Metalinguistic, Shmetalinguistic: the phonology of shmreduplication

1
Metalinguistic, Shmetalinguistic the phonology
of shm-reduplication

• Andrew Nevins, MIT
• Bert Vaux, Harvard
• Chicago Linguistics Society
• April 10, 2003

2
Standard Definitions of Shm

• In English, the initial consonants, if any, are
  replaced by a specified consonant or cluster.
  (McCarthy Prince 198671)
• What about breakfast-shmreakfast? (Not only used
  in Mall Rats, but the screen name of an AOL
  user).
• What about obscene, obschmene ?
• What about árcade, shmárcade ?

3
Research Questions

• How much variation is there in shm- reduplication
  across speakers?
• How does it vary depending on the input
• Segmental features
• Stress
• Compounds
• Phrases

4
Investigation of Shm- also Yields Research
Results on

• Phonological Landmarks
• Grammatical status of glides
• Avoidance Phenomena in Fixed-Segmentism
• Ineffability due to Lexical Blocking
• The syntax of echo reduplication

5
Anchor Points Phonological Landmarks

• Yu (2002), Nevins Vaux (2003a)
• Phonological Rules can only refer to a restricted
  set of Anchor Points.
• A process like infixation/reduplication cannot
  target anything besides
• First vowel, First consonant, First onset, First
  foot, stressed syllable, Last syllable. Rules
  cant count. They can only refer to prominent
  landmarks (Pierrehumbert and Nair 1995).

6
Sample Anchor Points for Infixation and
Reduplication

• First vowel
• ka-r-chet (Katu, Costello 1988)
• Ma-m-vit (Pima, Riggle 2003)
• First consonant
• g-ab-abuji (Mangarrayi, Downing 2002)
• n-ar-aho (Sundanese, Robbins 1959)
• k-ni-aati (Leti, Blevins 1999)
• Stressed syllable
• e-goddamned-váporate (English expletive, McCarthy
  1982)
• hu-ga-gándo (Chamorro, Topping 1973)

7
Research Hypothesis

• The restricted set of anchor points accounts for
  all of the variation in shm reduplication
• Speakers will vary as to where they insert shm-,
  but it will only be at an anchor point
• The underdetermined data in the input for shm-
  examples leads speakers to postulate divergent
  rule targets, compatible with the data for simple
  forms, but which reveal variation on more complex
  forms
• bagel-shmagel is compatible with first C, first
  onset, before first vowel,.. the learners
  choice among these will reveal itself only on
  breakfast

8
The Fundamental Operation for Shm

• Add a new precedence relation from the last
  segment of the input word to shm
• Add a new precedence relation from shm- to an
  anchor point (for most speakers, the first vowel)
• Raimy (2001) t? ?b ?l ?

e
a
?
m
We adopt Raimys model because it allows explicit
formalization of the target of a
morphophonological rule resulting in
reduplication
9
Alderete et al.s 1999 shm analysis

• Based on violation counting
• Table, ?m, RED are inputs
• MaxIO gtgt MaxBR, so including all inputs is more
  important than BR-identity
• But ?mtable is out one of ?m or t must be chosen
• MaxIO picks ?m over t since 2 violations worse
  than 1
• Counting predicts shmeel-shmeel should be optimal
  for eel ?m, and in simple reranking
  string-string optimal for string ?m
• Also fails for languages where phonotactics would
  allow FixSegInitial Onset but dont (e.g. Hindi
  roti-voti, vroti cf. also ? wig, shwig where
  BRgtgtIO)

10
What does shm- mean?

• Dismissive/Pejorative
• I care so little about ___ that I will pronounce
  it flagrantly incorrectly, so there comment
  from one respondent
• Topic-comment binomial pair, often with list
  intonation
• It means very in the case of fancy comment
  from one respondent

11
The history of shm
..der shm iz universalizirt gevorn ersht baym
sof nayntsetn yorhundert (Weinreich 1973)

• Mark Southern, Contagious Couplings
• Shm- arose from a mix of Turkic Echo m- and East
  Slavic sh-. First attested use in Yiddish
  Shmallig (c.1600 to disparage hallig holy)
• Arose in American English in late 19th century
  and became integrated into common usage in 1930s
  (Lockwood 1978) OED cites first use as crisis,
  shmisis (Gollers Five Books of Mr. Moses,
  1929)
• Yiddish had variants such as shlofn-pofn (sleep
  and stuff) and gelt-shpelt

12
Methodology

• How does shm-reduplication really work?
• http//www.ai.mit.edu/projects/dm/shm
• 55 questions, 180 respondents (to date)
• Questions are multiple choice
• Order of questions is randomized for each survey
  respondent

13
Sample Question

• Person A Who did that awful painting hanging in
  your basement?
• Person B (pointing at her husband) Umm, Lee did.
• Person A Lee did, _____! I know that you painted
  it, you untalented fop!
• A. Lee shmid
• B. Shmee shmid
• C. Shmee did
• D. Nothing sounds good here
• E. Other
• Comments on this question
  

14
Objects of inquiry I simple

• M-initial inputs (massage)
• Glide-initial inputs (union)
• Final-? Stress (arcade) not ambisyllabic
• Non-initial ? Stress (terrific) ?ambisyllabic
• Compounds (walkman, cookie jar)
• Phrasal inputs (Lee did, going to the beach, cats
  and dogs, figs and dates)

15
Objects of inquiry II trickier cases

• Can the output coincide with an existing word?
  Will intonation help?
• Waltz, ear, floozie, luck, Joe
• Do speakers allow for identical base and
  reduplicant?
• Schmidt, schmooze
• How are sequences of identical elements dealt
  with?
• Ishmael, Ashmont
• Will onset complexity be tolerated in
  reduplicant?
• Breakfast, shrapnel, schlitz, shnozz

16
Data will be presented in percentages

• Our survey presented questions in randomized
  order, and people sometimes quit after a few
  questions, so not all questions had equal numbers
  of responses

17
Phonological Ambiguity

• table-shmable is ambiguous the way I saw the man
  with the telescope is
• Shm-reduplication targets 1st syllable versus
  stressed syllable
• Second copy places shm at target in stressed
  syllable (terrific-teshmiffic), versus second
  copy starts with shm at target in stressed
  syllable (terrific-shmiffic)
• Shm-reduplication targets material after 1st
  consonant versus 1st onset

18
Complex Onsets

• Breakfast, shmreakfast (10)
• Schlitz, Shmlitz (8)
• Broom, Shmroom (36) on follow-up survey
• Dwarf, Shmwarf (29) on follow-up survey
• Floss, Shmloss (16) on follow-up survey
• These speakers target the segment immediately
  after the first consonant, rather than after the
  first onset

19
Union, Shmyoonion

• 20 kept the glide
• 76 didnt
• 14.3 kept it in confusion, conshmjusion
• 4 Wig, shmwig

20
Glide Variation Representational Status or Rule
Variable

• Barlow (2001) There are two representations
  people have for /spju/ one syllabifies /j/ in
  the onset, and one in the nucleus
• Nevins Vaux (2003a), Idsardi Raimy (2003)
  /j/ is always part of the nucleus. Variation in
  reduplication obtains due to two different rule
  variables

21
Two variables for rule targets

• First Nuclear Segment cute, shmjut
• First Vowel cute, shmute
• Speaker variation depends on the target of the
  reduplicative pointer.
• Advantage uniform representational status of
  prevocalic glide, explains why speakers may treat
  /ju/ sequences differently in different language
  games (which happens!)

22
anterior Dissimilation

• Apparent dissimilatory effect triggered by S tS
  Z s z
• tS witches, smitches (17) rich ? s- (10)
• S Ashmont smashmont (18)
• Ishmael smishmael (15)
• ash, smash (8) ash shmass (2)
• gibberish smibberish (3)
• sZ massage smassage (5)
• s circus, smircus (4)
• z schnozz smozz (3)

difference as function of distance?
reduced numbers due to lexical blocking by smash?
stridstrid X X
-ant aant
23
Non-Initial Stress Obscéne,

• Obschmene 44 and schmobscene 46!
• Respondents note understandable, undershmandable
  as well, and terrific, tershmiffic (6.5 kept
  r, suggesting ambisyllabicity for those
  speakers)
• Conclusion Shm- rule for some speakers must
  refer to stressed syllable
• Shm on both initial and stressed attested
  forbidden, shmorshmidden
• Targeting of stressed syllable not limited to
  2nd regulations, regushmations

24
Non-initial Stress

• Arcade, SHMARcade 46
• Arcade, shmarCADE 44
• Of those who stressed first syllable of
  reduplicant, they stress first syllable of arcade
  as well, due to either rhythm rule or desire for
  prosodic identity
• PERmit, SHMERmit -- 10 say can mean the noun
  (pérmit) as well as the verb (permít)

25
Copy starting from stressed ?

• Massage, Shmage (6) Copy from stressed syllable
  onward
• These speakers are also using stressed syllable
  as an anchor point (cf. ma-shmage), but
  truncating all preceding material
• (Also shmerage, sa-shmage, ma-shmage)
• Confusion, shmooshun attested in Spitzer 1952

26
Word Formation and Reduplication

• Lidz (2001) on Kannada echo can apply inside or
  outside of case marker -annu
• baagil-aanu giigil-annu (doors-acc. shmoors-acc.)
• baagil-giigil-annu (doors-shmoors-acc.)
• Variation in reduplicant is the result of
  triggering morpheme merging at the level of the
  NP or the DP

27
Variation in Compounds

• Shmookie cutter (69) 1st noun
• Shmookie shmutter (10) Both nouns
• Cookie shmutter (10) Second noun
• Compare exocentric compound walkman, shmalkman
  96
• Names Donald Shmumsfeld (26.5), Shmonald
  Rumsfeld (18.5)

28
IP, ShmIP?

• Jacob wants a laptop,
• Jacob wants a shmaptop (11.5)
• Ineffable for whole sentences (80)
• Going to the beach
• Shmoing to the beach (10)
• Going to the shmeach (12)

29
The Syntax of Shm-

• Travis (2001) Syntactic reduplication (e.g. I
  need a DOCTOR doctor) involves movement of a copy
  to check a feature.
• Shm-reduplication can never appear in argument
  position we argue it involves obligatory
  movement to TopP (often with resumption).
• I dont want to go to Europe, Shmeurope
• Europe, Shmeurope, who wants to go there!

30
Variation Recap

• First ? versus stressed ?
• massage shmassage mashmage
• First consonant versus first onset
• tree shmree shmee
• First nuclear segment versus first vowel
• union shmyoonion shmoonion
• Level of syntactic attachment
• cookie cutter cookie shmutter shmookie cutter

31
Non-identity Metalinguistic?

• even English speakers with little experience of
  the phenomenonreport that words already
  beginning with the cluster shm cannot enter into
  this pattern shmaltz-shmaltz (with the intended
  reading). An English speaker with considerable
  experience of the same phenomenon reports that
  words of this type systematically have initial
  shp instead shmaltz-shpaltz. (McCarthy and
  Prince 198668)
• words that already begin with shm- do not easily
  undergo this processthis can be attributed to a
  metalinguistic requirement that the reduplicant
  be different from the original word (Sanders
  2000)
• Direct counterexamples to universal
  anti-identity of echo words comes from Tamil ki-
  reduplication and Bengali ?- reduplication, which
  allow identity

32
Schmuck
This sounds whimsical. If shm- reduplication
comes from Yiddish, as does schmuck, and if part
of the humor is to use a Yiddishism, then it
sounds redundant to use shm-reduplication here.

• ineffable 70 81/115
• shluck 10 11/115
• shnuck 4 5/115
• shmuck 3 4/115
• fluck 3 3/115
• shpuck 2 2/115
• vluck, shmluck, shuck, shfuck, shvuck, smuck,
  fuck, shmukster, schmuck, schnook, this I know
  already!, my ass, (Bronx cheer)
• Similar results for schmooze (Q20)
• ineffable (62.5), shnooze (12.5), flooze (4),
  shmooze (4), shpooze (4), shlooze (3), vlooze,
  shplooze, shmmooze, mooze, wooze, commooze, my
  ass

33
Ineffability

• A fundamental axiom of OT is that all
  constraints are violable (Gerfen 2001)
• Problem ineffability/null outputs
• Some output should be optimal
• Prince and Smolensky 1993 Null Parse
• Vaux 2003
• Null Parse candidate has to be stipulated to
  satisfy all well-formedness and faithfulness
  constraints
• Fails to capture intuition that shm-reduplication
  of schmuck produces no output, rather than a
  phonetically null output (in fact, where do
  periphrastic outputs come from).
• Orgun and Sprouse 1999 ordering paradox in
  Turkish
• Null output for my C (doyum,doum,dom) requires
  Dep gtgt Mparse
• Epenthesis in cardat araba-ya requires
  Mparse gtgt Dep
• Conclusion some grammars must be able to contain
  inviolable constraints
• What drives the inviolable constraint involved in
  echo reduplication?
• Paradigm gap? (Albright, Bresnan, etc.)
• Spanish I abolish foregoed.forewent amnt
  in non-Irish English
• Cant be simple paradigm gap since ineffability
  with shm- is robustly productive
• Nonrecoverability?
• one might not know in schmidt schmidt, luck
  shmuck that shm-reduplication has occurred

34
Avoidance Phenomena

• When the fixed segmentism would result in
  identical base reduplicant, avoidance occurs
• Hindi aam-vaam, paani-vaani, vakil-shakil
• Turkish kitap-mitap, masa-masa
• Turkish cip-ciliz, dop-doluz, dim-dik
• Antifaithfulness does not explain why it is
  precisely the fixed segment that dissimilates

35
Coprecedence and Allomorphy

• Avoidance allomorphy looks like long-distance
  dissimilation. But when v- and v- are
  coprecedent, this is a local relationship, and
  provides a context for dissimilation.
• Shm- shm- dissimilation occurs not as a
  metalinguistic correction, but due to a simply
  stated allomorphy rule based on coprecedence

36
A precedes B, Z precedes B AZ are Coprecedents

• p aa n i - vaani
• p precedes aa
• aa precedes n
• n precedes i
• i precedes v
• v precedes a

p and v both precede aa They are coprecedent
In the cases of masa masa, vakil vakil, schmuck
schmuck, coprecedence between identical segments
obtains, providing a local context for a
dissimilation or allomorphy rule
37
Lexical Blocking

• Joe, luck, waltz, ear (QQ 39-42)
• Person A Joes a really cool guy. He gave me a
  ride home today when my car wasnt
  working.Person B Joe, _____! I really dont see
  why everyone thinks hes so great!
• shmo 66 77/116
• blocked 34 39/116
• Person A No one will date me, because my left
  ear is deformed.Person B Ear, _____. Youre
  very attractive, and youll find someone soon
  enough.
• shmear 88 98/111
• blocked 10 11/111
• smear 2 2/111

i think you could get away with it, but schmear
is already a word. not as bad as schmuck.
Schmeer, shmear 1. Bribery, corruption, flattery
2. the whole schmeer, everything, everything
possible or available, every aspect of the
situation. OED
38
Contributions

• Speakers have clear and consistent linguistic
  intuitions, suggesting that shm-reduplication is
  computed in the grammar
• Identification of several distinct subtypes of
  shm- reduplication
• variation is principled supports a theory of
  anchor points.
• (generally) formed through underdetermined
  exposure
• Ineffability phenomena require language-particular
  inviolable constraints (as in derivational
  phonological models or Orgun Sprouses Control
  Module)
• The target of dissimilation in avoidance is
  stateable through formal precedence relations and
  not blanket antifaithfulness

39
References
Alderete et. al. 1999. Reduplication with Fixed
Segmentism. Linguistic Inquiry/ ROA. Barlow,
Jessica. 2001. Individual Differences in the
production of initial consonant sequences in Pig
Latin. Lingua 111667-696. Idsardi, William
Eric Raimy. 2003. Phonological Representations
and the Delphic Oracle. Manuscript, University of
Delaware and Swarthmore College. McCarthy, John
and Alan Prince. 1986. Prosodic Morphology 1986.
Manuscript, Rutgers and UMass Amherst. Orgun,
Cemil Orhan and Ronald Sprouse. 1999. From MPARSE
to CONTROL deriving ungrammaticality. Phonology
16.2191-224. Fitzpatrick, Justin and Andrew
Nevins. 2002. Phonological Occurrences Relations
and Copying. Presented at North American
Phonology Conference (NaPhC) 2, Montreal. Lidz,
Jeffrey. 2001. Echo Formation in Kannada. The
Linguistic Review. Nevins, Andrew and Bert Vaux.
2003. Underdetermination in Language Games
Dialects of Pig Latin. Presented at the LSA,
Atlanta Raimy, Eric. 2001. The Phonology and
Morphology of Reduplication. Mouton. Travis,
Lisa. 2001. A syntacticians view of
reduplication. Presented at NELS. Vaux, Bert.
2003. Why the phonological component must be
serial and rule-based. Manuscript, Harvard
University (presented at LSA). Yu, Alan. 2002.
Understanding infixation as infixation. Presented
at NaPhC, on ROA.