Collaborations:As far as different subfields, we’re all,

“Ain’t no reason we shouldn’t work together”

John RickfordThomas Wasow

Stanford University

Unlikely Collaborators

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.

• Advisor

• Focus

• Methods

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTime™ and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Sociolinguistic interviews & observations

Introspection

“E-language”

Shared, social phenomenon

“I-language”

Feature of individual psychology

But Friends & Neighbors Talk

John’s house

Tom’s house

Common Interest: Alternations

• Generative grammarians look at what forms may alternate with others

• Variationists look at factors conditioning alternations

• Both concerned with the question of why forms alternate

• But focus on different aspects of the answer

However…

• Focus on different aspects of a question can be complementary, not competitive

• Language is a product of both individual psychology and social interaction, so both perspectives contribute

• Collaborations across subdisciplines are both possible and enlightening (cf “Sociolinguistics & Phonology” plenary by Beth Hume & Naomi Nagy & other papers, LSA Chicago, Jan ‘08)

Overview

• Summarize the four projects we have collaborated on so far:– Topic-restricting as far as– Negative inversion in African-American English– Intensive and Quotative all– Relativizer omission across dialects

• Draw lessons from each and from them collectively

The as far as Project

• “Syntactic Variation and Change in Progress” Language 1995 (71.1:102-31), by Rickford, Wasow, Mendoza-Denton, & Espinoza (1 grad, 1 undergrad student)

• For years, as part of interest in relatively understudied syntactic variation, John had been collecting examples like:As far as filling out the details Ø, that isn’t a problemPeople think I’m constantly in motion, as far as making films Ø– When is the sentence coda (goes or is concerned) omitted?– Is omission of coda increasing?

• Tom joined the project after John asked him about the syntax of the NP after as far as

What we did

• Collected examples – First through listening and reading– Then through corpus searches

• Read usage manuals on this alternation• Did questionnaire study, getting intuitive well-

formedness ratings (4-point scale)• Coded for factors we thought might matter• Ran VARBRUL analyses

Our Findings

• Studying syntactic variation requires methods other than sociolinguistic interviews & recordings (egs overheard on fly, corpora, elicited intuitions)

• Various factors favor coda omission: spoken modality, younger speaker, female speaker, syntactically complex NP, prosodically branching NP, sentence-initial position of as far as

• Coda omission has been increasing, both in frequency and in range of environments where it occurs (in keeping with Bailey’s wave model)

Factors in afa verb absence

Implications for models of language change

Historical evidence for wavelike spread

• Env a: Sentential NPs, e.g.(61) And I will own to you, (I am sure it will be safe), that so far as our living with Mr. Churchill at Enscombe Ø, it is settled. [1816, Jane Austen, Emma, p. 460]

• Env b: Prepositional or Conjoined NPs, e.g.(66) The cabin ... was in perfect condition so far as frame and covering Ø until 1868. [1939, Henry Seidel Canby, Thoreau]

• Env c: Simple NPs, e.g. (3) As far as the white servants Ø, it isn't clear. [Renee Blake, 22, 1987]

But We Learned Much More

• A major turning point in Tom’s career– Got him started thinking about syntactic

complexity, a topic central to the next decade of his work

– Convinced him of the importance of looking at usage

• Also valuable for John: – first paper whose data did not come from creoles

or AAVE – Contributions to models of lg change

Syntactic Complexity

• Coda omission sounds better when NP is longer & more complex:As far as Gore Ø, I make no predictionsAs far as the former Vice President who won the Nobel

Prize Ø, I make no predictions

• VARBRUL indicates complexity (depth of embedding) matters more than length

• Later found some evidence for same conclusion regarding heavy NP shift, the dative alternation, and the verb-particle construction

Judgment Data

• Questionnaire: 20 sentences, rated by 180 speakers on a 4-level scale of acceptability

• Results showed clear patterns, but none were categorical: of 80 possible responses (20 sentences x 4 ratings), none occurred fewer than 10 times.

• Bottom line: well-formedness judgments from a single speaker are highly suspect

• Aside: People who said they would never use coda-less afa sentences were heard doing so

Corpus Data

• Reviewer questioned the high rate of coda omission we reported, suggesting sentences w/o coda might be more noticeable

• Searched a small portion of the Switchboard corpus and found a much higher level of coda omission there

• Convinced us of the value of electronic corpora

• Corpus studies continue to play a big role in our research

AAVE Negative Inversion

• “An Optimality Theoretic Approach to Variation in Negative Inversion in AAVE” Natural Language and Linguistic Theory 1996 (14.3:591-627). Sells, Rickford, & Wasow

• Dealt with examples like:Can’t nobody beat ‘emAin’t nothin’ went down

• Goals: – Improve on old analysis of Labov’s, which posited two distinct

analyses– Explain why this inversion is limited to negatives

• Method: Elicited judgments from native AAVE speakers

Origins of This Collaboration

• Tom got interested in AAVE by student in intro syntax course (Renée Blake), who noted divergences in judgments

• Formed reading group on AAVE syntax with John, Peter Sells, Arnold Zwicky, and (later) Lisa Green, who visited as a postdoc

• Won’t go into details about our analysis

AAVE as a Real Language

• Linguists claimed for decades that AAVE is not just bad English, “sloppy talk”, etc.

• But most previous work focused on social factors surrounding AAVE and its relationship to standard American English

• This paper was a formal syntax paper, whose object language happened to be AAVE

• Appeared the same year as the Ebonics controversy in Oakland

Syntactic Variation and OT

• This paper was also an early application of Optimality Theory to syntax

• Because the phenomenon exhibits variation, we proposed that individual speakers may allow multiple constraint rankings

• This idea was being explored in morphology at the same time by our then-student, Arto Anttila

• Anttila has since applied it very insightfully to syntactic variation, as well

The ALL Project

• “Intensive and Quotative ALL: Something Old, Something New” American Speech, 2007. Rickford, Buchstaller, Wasow, & Zwicky.

• “The Sociolinguistics of a Short-Lived Innovation: Tracing the development of quotative all across real and apparent time” under revision. Buchstaller, Rickford, Traugott, Wasow, & Zwicky.

• “The lady was al demonyak: historical aspects of Adverb all” English Language & Linguistics. Buchstaller & Traugott.

• Lots of other collaborators: In addition to co-authors, students Zoe Bogart, Tracy Conner, Rowyn McDonald, Nick Romero, Laura Whitton, Kathryn Campbell-Kibler, Lauren Hall-Lew, Laura Staum and others …

The PhenomenaShe was all mad at me.

Origins of the ALL Project: John’s Dream

Email sent 3/18/04, to full departmental mailing list

OK, this isn't Black History month, and this is not about MLK Jr.'s famous speech.

It IS about a dream I had a few hours ago that we (Stanford Linguistics faculty and students) were working on a common linguistic project that drew on the expertise and experience of our respective subfields, showcasing the value of collective effort on a scale unprecedented in linguistics, and highlighting what is distinctive about linguistics at Stanford…. I'm here at 5 am trying to see if there's any hope of making the dream a reality. We could devote a small portion of our time and say, one faculty meeting or colloquium a month to the collective enterprise, building up a pool of data and readings ..., and resulting in conference presentations and publications under a common name, like the "Stanford Linguistics Collective"

Data for ALL Study

• 1990/94 recordings of native California adolescents & young adults collected by Ann Wimmer (Stanford undergrad senior thesis) & Carmen Fought (Pitzer College, LA area)

• New 2005 recordings of high school & college students from Palo Alto, Stanford, & San Francisco

• Multi-source corpus--incl. examples from conversation & culled from publications (Waksler, American Speech 2001), web pages, TV series (Buffy the Vampire Slayer) & movies (Clueless). Lots of all tokens (597 intensifiers, 253 quotatives), but not accountable like recorded corpora.

• Google Newsgroups Corpus, 1981-2005. Billions of words, including at least 354 examples of quotative all.

Results: Intensive ALL

• Intensive all is old, examples going back to 11th & 13th c.

• But extension to full tensed verbs is new: she all walks in ... I all screamed...

• In 2005 recordings, most common with adjectives (all spastic), PPs (all in bed), & Verb+ing (all laughing)

• As a booster, a degree modifier that intensifies the property of its head, intensive all (7% in our corpus) is not as frequent as really (52%), so (19%), and very (9%), but it is more frequent than totally (3%), and is popular with adjectives denoting physical property (all shiny), age, color, & speed.

QuickTime™ and ampeg4 decompressor

are needed to see this picture.

Boy meets girl, girl teases boy, boy looks for something to destroy. He’s into her, she’s onto him, and that’s the way it’s always been. She’ll be with you if you want her to, unless she finds out that you do. Then somehow she won’t want to be, it turns around so suddenly. And

I’m like yeah, but she’s all no, and I’m all come on baby, let’s go, and she’s like I don’t think so, and I’m going...

The search for love and happiness turns out to be a game of chess. You can’t move or you flip the board, and you’re lying in pieces on the floor.

I’m like um, and she’s all hey, and I’m all come on baby, let’s play, and she’s like that’s okay and I’m going...

Every day I just want to say I love her madly, but I do it so badly, that when I do, I can't get through. If she even listens, she’s way off in the distance.

Quotatives: “I’m Like Yeah, But She’s All No” The Mr. T Experience

Success in these relationships rests more or less on gamesmanship, and these are ships that I can’t board, or keep in order or afford.

I’m like yeah, but she’s all no, and I’m all come on baby, lets go, and she’s like I don’t think so, and I’m going. I’m going. I’m going..

Results: Quotative ALL• It’s new: 1st noted 1982. (Switchboard, 1988-92 & Santa

Barbara Corpus of Spoken American English I each has only 1 example)

• In 1990/94 corpus, all is primary quotative introducer (all used 46%, like 17%, unframed 16%, say 11%, Other 8%, go 2%) & VARBRUL analysis shows primary favoring factor is Present tense, then Quoted Speech (vs. Thought), then Perseverance (quotative all in 5 preceding lines)

• In 2005 corpus, all much less frequent (4%) as quotative introducer, overtaken by like (69%), & all like has emerged. VARBRUL shows tense not significant, but Speech still favors, & perseverance disfavors all.

• Google newsgroups data also suggest quotative all peaked in 1999, then declined steeply.

Lessons from the ALL Project

• Even at its peak, quotative all had somewhat different usage patterns from other quotatives

• Linguistic innovations can be short-lived

• Mix of sociolinguistic interviews and corpus work can be productive

• Large, heterogeneous groups of linguists are capable of working together productively

Relativizer Omission

• Ongoing work -- no joint publications yet, but separate conference presentations (Tom et al since 2004, John & Laura 2008) & in press articles

• First time we both worked on the same alternation separately before collaborating

• Tom’s work mostly with Florian Jaeger, plus sometimes Roger Levy or David Orr

• John’s work w grad student Laura Smith & w undergrads Pat Callier, Cole Paulson, Doug Kenter plus consultant Bob Bayley (UCD)

The Phenomenon

• Alternation in relative clauses

The last movie which/that I saw was Ironman

The last movie Ø I saw was Ironman

• Both interested in factors influencing omission

• Tom’s work largely on the Switchboard corpus

• John studying AAVE and Appalachian English and Caribbean English creoles (in Jamaica, Guyana, Barbados)

Tom’s Issues

• What can relativizer omission tell us about syntactic processing?

• How do grammar and processing influence one another?

• Proposal from Wasow, Jaeger, & Orr (in press): The more predictable the occurrence of a non-subject relative clause is, the less likely it is to have a relativizer

• If this is due to processing, it should be the same across dialects

John’s Issues• Does relativizer choice (including omission)

pattern alike in AAVE, Appalachian (White vernacular) and Caribbean creoles?

• How do relativizer patterns in these lang-uages compare to those in colloquial and standard English? (NO quan. data avail.)

• What, if anything, can be concluded about the origins of AAVE from relativizer patterns (cf.Tottie & Harvie in Poplack, ed., 2000)?

• How test predictability hypothesis without tagged/parsed computer corpora?

Projects Mutually Reinforcing

• Predictability hypothesis needs to be tested on wider range of data

• Processing explanations of relativizer patterns undermine historical arguments based on them.

• AAVE, Appalachian & creoles among dialects that allow relativizer omission more freely in subject RCs:

He the man Ø got all the records

A yuh Ø mek dem bad (Lionheart Gal, p. 8,J’ca)

• This is another place to test both the predictability hypothesis and similarities across dialects

• So are it-clefts: It’s in the bed (that) they hid the gun.

Take Away Lessons• Don’t compartmentalize! Lx today requiring more breadth than

before, combining observations & experiments (Campbell-Kibler), diff subfields (phonetics/socio) & fields (Psych, Sociol)

• Language is a multi-faceted phenomenon

– Has an intricate internal structure, represented in the minds of speakers, influenced by human information processing mechanisms

– Is a powerful social tool, shaping and shaped by our interactions

– These facets are not independent

– We learn more about language when we attend to both

• Computational tools changing both sociolinguistics and syntax in ways that bring their methods and results closer together

• Larger theorizing & model building? (What we haven’t yet done)

Thanks to• NSF grant 1101381-100-QAOLE

(Rickford, PI)• Grants from Stanford Humanities Lab & the

Vice-Provost for Undergraduate Education• Too many individuals to fit on this slide,

notably our other collaborators & students• …and Beth Hume and the other meeting

organizers for inviting (and helping) us.• Slides available soon at

http://www.stanford.edu/~wasow/wasow.html