1. Some aspects of second-language phonology: The characterization of interlanguage grammars.
Fred R Eckman
University of Wisconsin-Milwaukee
Phonology Forum 19 – 21 August 2015
Osaka, Japan

2. Overview
Characterization of the goals and methods of second-language (L2) phonology
Interpretation of the progress in the field
Propose an answer:
What do we think we understand about the pronunciation patterns of L2 learners?

3. Overview
Divide research on L2 phonology into two fundamental periods, or research programs:
Pre-interlanguage hypothesis
Post-interlanguage hypothesis

4. Overview Interlanguage hypothesis
The claim that L2 learners construct their own version of the target language (TL) that they use to produce and comprehend utterances in the TL.

5. Overview
Many possible responses to the question of what we believe we understand;
Numerous interpretations of goals and findings of L2 phonology.

6. Overview Roadmap: Describe some terms, and lay out assumptions
Beginning: (pre-ILH) Predicting learner difficulty
Section 1: Contrastive Analysis Hypothesis
Section 2: Error Analysis: Systematicity of learner errors
Progress: (post-ILH) Learner-grammars
Section 3: Interlanguage Hypothesis (ILH)
Section 4: Characterization of IL grammars
Section 5: Intermediate stages of learning.
Conclusion

7. Introduction L2 versus L1
Time of life during acquisition
L1 = acquisition during early childhood
End state is always that of a native speaker

8. Introduction L2 versus L1
L2 = acquisition during adulthood
End state = non-native-speaker;
Pronunciation patterns generally distinct from native speakers;
Learner patterns characterized as errors;
Errors are assumed to derive from differences between the native language (NL) and target language (TL).

9. Introduction Fundamental assumption:
no one can give the language to a learner;
the learner must create the language.
The goal of the linguist is to understand the nature of the learning process.

10. Introduction
One of the major goals of L2 phonology, therefore, is the characterization and explanation of intermediate stages in the acquisition of the target-language (TL) pronunciation patterns.

11. Introduction Principal methodology:
Consider the target structure the learners are aiming for;
Note where the learners hit;
Theorize as to why the learners missed the target in the way they did.

12. Introduction Example: Children L1 Target word: ran [ɹæn]
Early stage: ran [ɹæn]
Later stage: *runned, *ranned [ɹʌnd], [ɹænd]
Final stage: ran [ɹæn]
Explanation: memorization followed by overgeneralization

13. Introduction Example: L2 learners Target word: tag
Target pronunciation: [tæɡ]
Learner pronunciation: [tæk]
Explanation (?)

14. Section 1 Early L2 Phonology
Explain/predict learning difficulty on the basis of differences between the NL and TL.

15. Section 1 Contrastive Analysis Hypothesis (CAH)
Lado (1957)
“We assume that the student who comes in contact with a foreign language will find some features of quite easy and others extremely difficult. Those elements that are similar to his native language will be simple for him, and those elements that are different will be difficult”. (Lado 1957: 2)

16. Section 1 CAH
Postulated two hypothetical constructs, NL & TL NL TL

17. Section 1 CAH
Claim: L2 learning difficulty, as reflected in learners’ errors, should always involve structures where the NL and TL are different.

18. Section 1: CAH Additional hypotheses
Hierarchies of difficulty; degree of difficulty: In general, allophones reflected maximum difficulty

19. Section 1: CAH Additional hypotheses
Degree of difficulty Maximum difficulty “… stated in more general terms, when one significant unit or element in the native language equates bilingually with two significant units in the target language, we have maximum learning difficulty.” (Lado 1957: 15)

20. Section 1: CAH Example
Splitting NL allophones into separate phonemes in the TL: NL Spanish: [d], [ð] allophones of /d/ TL English: /d/ & /ð/

21. Section 1: CAH Additional hypotheses
Hierarchies of difficulty
Stockwell & Bowen (1965)
Postulated 8 levels of difficulty based on whether the NL-TL distinctions involved phonemic contrasts, allophonic differences, or the absence on the sound segments in one of the languages.
The four highest levels of difficulty involved allophonic differences in either the NL, the TL or both.

22. Section 1: CAH Additional hypotheses
Hierarchies of difficulty
Hammerly (1982)
Postulated 6 levels of difficulty, with 4 of the highest levels involving allophonic distinctions in either the NL. TL or both.

23. Section 1: CAH Studies and results
Mixed support
Some findings showed that L2 errors can be attributed to NL-TL differences, e.g., Suter, (1976);
Some studies showed that L2 errors were independent of NL-TL differences, e.g., Wode, (1976).

24. Section 1: CAH Studies and results
Mixed support for CAH led to some linguists postulating a weaker form of the hypothesis in which some learner errors could possibly, but not necessarily, be explained on the basis of NL-TL differences.

25. Section 1: CAH Studies and results
“An evaluation of this strong version of the contrastive analysis hypothesis suggests that it makes demands of linguistic theory, and, therefore, of linguists, that they are in no position to meet.” (Wardhaugh, 1970: 125); “The weak version requires of the linguist only that he (sic) use the best linguistic knowledge available to him in order to account for observed difficulties in second language learning.” (Wardhaugh, 1970: 126)

26. Section 1: CAH Studies and results
“It does not require what the strong version requires, the prediction of those difficulties …. (Wardhaugh, 1970: 126) “The weak version leads to an approach which makes fewer demands of contrastive theory than does the strong version. It starts with the evidence provided by linguistic interference and uses such evidence to explain the similarities and differences between systems.” (Wardhaugh, 1970: 126)

27. Section 1: CAH Legacy
At least some NL-TL differences are recognized as being the explanation for L2 pronunciation patterns. There is no L2 phonological framework that excludes such differences as a possible factor in explaining L2 pronunciation patterns.

28. Section 1: CAH Legacy
Focus turned to the systematicity of the errors of L2 learners. Deviations from the TL pronunciation patterns (i.e., errors) were characterized in terms of learning strategies. This led to the explanatory framework of Error Analysis.

29. Section 2: Error Analysis
Focus on the systematicity of the L2 learners’ errors of L2;
EA allowed, but did not require, explanation in terms of NL-TL differences;
EA characterized the errors in terms of learning strategies

30. Section 2: Error Analysis
Language viewed as a system of rules;
One of the goals is to characterize the nature of this set of rules (grammar) and how they are learned.

31. Section 2: Error Analysis Sample L2 Data: NL = Spanish
L2 pronunciation TL pronunciation Gloss
1. [ɾɛt] [ɹɛd] red
[ɾɛðəɾ] [ɹɛɾəɹ] redder
[bik] [bɪɡ] big
[biɡəɾ] [bɪɡəɹ] bigger
[fʌsi] [fʌzi] fuzzy
6. [sik] [sɪk] sick
[sikəst] [sɪkəst] sickest
[wɛt] [wɛt] wet
[wɛtəɾ] [wɛtəɹ] wetter
[fɾis] [friz] freeze
11. [smut] [smuð] smooth
12. [smuðəɾ] [smuðəɹ] smoother

32. Section 2: Error Analysis Sample L2 Data
Several forms show the influence of the NL, e.g., (1, 2, 5, & 10); Others (1 & 3) are not NL-like; EA assumes that the L2 learner knows the TL-like forms, but cannot pronounce them; EA postulated generalizations to map TL forms onto L2 forms: Final voiced obstruents in the TL replaced by corresponding voiceless obstruents in the L2

33. Section 2: Error Analysis Legacy
Reflected the view that:
L2 learners’ errors are systematic;
The systematicity was across systems, between the TL and the L2 pronunciations;
it was assumed that the L2 learner knew the TL-like representations, but could not pronounce them;

34. Section 3 Interlanguage Hypothesis (ILH)
Language learners
construct their own system of the TL;
use this system to produce and comprehend utterances of the TL.

35. Section 3 Interlanguage Hypothesis
The construct of a learner language was proposed independently by three scholars:
Linguist Term given to learner-language system
Corder (1971): idiosyncratic dialects
Nemser (1971): approximative system
Selinker (1972): interlanguage

36. Section 3 Interlanguage HypothesisTL NL IL

37. Section 3 Interlanguage Hypothesis
IL grammar may contain:
structures from the NL;
structures learned on the basis of input from the TL, and
structures independent of both the NL and TL.

38. Section 3 Interlanguage Hypothesis
Postulation of a third construct
Empirical evidence for the necessity of an IL
regularity in the productions of an L2 learner that begs explanation
but cannot be linked to the NL, and
also cannot be attributed to the TL input, and
therefore necessitates the positing of a third system.

39. Section 3 Interlanguage Hypothesis
A consequence of the ILH is that L2 learning becomes the acquisition of an interlanguage.

40. Section 3: ILH Sample L2 Data: NL = Spanish (repeated from above)
L2 pronunciation TL pronunciation Gloss
1. [ɾɛt] [ɹɛd] red
[ɾɛðəɾ] [ɹɛɾəɹ] redder
[bik] [bɪɡ] big
[biɡəɾ] [bɪɡəɹ] bigger
5. [fʌsi] [fʌzi] fuzzy
6. [sik] [sɪk] sick
[sikəst] [sɪkəst] sickest
[wɛt] [wɛt] wet
[wɛtəɾ] [wɛtəɹ] wetter
[fɾis] [friz] freeze
11. [smut] [smuð] smooth
12. [smuðəɾ] [smuðəɹ] smoother

41. Section 3 Interlanguage Hypothesis
Empirical differences between EA analysis and IL analysis
EA assumes the learner knows or “has” TL forms and generalizations map differences between learner forms and TL forms;
IL analysis treats learner-data as part of a language, and analyzes them accordingly.

42. Section 3: ILH IL Analysis
Empirical differences between EA analysis and IL analysis
Evidence that the learner’s lexical representation for red and big are /ɾɛd/ and /biɡ/, respectively:
Alternations between word-final voiceless obstruents and intervocalic voiced obstruents

43. Section 3: ILH IL Analysis
Pedagogical differences between EA analysis and IL analysis
Evidence that the learner’s lexical representations for red and big are /ɾɛd/ and /biɡ/, respectively:
The IL system of this L2 learner has much that is TL-like, despite errors;
This L2 learner does not need to learn that the English word red ends in a final voiced obstruent;
This L2 learner needs to learn not to devoice the obstruent in final position.

44. Section 3: ILH IL Analysis
Pedagogical differences between EA analysis and IL analysis
Evidence that the learner’s lexical representation for smooth and smoother are /smud/ and /smudəɾ/, respectively:
This L2 learner has the near-TL-like pronunciation, [smuðəɾ], for the wrong reason;

45. Section 3: ILH The above IL pattern is independent of both NL and TL
does not derive from the NL, Spanish;
nor does it come from the TL, English.

46. Section 3: ILH
This pattern, exhibited by the grammars of a number of languages of the world:
Catalan
German
Polish
Russian

47. Section 3: ILH Language-hood of interlanguages:
an L2 pattern that is independent of both the NL and TL,
but that is attested in the grammars of other languages of the world.

48. Section 3: ILH Legacy Interlanguages
grammars constructed by L2 learners;
languages in their own right;

49. Section 3: ILH Legacy
Hypotheses about L2 acquisition are assertions about the state of a learner’s IL grammar, and therefore must be addressed by individual, rather than aggregate, data.

50. Section 3: ILH Legacy
Postulation of an IL constituted a break-through in L2 phonology in that it changed the research program from one of attempting to predict learning difficulty to one of explaining why IL grammars are the way they are.

51. Section 4 Characterization of IL grammars
Research programs in L2 phonology:
L2 phonologists try to understand IL grammars in terms of what phonologists believe they know about L1 grammars.
Explanation format: IL phonologies are the way they are, in part, because ILs are languages, and therefore subject to at least some of the same constraints as L1 grammars.

52. Section 4 Characterization of IL grammars
Some properties of IL grammars resemble L1 grammars, and other characteristics differ from L1 grammars;
Research programs in L2 phonology turned to theorizing about why ILs are the way they are on the basis of factors that are
Internal to the learner
External to the learner

53. Section 4 Characterization of IL grammars
Internal to the learner
Constraints on
grammars
learnability
simplicity

54. Section 4 Characterization of IL grammars
External to the learner
TL input: frequency of occurrence
Phonological saliency
Usage-based factors

55. Section 4 Characterization of IL grammars
Some L2 phonologists try to explain the nature of L2 phonologies by invoking principles used to characterize L1 phonologies.

56. Section 4 Characterization of IL grammars
Explanations of IL phonologies couched within the same frameworks as L1 phonologies
Generative phonology
Optimality Theory
Usage-based approaches
Phonetic approaches
Speech Learning Model (SLM; Flege, 1995)
Perception Assimilation Model (PAM; Best, 1995)

57. Section 4 Characterization of IL grammars
Issues in L2 phonologies are similar to those encountered in L1 phonologies
Competence versus performance
Variability

58. Section 5: IL as intermediate stage
Generative grammar: parameters Broselow & Finer (1991) The MSD was proposed to account for the systematic, cross-linguistic variation found among the world’s languages in allowable onset clusters.

59. Section 5: IL as intermediate stage
Generative grammar: parameters Broselow & Finer (1991)
Minimal Sonority Distance parameter in the acquisition of some English onset clusters by native speakers of Japanese and Korean.
Findings were that the attested IL grammars fell in between the onset cluster patterns of the NL and the TL, but that could nevertheless be characterized by the MSD.

60. Section 5: IL as intermediate stage
Phonetic models Flege’s (1995) Speech Learning Model (SLM) Best’s (1995) Perception Assimilation Model (PAM)

61. Section 5: IL as intermediate stage
SLM and PAM phonetic models
Based on careful phonetic analyses, these models have sought to explain the IL phonologies of L2 learners in terms of their perception of non-native speech as phonetic categories that are similar or different from the NL categories.

62. Section 5: IL as intermediate stage
Example of SLM postulate and hypothesis
Postulate 1: The mechanisms and processes used in learning the L1 sound system, including category formation, remain intact over the life span, and can be applied to L2 learning.
Hypothesis 1: Sounds in the L1 and L2 are related perceptually to one another at a position-sensitive allophonic level, rather than at a more abstract phonemic level.

63. Section 5: IL as intermediate stage
Sample study: Flege & Eefting (1987)
The study analyzed the production of onset stops in English by native speakers of Spanish pronounced English stops with Voice Onset Times (VOT) that were reliably longer than those of native English speakers, but were also significantly shorter than those of monolingual Spanish speakers.

64. Section 5: IL as intermediate stage
Optimality Theory (example)
Hancin-Bhatt (2000): Characterized the acquisition of English codas by native speakers of Thai.
OT analyses of L2 phonologies explicitly claim that IL phonologies (in terms of constraint rankings) differ from L1 phonologies in the same way that L1 phonologies differ from each other.

65. Section 5: IL as intermediate stage
Derived Environments (Eckman & Iverson, 2013)
Allophonic splits: “maximum difficulty” (Lado, 1957)
The acquisition of splitting NL allophones into TL phonemes involves two implicationally related stages.
NL allophonic pattern is transferred to IL.
Once the learner acquires the contrast in some words, the NL pattern can apply in the IL only in derived environments.

66. Section 5: IL as intermediate stage
Derived Environments: Allophonic splits = “maximum difficulty” (Lado, 1957)
Stages of acquisition:
No contrast – Learner errs on contrast in all words
Partial contrast – Learner has the contrast in some (morphologically simplex) words; Learner errs on contrast only across morpheme boundaries
Full contrast – Learner has mastered the contrast in all environments

67. Section 5 Interlanguage As intermediate stages of acquisition
Covert contrasts: A statistically significant acoustic distinction made by a learner between TL phonemes that is nevertheless not perceived by native speakers of the TL.

68. Section 5: IL as intermediate stage
Covert contrasts have been attested for more than 30 years in L1 acquisition and in disordered speech.
Macken & Barton (1980)
Gierut & Dinnsen (1986)

69. Section 5: IL as intermediate stage
Covert contrasts have been attested in L2 acquisition only recently and sparsely .
Lim & Oh (2008)
Eckman, Iverson & Song (2014)
Eckman, Iverson & Song (2015)

70. Section 5: IL as intermediate stage
Covert contrasts have been hypothesized as an intermediate stage in L1 acquisition and in research on disordered speech.
Findings suggest that transcription data alone are not sufficient for analysis of the acquisition of phonemic contrasts.
I will have more to say about covert contrasts on Friday.

71. Summary and Conclusions
Analysis of L2 phonology has taught us that
NL can constrain the acquisition process;
L2 “errors” are systematic and show as much about what a learner knows about the TL as what the learner does not know;
L2 learners construct their own version of the TL grammar;
L2 acquisition becomes the learning of an IL;
IL grammars are intermediate stages of acquisition.

72. Thank You!