This is something I’ve been meaning to post for a while now, dealing partially with online data collection, partially with Judeo-Spanish phonology. Eric’s recent post mentions a talk by Bert Vaux on the pros and cons of using Google for linguistic research (see May 6, 2005, of the colloquium schedule here, or alternatively, here), which got me thinking about the use of linguistic data from online sources. On issues related to Google searching, see this article from The Economist, as well as several posts on the Language Log, e.g., by Mark Liberman, Geoffrey Pullum, and Philip Resnik. Having done some reading on phonological variation in Judeo-Spanish, I began to wonder whether the Internet might offer some authentic examples of a particular series of phonological innovations involving /we/ diphthongs.
Here’s a description of these changes, based on Penny (1992, 2000:179) and Sala (1965:546-7):
- (1) Word-initial /nwe/ is regularly modified to /mwe/, such that standard Spanish nuevo “new”, nueve “nine”, nuestro “our” correspond to Judeo-Spanish muevo, mueve, muestro.
- (2) The diphthong /we/ is reinforced to [ɣwe] in syllable-initial position, as in non-standard Peninsular and American Spanish (e.g., hueso /weso/ [ɣweso] “bone”).
- (3) /we/ is also reinforced when preceded by a consonant in the same syllable, e.g., luego “later” > *[lɣwe-ɣo], which leads to resyllabification and the addition of a vowel before the lateral: [el-ɣwe-ɣo]. When the preceding consonant is voiceless /s/, the reinforcing element is realized as a voiceless labiovelar fricative, e.g., sueño “sleep, dream” > *[sʍwe-ño] > [es-ʍwe-ño] , sometimes written as esfueño. N.B: Sala(1965:547) claims that the epenthetic fricative is [χ], which has undergone a hypercorrective shift to [f] in some Judeo-Spanish dialects. Penny (1992:137) argues that labiovelar [ʍ] is variably perceived as either labial(dental) /f/ or velar /x/.
In Kilgarriff and Grefenstette’s (2003:339) Altavista-based estimates of web size for 32 different languages, Judeo-Spanish doesn’t even appear, at least not as a category separate from “Spanish”. As it turns out, however, the best place to find authentic examples from actual usage was in the searchable archives of an email correspondence circle at Yahoo! Groups. Ladinokomunita was founded by moderator Rachel Amado Bortnick of Brookhaven Community College and has been in operation since January 5, 2000. See the group’s web site, and a multilingual dictionary, under development by Güler Orgun, Antonio Ruiz Tinoco, and other collaborators. The group now has over 13,100 posts as of August 23, 2005. Obviously, the size of this corpus does not come close even to the 10,332,000 pages in Albanian purported to exist on the web in 2003 (Kilgarriff and Grefenstette 2003:339), but the disparity seems unavoidable, given the precarious status of Judeo-Spanish as an endangered language.
Although the “Search Messages” feature in Yahoo! Groups seems pretty limited, e.g., no wildcard search capability, I was able to find some interesting examples of each of the innovations in (1) through (3), based on almost six years worth of posts made by the members of Ladinokomunita:
Group 1: The /nwe/ > /mwe/ shift
| Standard Spanish | Judeo-Spanish (per Ladinokomunita) | nuera "daughter-in-law" | nuera ~ elmuera ~ ermuera |
| nuez "walnut" | muez, muezes (pl.) |
| nueve "nine" | mueve dizimueve (< diecinueve "nineteen") |
| <inuevo "new" | muevo de muevo ~ demuevo (< de nuevo "again") muevamente (< nuevamente "again") |
| nuestro "our" | muestro |
Group 2: /we/-strengthening
| Standard Spanish | Judeo-Spanish (per Ladinokomunita) |
| huerta "garden" (h is silent) | huerta ~ guerta guertisika (dimin.) guertelano ~ guertalano "orchard" |
| huérfano "orphan" | guerfano guerfanikos (dimin.) |
| huevo "egg" | huevo ~ guevo ~ guebo |
| hueso "bone" | gueso |
| huelo "I smell" | guelo los golores de las roskas "I smell the smells of the cakes/bread rolls" (< olores "smells") |
| abuelo, abuela "grandfather, grandmother" | abuelo ~ aguelo, abuela ~ aguela |
| bueno "good" | bueno kon karinio y muncho mazal gueno "with affection and a lot of good luck" |
Group 3: /we/-strengthening after tautosyllabic non-nasal C
| Standard Spanish | Judeo-Spanish (per Ladinokomunita) |
| tuerto "one-eyed" | tuerto ~ tuguerto si no te va a pasar un diya todo tuguertado "or else you’ll have a bad day (?)" (cf. a tuertas "in a way that is opposite of what should be done" |
| jueves "Thursday" | djueves ~ jueves ~ djugueves ~ jugueves (dj, j = [ʤ]) |
| duele "hurts" | duele ~ erguele ~ deguele Su korason le erguelia "His heart was hurting him" |
| rueda "wheel" | rueda ~ rugueda |
| trueno "thunder" | trueno ~ turguelo ~ turgelo |
| suegro, consuegro "father-in-law" suegra, consuegra "mother-in-law" |
suegro ~ shuegro ~ esuegro ~ eshuegro ~ esguegro ~ kosuegro ~ koshuegro ~ kosfuegro (h = [h]) suegra ~ esuegra ~ eshuegra ~ esguegra ~ esguefra ~ kosuegra ~ koshuegra agora todos estan komo una famiya, i koshuegrando uno kon otro "now all are like a family, and becoming parents-in-law with one another" |
| sueño "sleep, dream" | suenyo ~ suenio ~ esuenyo ~ shuenyo ~ eshuenyo ~ sfuenyo ~ esfuenyo ~ esfuenio ~ esfueniyo Ma se ke yo esto eshuenyando un eshuenyo imposivle agora "But I know that I am dreaming an impossible dream now" |
The first thing to note is the range of orthographic variation in the Judeo-Spanish examples. Many of the innovative forms compete with the corresponding standard form, especially in the third group, e.g., tuerto ~ tuguerto “one-eyed”. Such competition seems intermediate in the second group but is virtually absent in the first group involving nasals: only nuera “daughter-in-law” is attested in addition to the innovative elmuera and ermuera. Presumably, these variants avoid association with muera, present subjunctive of morir ~ murir “to die” (Penny 1992:138).
The examples also suggest some intriguing phonological patterns. First, all of the standard forms in groups one and three have /we/ appearing after a coronal consonant. The target of innovation seems to be a heterorganic sequence of a coronal consonant followed by a labiovelar glide. However, group two does have two innovative forms in which /bw/, with a shared labial component, shifts to /gw/, with a shared velar component.
Second, the innovations can be viewed as different strategies for avoiding the marked coronal+/we/ configuration. This is achieved via place assimilation in group one, whereby /n/ changes to agree in [labial] with the following labiovelar glide. Assimilation to the velar component is plausibly ruled out by a phonotactic restriction against velar nasals in onset. In contrast, /s/ in /swe/ does not undergo total place assimilation like /n/ does in /nwe/. Sibilant place is maintained most likely due to the greater salience of internal place cues of sibilants in comparison with nasals. Place agreement is achieved via the insertion of [f] or [x] ([h]?) which share a labial and velar component, respectively, with the following glide. In the remaining cases, vowel epenthesis serves to break up the coronal+/w/ sequence. I wonder whether this last strategy might be another case of vowel epenthesis as a backup to assimilation, as Eric has argued elsewhere.
Finally, one might ask whether these innovations reflect changes in the status of prevocalic glides in Judeo-Spanish, at least for the labiovelar glide. In the literature on Spanish, prevocalic glides are argued to form complex nuclei with following vowels (see the discussion in Harris and Kaisse 1999:126-129). As summarized by these authors,
“Spanish prevocalic glides form onsets when no less sonorous segment is available to fill that position but are assigned to rhymes (complex nuclei) when a consonantal onset is available. These arguments hold mutatis mutandis in all the major dialects of Spanish, and no viable counterarguments are known” (p. 129).
“No motivation for a process that moves a glide into an already filled onset is known” (p. 128).
The reinforcement of initial /we/ no doubt satisfies the need for a syllable onset in group two examples like guerta “garden” and guerfano “orphan”. Some additional process may be at work in the examples of groups one and two, in which /w/ is moved into an already filled onset. In fact, co-occurrence restrictions involving place of articulation in consonant+glide sequences are often used as a diagnostic of the onset status of the prevocalic glide (although see Yip 2003). The patterns of epenthesis and place assimilation observed in /Cwe/ sequences could be epiphenomena of a deeper structural innovation in Judeo-Spanish, namely the movement of /w/ out of the nucleus. Does the palatal glide /j/ show similar behavior? Why or why not? It would be interesting to see whether examples from Ladinokomunita might give some insight into these questions.
References
Baković, Eric. To appear. Local Assimilation and Constraint Interaction. The Cambridge Handbook of Phonology ed. by P. de Lacy. Cambridge: Cambridge University Press.
Kaisse, Ellen, and James Harris. 1999. Palatal Vowels, Glides and Obstruents in Argentinian Spanish. Phonology 16:117-190.
Kilgarriff, Adam, and Gregory Grefenstette. 2003. Introduction to the Special Issue on Web as Corpus. Computational Linguistics 29(3):333-347.
Penny, Ralph. 1992. Dialect Contact and Social Networks in Judeo-Spanish. Romance Philology 46(2):125-140.
———-. 2000. Variation and Change in Spanish. Cambridge: Cambridge University Press.
Sala, Marius. 1965. La organización de una ‘norma’ española en el judeo-español. Actas del II Congreso de la Asociación Internacional de Hispanistas ed. by Jaime Sánchez Romeralo and Norbert Polussen, 543-550. Nijmegen: The Spanish Institute of The University of Nijmegen.
Yip, Moira. 2003. Casting Doubt on the Onset/Rime Distinction. Lingua 113(8):779-816.