are governed by considerations like aerodynamic and perceptual factors that themselves are not directed either (cf. Ohala, who argues that “sound change, at least at its very initiation, is not teleological. It does not serve any purpose at all” (2005: 683)). Joseph has emphasized the importance of distinguishing between “what speakers do versus what linguists do” (1992: 131) and of recognizing that the generalizations that speakers make may not reach the same level of generality of those that linguists make (1992, 1997). Speakers’ behavior often suggests that they are not concerned with (and indeed may not really be aware of) the structural patterns that interest linguists. The regularized outcomes typically associated with analogical processes may be epiphenomena rather than goals. If this is so, then the fact that analogy can create suppletion presents no paradox, for speakers aren’t trying to change the system; they are changing pieces of the system without concern for the overall effect. This view casts in a rather different light the mechanism behind patterns of change. On the one hand, some linguists see this as a kind of teleological process; Maiden claims, “The very fact that speakers make ‘morphomic’ generalizations […] suggests that they seek out, and seek to reinforce, paradigmatic patterns of maximal generality” (2011b: 267). The data presented here support a local, non-teleological perspective. As Joseph puts it, speakers “often act as if they were in a fog, by which is meant not that they are befuddled but that they see clearly only immediately around them, so to speak, and only in a clouded manner farther afield” (1992: 140). 7.2

Mechanisms employed in analogical change

This localized perspective on analogical change fits several key aspects of the picture. It corresponds well with the tendency for analogy to affect members of an inflectional paradigm but not necessarily derived elements (as in Latin honor and honestās) and with the semantic relations discussed in the preceding section. The paradigmatic patterns mentioned earlier are consistent with Bybee’s arguments regarding relatedness among forms, including her notions of relevance and lexical strength (1985). These patterns are also consistent with many of the points raised by scholars investigating autonomous aspects of morphology, especially as seen in the work of Maiden (2005, 2011a, 2011b, 2011c). © 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 193

8. Frequency and memory Analogy and suppletion are both tied to frequency, but very little research, if any, links them directly to each other (cf. volumes on frequency by Bybee & Hopper 2001 and Bybee 2007). The fact that analogy and suppletion are linked to each other raises new questions about how each relates to frequency and, naturally, how frequency relates to cases involving both analogy and suppletion. Further, frequency effects raise questions about memory that must be addressed in light of these data. As mentioned in Section 6, semantics and frequency are also linked. Specifically, Veselinova’s cross-linguistic analysis shows that verbs with certain meanings – such as be, come, go, say, and do – tend to be more frequent than others (2006: 104). To the extent that semantic and pragmatic factors drive frequency and that frequency affects the likelihood of analogically-based suppletion on the one hand and analogical regularization on the other, it is likely that many of the hallmarks surrounding suppletion derive – directly or indirectly – from semantic and pragmatic factors. According to Fertig, “Semantic generality will invariably and necessarily correlate strongly with token frequency” (1998: 1079). The view of certain meanings as ‘general’ is both quite common and vastly oversimplified. Consider Bybee et al., who claim, “The more generalized movement verbs go and come, however, lack specifics concerning the nature of the movement and are thus appropriate in a much wider range of contexts. […] It is lexical items of this degree of generality that […] enter into grammaticization” (1994: 5). Juge argues that generality per se is not the key factor in such developments, but rather the high degree of polysemy that such verbs exhibit (2002: 180). As we saw above with analogy as a regularizing and a suppletivizing force, frequency and suppletion present an apparent paradox: How is it that frequency can be tied to both the creation of suppletion and resistance to regularization? The correlations among type frequency, token frequency, suppletion, and regularization are well-established. Items with high token frequency are resistant to analogical regularization and are particularly susceptible to irregular sound change (cf. English, where high-frequency said has /ɛ/ in contrast to the full diphthong /ej/ in paid). It is telling that the cases of suppletion created analogically in Galician have occurred in the relatively lower-frequency categories of Present Subjunctive and Imperfect Indicative and not in higher-frequency categories like the Present Indicative or the Preterit. Slips of the tongue suggest that one factor in analogical regularization is access to stored forms. Bybee provides evidence for the view that forms with high token frequency – regular or irregular – have greater lexical strength. Thus, speakers are much more likely to momentarily forget an irregular form and substitute © 2013. John Benjamins Publishing Company All rights reserved

194 Matthew L. Juge

a regularized form if the form is of relatively low frequency. Bybee’s citation of variation in past tense forms in English supports this view. She reports variation for creep (creeped, crept), leap (leaped, leapt), and weep (weeped, wept) in contrast with the lack of variation for the much more frequent verbs keep, leave, and sleep (2007: 29).

9. Conclusions Analogy – in its various guises – has been largely ignored as a factor in the development of suppletion (it has received more attention with respect to the elimination of suppletion). Where it has received attention, one main focus has been on paradigmatic distribution, but the proposals have not adequately accounted for the data. As a source of suppletive forms, analogy has apparently escaped notice almost completely. Not only is analogy important in cataloging the ways in which suppletion develops but it also promises to clarify issues like the interactions among the various types of suppletion, including the understudied phenomenon of overlapping suppletion (in which distinct lexemes share suppletive forms), with other areas, including semantics, psycholinguistics, and neurolinguistics. Further examination of these relationships – using data from both familiar languages and minority languages – will advance our understanding of diachronic processes and synchronic patterns and improve our ability to confidently apply our knowledge to lesser-known languages and families.

Abbreviations 1 2 3 aux ind pres pret pl sg

first person second person third person auxiliary Indicative Present Preterit plural singular

© 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 195

References Alessio, Giovanni. 1951–1955. Grammatica storica francese. Vol. 2. Bari, Italy: Casa editirice “Leonardi da Vinci”. Aronoff, Mark. 1994. Morphology by Itself: Stems and Inflectional Classes. Cambridge, MA: MIT Press. Aski, Janice M. 1995. “Verbal Suppletion: An Analysis of Italian, French and Spanish to go”. Linguistics 33.403–432. Bauer, Laurie. 2003. Introducing Linguistic Morphology. 2nd ed. Washington, DC: Georgetown University Press. Bybee, Joan. 1985. Morphology: A Study of the Relation Between Meaning and Form. (= Typological Studies in Language, 9.) Amsterdam/Philadelphia: John Benjamins. Bybee, Joan. 2007. Frequency of Use and the Organization of Language. Oxford: Oxford University Press. Bybee, Joan & Paul Hopper. 2001. Frequency and the Emergence of Linguistic Structure. Amsterdam & Philadelphia: John Benjamins. Bybee, Joan, Revere Perkins, & William Pagliuca. 1994. The Evolution Of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: University of Chicago Press. Clarke, Sandra. 2004. “Newfoundland English: Morphology and Syntax”. A Handbook of Varieties of English: A Multimedia Tool ed. by Bernd Kortmann, Kate Burridge, Rajend Mesthrie, Edgar W. Schneider & Clive Upton, Vol. 2, 313–318. Berlin: Mouton de Gruyter. Corbett, Greville G. 2007. “Canonical Typology, Suppletion, and Possible Words”. Language 83.8–42. Decurtins, Alexi. 1958. Zur Morphologie der unregelmässigen Verben im Bündnerromanischen: Historisch-deskriptive Studie mit besonderer Berücksichtigung der Sur und Sutselvischen. Bern: A. Francke. Dressler, Wolfgang U. 1985. “On the Predictiveness of Natural Morphology”. Journal of Linguistics 21:2.321–337. Elcock, W. D. 1959. “Zur Morphologie der unregelmäßigen Verben im Bündnerromanischen by Alexi Decurtins”. The Modern Language Review 54:4.612–613. Fertig, David. 1998. “Suppletion, Natural Morphology, and Diagrammaticity”. Linguistics 36:6.1065–1091. Haiman, John & Paola Benincà. 1992. The Rhaeto-Romance Languages. London: Routledge. Hinzelin, Marc-Olivier. 2011. “Syncretism and Suppletion in Gallo-Romance Verb Paradigms”. Morphological Autonomy: Perspectives from Romance Inflectional Morphology ed. by Martin Maiden, John Charles Smith, Maria Goldbach & Marc-Olivier Hinzelin, 287–310. Oxford: Oxford University Press. Hippisley, Andrew, Marina Chumakina, Greville Corbett & Dunstan Brown. 2004. “Suppletion: Frequency, Categories and Distribution of Stems”. Studies in Language 28:2.387–418. Hock, Hans Heinrich. 1991. Principles of Historical Linguistics. 2nd ed. Berlin: Mouton de Gruyter. Hock, Hans Henrich. 2003. “Analogical Change”. Handbook of Historical Linguistics ed. by Richard Janda & Brian D. Joseph, 441–460. Malden, MA: Blackwell. Joseph, Brian D. 1992. “Diachronic Explanation: Putting Speakers Back into the Picture”. Explanation in Historical Linguistics ed. by Garry W. Davis & Gregory K. Iverson, 123–144. Amsterdam/Philadelphia: John Benjamins.

© 2013. John Benjamins Publishing Company All rights reserved

196 Matthew L. Juge

Joseph, Brian D. 1997. “How General are Our Generalizations? What Speakers Actually Know and What They Actually Do”. ESCOL ’96. Proceedings of the Thirteenth Eastern States Conference on Linguistics ed. by Anthony D. Green & V. Motapanyane et al., 148–160. Ithaca: Cascadilla Press. Juge, Matthew L. 1999. “On the Rise of Suppletion in Verbal Paradigms”. Proceedings of the Twenty-Fifth Annual Meeting of the Berkeley Linguistics Society ed. by Steve S. Chang, Lily Liaw & Josef Ruppenhofer, 183–194. Berkeley: BLS. Juge, Matthew L. 2002. Tense and Aspect in Periphrastic Pasts: Evidence from Iberian Romance. Ph.D. dissertation, University of California, Berkeley. Juge, Matthew L. 2007. “Metaphor and Teleology Do Not Drive Grammaticalization”. Historical Linguistics 2005: Selected Papers from the 17th International Conference on Historical Linguistics, Madison, 31 July – 5 August 2005 ed. by Joseph C. Salmons & Shannon DubenionSmith, 33–48. Amsterdam & Philadelphia: John Benjamins. Juge, Matthew L. 2010. “Morphological Mismatch and Temporal Reference in Interaction with Lexical Semantics in Spanish”. Romance Philology 64.209–221. Kuryłowicz, Jerzy. 1947. “La nature des procès dits analogiques.” Acta Linguistica 5.121–138. Reprinted in Readings in Linguistics ed. by Eric P. Hamp, Fred W. Householder & Robert Austerlitz, 2.158–174. Chicago: University of Chicago Press, 1966. Maiden, Martin. 1992. “Irregularity as a Determinant of Morphological Change”. Journal of Linguistics 28.85–312. Maiden, Martin. 1999. “‘Perfecto y tiempos afines’. History of an Ibero-Romance Morphome”. Oxford Working Papers in Linguistics, Philology and Phonetics 4 ed. by Raffaella Folli & Roberta Middleton, 70–83. Oxford: Faculty of Linguistics, Philology and Phonetics, University of Oxford. Maiden, Martin. 2005. “Morphological Autonomy and Diachrony”. Yearbook of Morphology 2004 ed. by Geert Booij and Jaap van Marle, 137–175. Dordrecht: Springer Verlag. Maiden, Martin. 2011a. “Morphological Persistence”. The Cambridge History of the Romance Languages: Volume 1: Structures ed. by Martin Maiden, John Charles Smith & Adam Ledgeway, 155–215. Cambridge: Cambridge University Press. Maiden, Martin. 2011b. “Morphophonological Innovation”. The Cambridge History of the Romance Languages: Volume 1: Structures ed. by Martin Maiden, John Charles Smith & Adam Ledgeway, 216–267. Cambridge: Cambridge University Press. Maiden, Martin. 2011c. “Morphomes and ‘Stress-conditioned Allomorphy’ in Romansh”. Morphological Autonomy: Perspectives from Romance Inflectional Morphology ed. by Martin Maiden, John Charles Smith, Maria Goldbach & Marc-Olivier Hinzelin, 36–50. Oxford: Oxford University Press. Mańczak, Witold. 1957–1958. “Tendences générales des changements analogiques.” Lingua 7.298–325, 387–420. Maschi, Roberta. 2007. “Analogy and Irregularity in Romance Verbal Morphology”. Proceedings of the Fifth Mediterranean Morphology Meeting ed. by Geert Booij, Bernard Fradin, Angela Ralli & Sergio Scalise, 125–140. http://mmm.lingue.unibo.it/proc-mmm5.php Morpurgo Davies, Anna. 1998. History of Linguistics, Volume IV: Nineteenth-Century Linguistics. London: Longman. Ohala, John J. 2005. “Phonetics and Historical Phonology”. The Handbook of Historical Linguistics ed. by Brian D. Joseph & Richard D. Janda, 669–686. Malden, MA: Blackwell. Osthoff, Hermann. 1899. Vom Suppletivwesen der indogermanischen Sprachen. Heidelberg: J. Hörning.

© 2013. John Benjamins Publishing Company All rights reserved



Analogy as a source of suppletion 197

Parkinson, Stephen. 1988. “Portuguese”. The Romance Languages ed. by Martin Harris & Nigel Vincent, 131–169. Oxford: Oxford University Press. Real Academia Galega (RAG). 2005. Normas ortográficas e morfolóxicas do idioma galego. Vigo: Instituto da Lingua Galega. Ronneberger-Sibold, Elke. 1990. “The Genesis of Suppletion Through Morphological Change”. Proceedings of the Fourteenth International Congress of Linguists: Berlin/GDR, August 10–15, 1987, Vol. I ed. by Werner Bahner, Joachim Schildt & Dieter Viehweger, 628–631. Berlin: Akademie-Verlag. Rudes, Blair A. 1980. “On the Nature of Verbal Suppletion”. Linguistics 18.655–676. Traugott, Elizabeth C. & Richard B. Dasher. 2005. Regularity in Semantic Change. Cambridge: Cambridge University Press. Veselinova, Ljuba. 2006. Suppletion in Verb Paradigms: Bits and Pieces of the Puzzle. Amsterdam & Philadelphia: John Benjamins. Weinreich, Uriel, William Labov & Marvin I. Herzog. 1968. “Empirical Foundations for a Theory of Language Change”. Directions for Historical Linguistics ed. by W. P. Lehman & Yakov Malkiel, 95–188. Austin: University of Texas Press. Werner, Otmar. 1987. “The Aim of Morphological Change is a Good Mixture – Not a Uniform Language Type”. Papers from the 7th International Conference on Historical Linguistics ed. by Anna Giacalone Ramat, Onofrio Carruba & Giuliano Bernini, 591–606. Amsterdam & Philadelphia: John Benjamins. Wolfram, Walt & Natalie Schilling-Estes. 2006. American English: Dialects and Variation. 2nd ed. Malden, MA: Blackwell.

© 2013. John Benjamins Publishing Company All rights reserved

© 2013. John Benjamins Publishing Company All rights reserved