A reanalysis of (non-)exceptional patterns in Bondu-so tongue root - PowerPoint PPT Presentation

1. ternary contrast on mid-vowel su�xes [ ATR] /-(i)loŋ/, [ ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ 2. incompatible with privative features (e.g. [ATR] /e, o/ vs. ∅ /ɛ, ɔ/) symmetric [ ATR] and [ ATR] su�x-controlled harmony neither feature is dominant/recessive (marked/unmarked) 3. directionally asymmetric bidirectional harmony lefuwards harmony bleeds rightwards harmony Basic generalisations Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Summary and implications : Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 5 / 46

[ ATR] /-(i)loŋ/, [ ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ 2. incompatible with privative features (e.g. [ATR] /e, o/ vs. ∅ /ɛ, ɔ/) symmetric [ ATR] and [ ATR] su�x-controlled harmony neither feature is dominant/recessive (marked/unmarked) 3. directionally asymmetric bidirectional harmony lefuwards harmony bleeds rightwards harmony Basic generalisations Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Summary and implications : 1. ternary contrast on mid-vowel su�xes Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 5 / 46

2. incompatible with privative features (e.g. [ATR] /e, o/ vs. ∅ /ɛ, ɔ/) symmetric [ ATR] and [ ATR] su�x-controlled harmony neither feature is dominant/recessive (marked/unmarked) 3. directionally asymmetric bidirectional harmony lefuwards harmony bleeds rightwards harmony Basic generalisations Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Summary and implications : 1. ternary contrast on mid-vowel su�xes ▶ [ + ATR] /-(i)loŋ/, [ − ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 5 / 46

symmetric [ ATR] and [ ATR] su�x-controlled harmony neither feature is dominant/recessive (marked/unmarked) 3. directionally asymmetric bidirectional harmony lefuwards harmony bleeds rightwards harmony Basic generalisations Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Summary and implications : 1. ternary contrast on mid-vowel su�xes ▶ [ + ATR] /-(i)loŋ/, [ − ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ 2. incompatible with privative features (e.g. [ATR] /e, o/ vs. ∅ /ɛ, ɔ/) Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 5 / 46

3. directionally asymmetric bidirectional harmony lefuwards harmony bleeds rightwards harmony Basic generalisations Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Summary and implications : 1. ternary contrast on mid-vowel su�xes ▶ [ + ATR] /-(i)loŋ/, [ − ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ 2. incompatible with privative features (e.g. [ATR] /e, o/ vs. ∅ /ɛ, ɔ/) ▶ symmetric [ + ATR] and [ − ATR] su�x-controlled harmony ▶ neither feature is dominant/recessive (marked/unmarked) Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 5 / 46

lefuwards harmony bleeds rightwards harmony Basic generalisations Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Summary and implications : 1. ternary contrast on mid-vowel su�xes ▶ [ + ATR] /-(i)loŋ/, [ − ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ 2. incompatible with privative features (e.g. [ATR] /e, o/ vs. ∅ /ɛ, ɔ/) ▶ symmetric [ + ATR] and [ − ATR] su�x-controlled harmony ▶ neither feature is dominant/recessive (marked/unmarked) 3. directionally asymmetric bidirectional harmony Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 5 / 46

Basic generalisations Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Summary and implications : 1. ternary contrast on mid-vowel su�xes ▶ [ + ATR] /-(i)loŋ/, [ − ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ 2. incompatible with privative features (e.g. [ATR] /e, o/ vs. ∅ /ɛ, ɔ/) ▶ symmetric [ + ATR] and [ − ATR] su�x-controlled harmony ▶ neither feature is dominant/recessive (marked/unmarked) 3. directionally asymmetric bidirectional harmony ▶ lefuwards harmony bleeds rightwards harmony Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 5 / 46

[dòɡ-ílòŋ] vs. [dɔ ̀ ɡ-íjɛ ́ ], not *[dɔ ̀ ɡ-ɪ́jɛ ́ ] displays 7V surface contrasts: paired [e, ɛ, o, ɔ] unpaired [i, u, a] – lacking *[ɪ, ʊ, ə] therefore do not display harmony alternations ( harmonically neutral ) What happens following unpaired /i, u, a/ harmony triggers? /bij-E/ ‘s/he laid down’ /ɡij-E/ ‘s/he killed’ /bar-E/ ‘s/he helped’ /paɡ-E/ ‘s/he tied’ * Unpaired high/low vowels trigger both [ ATR] harmony Abstract contrasts? Bondu-so vowel harmony is not always surface true Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 6 / 46

[dòɡ-ílòŋ] vs. [dɔ ̀ ɡ-íjɛ ́ ], not *[dɔ ̀ ɡ-ɪ́jɛ ́ ] therefore do not display harmony alternations ( harmonically neutral ) What happens following unpaired /i, u, a/ harmony triggers? /bij-E/ ‘s/he laid down’ /ɡij-E/ ‘s/he killed’ /bar-E/ ‘s/he helped’ /paɡ-E/ ‘s/he tied’ * Unpaired high/low vowels trigger both [ ATR] harmony Abstract contrasts? Bondu-so vowel harmony is not always surface true ▶ displays 7V surface contrasts: ▶ paired [e, ɛ, o, ɔ] ▶ unpaired [i, u, a] – lacking *[ɪ, ʊ, ə] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 6 / 46

What happens following unpaired /i, u, a/ harmony triggers? /bij-E/ ‘s/he laid down’ /ɡij-E/ ‘s/he killed’ /bar-E/ ‘s/he helped’ /paɡ-E/ ‘s/he tied’ * Unpaired high/low vowels trigger both [ ATR] harmony Abstract contrasts? Bondu-so vowel harmony is not always surface true ▶ displays 7V surface contrasts: ▶ paired [e, ɛ, o, ɔ] ▶ unpaired [i, u, a] – lacking *[ɪ, ʊ, ə] ▶ therefore do not display harmony alternations ( harmonically neutral ) ▶ [dòɡ-ílòŋ] vs. [dɔ ̀ ɡ-íjɛ ́ ], not *[dɔ ̀ ɡ-ɪ́jɛ ́ ] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 6 / 46

* Unpaired high/low vowels trigger both [ ATR] harmony Abstract contrasts? Bondu-so vowel harmony is not always surface true ▶ displays 7V surface contrasts: ▶ paired [e, ɛ, o, ɔ] ▶ unpaired [i, u, a] – lacking *[ɪ, ʊ, ə] ▶ therefore do not display harmony alternations ( harmonically neutral ) ▶ [dòɡ-ílòŋ] vs. [dɔ ̀ ɡ-íjɛ ́ ], not *[dɔ ̀ ɡ-ɪ́jɛ ́ ] What happens following unpaired /i, u, a/ harmony triggers? ▶ /bij-E/ → ⁇ ‘s/he laid down’ ▶ /ɡij-E/ → ⁇ ‘s/he killed’ ▶ /bar-E/ → ⁇ ‘s/he helped’ ▶ /paɡ-E/ → ⁇ ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 6 / 46

* Unpaired high/low vowels trigger both [ ATR] harmony Abstract contrasts? Bondu-so vowel harmony is not always surface true ▶ displays 7V surface contrasts: ▶ paired [e, ɛ, o, ɔ] ▶ unpaired [i, u, a] – lacking *[ɪ, ʊ, ə] ▶ therefore do not display harmony alternations ( harmonically neutral ) ▶ [dòɡ-ílòŋ] vs. [dɔ ̀ ɡ-íjɛ ́ ], not *[dɔ ̀ ɡ-ɪ́jɛ ́ ] What happens following unpaired /i, u, a/ harmony triggers? ▶ /bij-E/ → [bìj-è] ‘s/he laid down’ ▶ /ɡij-E/ → [ɡìj-ɛ ̀ ] ‘s/he killed’ ▶ /bar-E/ → [bàr–è] ‘s/he helped’ ▶ /paɡ-E/ → [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 6 / 46

Abstract contrasts? Bondu-so vowel harmony is not always surface true ▶ displays 7V surface contrasts: ▶ paired [e, ɛ, o, ɔ] ▶ unpaired [i, u, a] – lacking *[ɪ, ʊ, ə] ▶ therefore do not display harmony alternations ( harmonically neutral ) ▶ [dòɡ-ílòŋ] vs. [dɔ ̀ ɡ-íjɛ ́ ], not *[dɔ ̀ ɡ-ɪ́jɛ ́ ] What happens following unpaired /i, u, a/ harmony triggers? ▶ /bij-E/ → [bìj-è] ‘s/he laid down’ ▶ /ɡij-E/ → [ɡìj-ɛ ̀ ] ‘s/he killed’ ▶ /bar-E/ → [bàr–è] ‘s/he helped’ ▶ /paɡ-E/ → [pàɡ-ɛ ̀ ] ‘s/he tied’ * Unpaired high/low vowels trigger both [ ± ATR] harmony Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 6 / 46

* Harmony opacity via neutralisation: /ɡɪj-E/ /ɡɪj-ɛ/ [ɡìj-ɛ ̀ ] /bij-/ /ɡɪj-/ /suɡ-/ /ʤʊɡ-/ /bər-/ /paɡ-/ Abstract contrasts? (2) Distinct high/low vowel [ ± ATR]-harmony in Bondu-so [ + ATR] root [ − ATR] root [bìj-è] ‘s/he laid down’ [ɡìj-ɛ ̀ ] ‘s/he killed’ [sùɡ-è] ‘s/he went down’ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ [bàr–è] ‘s/he helped’ [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 7 / 46

* Harmony opacity via neutralisation: /ɡɪj-E/ /ɡɪj-ɛ/ [ɡìj-ɛ ̀ ] Abstract contrasts? (2) Distinct high/low vowel [ ± ATR]-harmony in Bondu-so [ + ATR] root [ − ATR] root /bij-/ [bìj-è] ‘s/he laid down’ /ɡɪj-/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-/ [sùɡ-è] ‘s/he went down’ /ʤʊɡ-/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bər-/ [bàr–è] ‘s/he helped’ /paɡ-/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 7 / 46

Abstract contrasts? (2) Distinct high/low vowel [ ± ATR]-harmony in Bondu-so [ + ATR] root [ − ATR] root /bij-/ [bìj-è] ‘s/he laid down’ /ɡɪj-/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-/ [sùɡ-è] ‘s/he went down’ /ʤʊɡ-/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bər-/ [bàr–è] ‘s/he helped’ /paɡ-/ [pàɡ-ɛ ̀ ] ‘s/he tied’ * Harmony opacity via neutralisation: /ɡɪj-E/ → /ɡɪj-ɛ/ → [ɡìj-ɛ ̀ ] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 7 / 46

Interim summary – the received story Bondu-so displays: ▶ bidirectional harmony * directionally asymmetric ▶ ternary contrasts on mid-vowel su�xes * not compatible with privative features ▶ abstract contrasts on high/low vowels * phonologically active but never surface Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 8 / 46

Outline 1 Introduction Background Problems with previous analyses 2 The reanalysis High/low vowel harmony neutrality In�ectional classes Reorganisation of the data Talk conclusions Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 9 / 46

bij-e ɡɪj-ɛ paɡ-ɛ bər-e – ɡij-ɛ – bar-e [bìj-è] [ɡìj-ɛ ̀ ] [pàɡ-ɛ] [bàr-è] Bondu-so involves a case of counterbleeding opacity (cf. Kiparsky 1973; Baković 2��9, 2�11): /ɪ, ə/ can trigger harmony but can’t surface surface [ α ATR]-harmony without obvious [ α ATR]-trigger (3) (3) Bondu-so harmony opacity via neutralisation /bij-E/ /ɡɪj-E/ /paɡ-E/ /bər-E/ Harmony Neutralisation ‘s/he laid down’ ‘s/he killed’ ‘s/he tied’ ‘s/he helped’ Problems Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 1� / 46

bij-e ɡɪj-ɛ paɡ-ɛ bər-e – ɡij-ɛ – bar-e [bìj-è] [ɡìj-ɛ ̀ ] [pàɡ-ɛ] [bàr-è] /ɪ, ə/ can trigger harmony but can’t surface surface [ α ATR]-harmony without obvious [ α ATR]-trigger (3) (3) Bondu-so harmony opacity via neutralisation /bij-E/ /ɡɪj-E/ /paɡ-E/ /bər-E/ Harmony Neutralisation ‘s/he laid down’ ‘s/he killed’ ‘s/he tied’ ‘s/he helped’ Problems Bondu-so involves a case of counterbleeding opacity (cf. Kiparsky 1973; Baković 2��9, 2�11): Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 1� / 46

/ɪ, ə/ can trigger harmony but can’t surface surface [ α ATR]-harmony without obvious [ α ATR]-trigger (3) bij-e ɡɪj-ɛ paɡ-ɛ bər-e – ɡij-ɛ – bar-e [bìj-è] [ɡìj-ɛ ̀ ] [pàɡ-ɛ] [bàr-è] Problems Bondu-so involves a case of counterbleeding opacity (cf. Kiparsky 1973; Baković 2��9, 2�11): (3) Bondu-so harmony opacity via neutralisation /bij-E/ /ɡɪj-E/ /paɡ-E/ /bər-E/ Harmony Neutralisation ‘s/he laid down’ ‘s/he killed’ ‘s/he tied’ ‘s/he helped’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 1� / 46

/ɪ, ə/ can trigger harmony but can’t surface surface [ α ATR]-harmony without obvious [ α ATR]-trigger (3) – ɡij-ɛ – bar-e [bìj-è] [ɡìj-ɛ ̀ ] [pàɡ-ɛ] [bàr-è] Problems Bondu-so involves a case of counterbleeding opacity (cf. Kiparsky 1973; Baković 2��9, 2�11): (3) Bondu-so harmony opacity via neutralisation /bij-E/ /ɡɪj-E/ /paɡ-E/ /bər-E/ Harmony bij-e ɡɪj-ɛ paɡ-ɛ bər-e Neutralisation ‘s/he laid down’ ‘s/he killed’ ‘s/he tied’ ‘s/he helped’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 1� / 46

surface [ α ATR]-harmony without obvious [ α ATR]-trigger (3) [bìj-è] [ɡìj-ɛ ̀ ] [pàɡ-ɛ] [bàr-è] Problems Bondu-so involves a case of counterbleeding opacity (cf. Kiparsky 1973; Baković 2��9, 2�11): ▶ /ɪ, ə/ can trigger harmony but can’t surface (3) Bondu-so harmony opacity via neutralisation /bij-E/ /ɡɪj-E/ /paɡ-E/ /bər-E/ Harmony bij-e ɡɪj-ɛ paɡ-ɛ bər-e Neutralisation – ɡij-ɛ – bar-e ‘s/he laid down’ ‘s/he killed’ ‘s/he tied’ ‘s/he helped’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 1� / 46

Problems Bondu-so involves a case of counterbleeding opacity (cf. Kiparsky 1973; Baković 2��9, 2�11): ▶ /ɪ, ə/ can trigger harmony but can’t surface ▶ surface [ α ATR]-harmony without obvious [ α ATR]-trigger (3) (3) Bondu-so harmony opacity via neutralisation /bij-E/ /ɡɪj-E/ /paɡ-E/ /bər-E/ Harmony bij-e ɡɪj-ɛ paɡ-ɛ bər-e Neutralisation – ɡij-ɛ – bar-e [bìj-è] [ɡìj-ɛ ̀ ] [pàɡ-ɛ] [bàr-è] ‘s/he laid down’ ‘s/he killed’ ‘s/he tied’ ‘s/he helped’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 1� / 46

ʧe – ʧi-u – tu ʧ-u [ʧe] [tu] [ ʧu ] the processes simply need to be independently motivated, as in (4) (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ i Palatalisation t ʧ e Deletion V ∅ V Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

ʧe – ʧi-u – tu ʧ-u [ʧe] [tu] [ ʧu ] (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ i Palatalisation t ʧ e Deletion V ∅ V Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: ▶ the processes simply need to be independently motivated, as in (4) Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

ʧe – ʧi-u – tu ʧ-u [ʧe] [tu] [ ʧu ] Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: ▶ the processes simply need to be independently motivated, as in (4) (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ { i Palatalisation t → ʧ / e Deletion V → ∅ / V Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

ʧi-u – tu ʧ-u [ʧe] [tu] [ ʧu ] Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: ▶ the processes simply need to be independently motivated, as in (4) (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ { i Palatalisation t → ʧ / ʧe – e Deletion V → ∅ / V Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

ʧi-u ʧ-u [ʧe] [tu] [ ʧu ] Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: ▶ the processes simply need to be independently motivated, as in (4) (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ { i Palatalisation t → ʧ / ʧe – e Deletion V → ∅ / V – tu Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

ʧi-u ʧ-u [ ʧu ] Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: ▶ the processes simply need to be independently motivated, as in (4) (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ { i Palatalisation t → ʧ / ʧe – e Deletion V → ∅ / V – tu [ʧe] [tu] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

ʧi-u ʧ-u Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: ▶ the processes simply need to be independently motivated, as in (4) (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ { i Palatalisation t → ʧ / ʧe – e Deletion V → ∅ / V – tu [ʧe] [tu] [ ʧu ] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

Acquisition of opaque patterns Vaux (2��8: p. 32) argues opaque patterns are �ne: ▶ the processes simply need to be independently motivated, as in (4) (4) Hypothetical counterbleeding opacity /te/ /to-u/ /ti-u/ { i Palatalisation t → ʧ / ʧe – ʧi-u e Deletion V → ∅ / V – tu ʧ-u [ʧe] [tu] [ ʧu ] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 11 / 46

/m?n/ /ɡɪj-E/ dɔɡ-ɛ – ɡɪj-ɛ – ⁇ ɡij-ɛ [mín] [ɡìj-ɛ ̀ ] ‘s/he waited’ ‘s/he killed’ Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) * neutralisation is not independently motivated, cf. (5) (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ Harmony Neutralisation [dɔ ̀ ɡ-ɛ ̀ ] ‘s/he lefu (it)’ The problem with opacity via absolute neutralisation Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

/m?n/ /ɡɪj-E/ dɔɡ-ɛ – ɡɪj-ɛ – ⁇ ɡij-ɛ [mín] [ɡìj-ɛ ̀ ] ‘s/he waited’ ‘s/he killed’ * neutralisation is not independently motivated, cf. (5) (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ Harmony Neutralisation [dɔ ̀ ɡ-ɛ ̀ ] ‘s/he lefu (it)’ The problem with opacity via absolute neutralisation Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

/m?n/ /ɡɪj-E/ dɔɡ-ɛ – ɡɪj-ɛ – ⁇ ɡij-ɛ [mín] [ɡìj-ɛ ̀ ] ‘s/he waited’ ‘s/he killed’ (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ Harmony Neutralisation [dɔ ̀ ɡ-ɛ ̀ ] ‘s/he lefu (it)’ The problem with opacity via absolute neutralisation Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) * neutralisation is not independently motivated, cf. (5) Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

/m?n/ /ɡɪj-E/ dɔɡ-ɛ – ɡɪj-ɛ – ⁇ ɡij-ɛ [mín] [ɡìj-ɛ ̀ ] ‘s/he waited’ ‘s/he killed’ The problem with opacity via absolute neutralisation Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) * neutralisation is not independently motivated, cf. (5) (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ Harmony Neutralisation [dɔ ̀ ɡ-ɛ ̀ ] ‘s/he lefu (it)’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

/m?n/ /ɡɪj-E/ – ɡɪj-ɛ ⁇ ɡij-ɛ [mín] [ɡìj-ɛ ̀ ] ‘s/he waited’ ‘s/he killed’ The problem with opacity via absolute neutralisation Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) * neutralisation is not independently motivated, cf. (5) (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ Harmony dɔɡ-ɛ Neutralisation – [dɔ ̀ ɡ-ɛ ̀ ] ‘s/he lefu (it)’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

/m?n/ /ɡɪj-E/ – ɡɪj-ɛ ⁇ ɡij-ɛ [ɡìj-ɛ ̀ ] ‘s/he killed’ The problem with opacity via absolute neutralisation Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) * neutralisation is not independently motivated, cf. (5) (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ Harmony dɔɡ-ɛ Neutralisation – [dɔ ̀ ɡ-ɛ ̀ ] [mín] ‘s/he lefu (it)’ ‘s/he waited’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

/ɡɪj-E/ ɡɪj-ɛ ɡij-ɛ [ɡìj-ɛ ̀ ] ‘s/he killed’ The problem with opacity via absolute neutralisation Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) * neutralisation is not independently motivated, cf. (5) (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ /m?n/ Harmony dɔɡ-ɛ – Neutralisation – ⁇ [dɔ ̀ ɡ-ɛ ̀ ] [mín] ‘s/he lefu (it)’ ‘s/he waited’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

The problem with opacity via absolute neutralisation Requirement of independent motivation is a ‘handicap for abstract analyses’ (Baković 2��9: p. 11) * neutralisation is not independently motivated, cf. (5) (5) Counterbleeding opacity in Bondu-so /dɔɡ-E/ /m?n/ /ɡɪj-E/ Harmony dɔɡ-ɛ – ɡɪj-ɛ Neutralisation – ⁇ ɡij-ɛ [dɔ ̀ ɡ-ɛ ̀ ] [mín] [ɡìj-ɛ ̀ ] ‘s/he lefu (it)’ ‘s/he waited’ ‘s/he killed’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 12 / 46

* Non-falsi�able : no independent way to con�rm/disprove abstract contrasts Can’t be observed; don’t turn up in acoustic analysis Don’t �gure in any other linguistic pattern * Circular : abstract contrasts only evidenced by distinct patterns in (2), which they’re supposed to explain (2) Distinct high/low vowel [ ATR]-harmony in Bondu-so [ ATR] root [ ATR] root /bij-/ [bìj-è] ‘s/he laid down’ /ɡɪj-/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-/ [sùɡ-è] ‘s/he went down’ /ʤʊɡ-/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bər-/ [bàr–è] ‘s/he helped’ /paɡ-/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Consequences of absolute neutralisation Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 13 / 46

Can’t be observed; don’t turn up in acoustic analysis Don’t �gure in any other linguistic pattern * Circular : abstract contrasts only evidenced by distinct patterns in (2), which they’re supposed to explain (2) Distinct high/low vowel [ ATR]-harmony in Bondu-so [ ATR] root [ ATR] root /bij-/ [bìj-è] ‘s/he laid down’ /ɡɪj-/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-/ [sùɡ-è] ‘s/he went down’ /ʤʊɡ-/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bər-/ [bàr–è] ‘s/he helped’ /paɡ-/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Consequences of absolute neutralisation * Non-falsi�able : no independent way to con�rm/disprove abstract contrasts Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 13 / 46

* Circular : abstract contrasts only evidenced by distinct patterns in (2), which they’re supposed to explain (2) Distinct high/low vowel [ ATR]-harmony in Bondu-so [ ATR] root [ ATR] root /bij-/ [bìj-è] ‘s/he laid down’ /ɡɪj-/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-/ [sùɡ-è] ‘s/he went down’ /ʤʊɡ-/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bər-/ [bàr–è] ‘s/he helped’ /paɡ-/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Consequences of absolute neutralisation * Non-falsi�able : no independent way to con�rm/disprove abstract contrasts ▶ Can’t be observed; don’t turn up in acoustic analysis ▶ Don’t �gure in any other linguistic pattern Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 13 / 46

Consequences of absolute neutralisation * Non-falsi�able : no independent way to con�rm/disprove abstract contrasts ▶ Can’t be observed; don’t turn up in acoustic analysis ▶ Don’t �gure in any other linguistic pattern * Circular : abstract contrasts only evidenced by distinct patterns in (2), which they’re supposed to explain (2) Distinct high/low vowel [ ± ATR]-harmony in Bondu-so [ + ATR] root [ − ATR] root /bij-/ [bìj-è] ‘s/he laid down’ /ɡɪj-/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-/ [sùɡ-è] ‘s/he went down’ /ʤʊɡ-/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bər-/ [bàr–è] ‘s/he helped’ /paɡ-/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 13 / 46

Consequences of absolute neutralisation * Theoretically/typologically irregular implications : ▶ bidirectional harmony * directionally asymmetric ▶ ternary contrasts on mid-vowel su�xes * not compatible with privative features ▶ abstract contrasts on high/low vowels * phonologically active but never surface Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 14 / 46

Root vowel representations Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations ▶ ‘correlation doesn’t imply causation’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations ▶ ‘correlation doesn’t imply causation’ ▶ [nòj-è] and [dɔ ̀ ɡ-ɛ ̀ ] are correlated for the harmony feature Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations ▶ ‘correlation doesn’t imply causation’ ▶ [nòj-è] and [dɔ ̀ ɡ-ɛ ̀ ] are correlated for the harmony feature ▶ but what’s the trigger and what’s the target? Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations ▶ ‘correlation doesn’t imply causation’ ▶ [nòj-è] and [dɔ ̀ ɡ-ɛ ̀ ] are correlated for the harmony feature ▶ but what’s the trigger and what’s the target? ▶ Is it /dɔ ̀ ɡ-E/ → [dɔ ̀ ɡ-ɛ ̀ ]? Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations ▶ ‘correlation doesn’t imply causation’ ▶ [nòj-è] and [dɔ ̀ ɡ-ɛ ̀ ] are correlated for the harmony feature ▶ but what’s the trigger and what’s the target? ▶ Is it /dɔ ̀ ɡ-E/ → [dɔ ̀ ɡ-ɛ ̀ ]? ▶ or /dOɡ-ɛ ̀ / → [dɔ ̀ ɡ-ɛ ̀ ]? Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations The question comes down to where the underlying contrast is – root or su�x? Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ − ATR] root /dɔɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations The question comes down to where the underlying contrast is – root or su�x? ▶ the near minimal pairs [bìj-è] and [ɡìj-ɛ ̀ ] suggest the su�x is contrastive Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ + ATR] root /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations The question comes down to where the underlying contrast is – root or su�x? ▶ the near minimal pairs [bìj-è] and [ɡìj-ɛ ̀ ] suggest the su�x is contrastive Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

Where have we gone wrong? Bidirectional [ + ATR] and [ − ATR] harmony in Bondu-so (1) UR of root Underspeci�ed su�x [ + ATR] su�x [ − ATR] su�x (perfective) (in�nitive) (mediopassive) [ + ATR] root /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [ + ATR] root /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Root vowel representations The question comes down to where the underlying contrast is – root or su�x? ▶ the near minimal pairs [bìj-è] and [ɡìj-ɛ ̀ ] suggest the su�x is contrastive 1. /dɔ ̀ ɡ-E/ → [dɔ ̀ ɡ-ɛ ̀ ] 2. /doɡ-ɛ ̀ / → [dɔ ̀ ɡ-ɛ ̀ ] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 15 / 46

(6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Preview: reanalysis implications Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

Preview: reanalysis implications (6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

Preview: reanalysis implications (6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive imperative Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [nój-ó] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] [dóɡ-á] Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

1. directionally asymmetric bidirectional harmony only uni-directional su�x-controlled harmony 2. ternary contrast on mid-vowel su�xes [ ATR] /-(i)loŋ/, [ ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ only [RTR] /ɛ, ɔ/ ~ (non-RTR) /e, o/ Preview: reanalysis implications (6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive imperative Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [nój-ó] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] [dóɡ-á] Important di�erences: Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

only uni-directional su�x-controlled harmony 2. ternary contrast on mid-vowel su�xes [ ATR] /-(i)loŋ/, [ ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ only [RTR] /ɛ, ɔ/ ~ (non-RTR) /e, o/ Preview: reanalysis implications (6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive imperative Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [nój-ó] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] [dóɡ-á] Important di�erences: 1. directionally asymmetric bidirectional harmony Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

2. ternary contrast on mid-vowel su�xes [ ATR] /-(i)loŋ/, [ ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ only [RTR] /ɛ, ɔ/ ~ (non-RTR) /e, o/ Preview: reanalysis implications (6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive imperative Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [nój-ó] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] [dóɡ-á] Important di�erences: 1. directionally asymmetric bidirectional harmony ▶ only uni-directional su�x-controlled harmony Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

only [RTR] /ɛ, ɔ/ ~ (non-RTR) /e, o/ Preview: reanalysis implications (6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive imperative Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [nój-ó] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] [dóɡ-á] Important di�erences: 1. directionally asymmetric bidirectional harmony ▶ only uni-directional su�x-controlled harmony 2. ternary contrast on mid-vowel su�xes ▶ [ + ATR] /-(i)loŋ/, [ − ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

Preview: reanalysis implications (6) Harmony variation across Bondu-so verbal classes UR of root perfective in�nitive mediopassive imperative Class A /noj-/ ‘sleep’ [nòj-è] [nój-ílòŋ] [nɔ ̀ j-íjɛ ́ ] [nój-ó] Class B /doɡ-/ ‘leave’ [dɔ ̀ ɡ-ɛ ̀ ] [dòɡ-ílòŋ] [dɔ ̀ ɡ-íjɛ ́ ] [dóɡ-á] Important di�erences: 1. directionally asymmetric bidirectional harmony ▶ only uni-directional su�x-controlled harmony 2. ternary contrast on mid-vowel su�xes ▶ [ + ATR] /-(i)loŋ/, [ − ATR] /-ijɛ/, ∅ (underspeci�ed) /-E/ ▶ only [RTR] /ɛ, ɔ/ ~ (non-RTR) /e, o/ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 16 / 46

only concrete or non-abstract /i, u, a/ 3. Abstract contrasts on high/low vowels /bij-E/ and /ɡɪj-E/ 4. Harmony counterbleeding opacity via neutralisation /ɡɪj-E/ /ɡɪj-ɛ/ [ɡìj-ɛ ̀ ] only transparent harmony neutrality: /ɡij-ɛ/ [ɡìj-ɛ ̀ ] Preview: reanalysis implications Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 17 / 46

only concrete or non-abstract /i, u, a/ 4. Harmony counterbleeding opacity via neutralisation /ɡɪj-E/ /ɡɪj-ɛ/ [ɡìj-ɛ ̀ ] only transparent harmony neutrality: /ɡij-ɛ/ [ɡìj-ɛ ̀ ] Preview: reanalysis implications 3. Abstract contrasts on high/low vowels ▶ /bij-E/ and /ɡɪj-E/ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 17 / 46

4. Harmony counterbleeding opacity via neutralisation /ɡɪj-E/ /ɡɪj-ɛ/ [ɡìj-ɛ ̀ ] only transparent harmony neutrality: /ɡij-ɛ/ [ɡìj-ɛ ̀ ] /bij-e/ /ɡij-ɛ/ /suɡ-e/ /ʤuɡ-ɛ/ /bar-e/ /paɡ-ɛ/ Preview: reanalysis implications 3. Abstract contrasts on high/low vowels ▶ /bij-E/ and /ɡɪj-E/ ▶ only concrete or non-abstract /i, u, a/ (7) No abstract contrasts: non-contrastive high/low vowels are harmonically neutral non-targets of tongue root harmony ATR class /-e/ RTR class /-ɛ/ [bìj-è] ‘s/he laid down’ [ɡìj-ɛ ̀ ] ‘s/he killed’ [sùɡ-è] ‘s/he went down’ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ [bàr–è] ‘s/he helped’ [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 17 / 46

4. Harmony counterbleeding opacity via neutralisation /ɡɪj-E/ /ɡɪj-ɛ/ [ɡìj-ɛ ̀ ] only transparent harmony neutrality: /ɡij-ɛ/ [ɡìj-ɛ ̀ ] Preview: reanalysis implications 3. Abstract contrasts on high/low vowels ▶ /bij-E/ and /ɡɪj-E/ ▶ only concrete or non-abstract /i, u, a/ (7) No abstract contrasts: non-contrastive high/low vowels are harmonically neutral non-targets of tongue root harmony ATR class /-e/ RTR class /-ɛ/ /bij-e/ [bìj-è] ‘s/he laid down’ /ɡij-ɛ/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-e/ [sùɡ-è] ‘s/he went down’ /ʤuɡ-ɛ/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bar-e/ [bàr–è] ‘s/he helped’ /paɡ-ɛ/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 17 / 46

only transparent harmony neutrality: /ɡij-ɛ/ [ɡìj-ɛ ̀ ] Preview: reanalysis implications 3. Abstract contrasts on high/low vowels ▶ /bij-E/ and /ɡɪj-E/ ▶ only concrete or non-abstract /i, u, a/ 4. Harmony counterbleeding opacity via neutralisation ▶ /ɡɪj-E/ → /ɡɪj-ɛ/ → [ɡìj-ɛ ̀ ] (7) No abstract contrasts: non-contrastive high/low vowels are harmonically neutral non-targets of tongue root harmony ATR class /-e/ RTR class /-ɛ/ /bij-e/ [bìj-è] ‘s/he laid down’ /ɡij-ɛ/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-e/ [sùɡ-è] ‘s/he went down’ /ʤuɡ-ɛ/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bar-e/ [bàr–è] ‘s/he helped’ /paɡ-ɛ/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 17 / 46

Preview: reanalysis implications 3. Abstract contrasts on high/low vowels ▶ /bij-E/ and /ɡɪj-E/ ▶ only concrete or non-abstract /i, u, a/ 4. Harmony counterbleeding opacity via neutralisation ▶ /ɡɪj-E/ → /ɡɪj-ɛ/ → [ɡìj-ɛ ̀ ] ▶ only transparent harmony neutrality: /ɡij-ɛ/ → [ɡìj-ɛ ̀ ] (7) No abstract contrasts: non-contrastive high/low vowels are harmonically neutral non-targets of tongue root harmony ATR class /-e/ RTR class /-ɛ/ /bij-e/ [bìj-è] ‘s/he laid down’ /ɡij-ɛ/ [ɡìj-ɛ ̀ ] ‘s/he killed’ /suɡ-e/ [sùɡ-è] ‘s/he went down’ /ʤuɡ-ɛ/ [ʤùɡ-ɛ ̀ ] ‘s/he recognised’ /bar-e/ [bàr–è] ‘s/he helped’ /paɡ-ɛ/ [pàɡ-ɛ ̀ ] ‘s/he tied’ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 17 / 46

eliminates all typologically and theoretically controversial generalisations In sum: // reinterpreting the direction of harmony in ambiguous cases (dɔ ̀ ɡɛ ̀ ) Preview: reanalysis implications Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 18 / 46

Preview: reanalysis implications In sum: // ▶ reinterpreting the direction of harmony in ambiguous cases (dɔ ̀ ɡɛ ̀ ) ▶ eliminates all typologically and theoretically controversial generalisations Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 18 / 46

Outline 1 Introduction Background Problems with previous analyses 2 The reanalysis High/low vowel harmony neutrality In�ectional classes Reorganisation of the data Talk conclusions Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 19 / 46

Outline 1 Introduction Background Problems with previous analyses 2 The reanalysis High/low vowel harmony neutrality In�ectional classes Reorganisation of the data Talk conclusions Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 2� / 46

If high/low vowels don’t trigger harmony (e.g. /ɡɪj-E/ /ɡɪj-ɛ/ [ɡìj-ɛ ̀ ]) then what is their actual behaviour? High/low vowel harmony neutrality Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 21 / 46

High/low vowel harmony neutrality If high/low vowels don’t trigger harmony (e.g. /ɡɪj-E/ → /ɡɪj-ɛ/ → [ɡìj-ɛ ̀ ]) ▶ then what is their actual behaviour? Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 21 / 46

Harmonically unpaired /i, u, a/ vowels are harmonically neutral (8) Bondu-so high and low vowel transparency a. /keʤ-iloŋ/ [kéʤ-ìlòŋ] ‘cut’-inf. b. /keʤ-ijɛ/ [kɛ ́ ʤ-íjɛ ́ ] ‘cut’-med-pass. c. /sem-anʤ-e/ [sém-ánʤ-è] ‘slaughter’-imperf.-2.pl. d. /sem-anʤ-ɛɛ/ [sɛ ́ m-ánʤ-ɛ ́ ɛ ̀ ] ‘slaughter’-imperf.-3.pl. High and low vowels are in other words phonologically inactive and invisible non-targets and non-triggers (transparent segments) High/low vowel harmony neutrality Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 22 / 46

(8) Bondu-so high and low vowel transparency a. /keʤ-iloŋ/ [kéʤ-ìlòŋ] ‘cut’-inf. b. /keʤ-ijɛ/ [kɛ ́ ʤ-íjɛ ́ ] ‘cut’-med-pass. c. /sem-anʤ-e/ [sém-ánʤ-è] ‘slaughter’-imperf.-2.pl. d. /sem-anʤ-ɛɛ/ [sɛ ́ m-ánʤ-ɛ ́ ɛ ̀ ] ‘slaughter’-imperf.-3.pl. High and low vowels are in other words phonologically inactive and invisible non-targets and non-triggers (transparent segments) High/low vowel harmony neutrality Harmonically unpaired /i, u, a/ vowels are harmonically neutral Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 22 / 46

High and low vowels are in other words phonologically inactive and invisible non-targets and non-triggers (transparent segments) High/low vowel harmony neutrality Harmonically unpaired /i, u, a/ vowels are harmonically neutral (8) Bondu-so high and low vowel transparency a. /keʤ-iloŋ/ [kéʤ-ìlòŋ] ‘cut’-inf. b. /keʤ-ijɛ/ [kɛ ́ ʤ-íjɛ ́ ] ‘cut’-med-pass. c. /sem-anʤ-e/ [sém-ánʤ-è] ‘slaughter’-imperf.-2.pl. d. /sem-anʤ-ɛɛ/ [sɛ ́ m-ánʤ-ɛ ́ ɛ ̀ ] ‘slaughter’-imperf.-3.pl. Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 22 / 46

High/low vowel harmony neutrality Harmonically unpaired /i, u, a/ vowels are harmonically neutral (8) Bondu-so high and low vowel transparency a. /keʤ-iloŋ/ [kéʤ-ìlòŋ] ‘cut’-inf. b. /keʤ-ijɛ/ [kɛ ́ ʤ-íjɛ ́ ] ‘cut’-med-pass. c. /sem-anʤ-e/ [sém-ánʤ-è] ‘slaughter’-imperf.-2.pl. d. /sem-anʤ-ɛɛ/ [sɛ ́ m-ánʤ-ɛ ́ ɛ ̀ ] ‘slaughter’-imperf.-3.pl. High and low vowels are in other words phonologically inactive and invisible ▶ non-targets and non-triggers (transparent segments) Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 22 / 46

Transparent segments (e.g. /i, u, a/) co-occur with non-RTR /e, o/ vowels e.g. /bèl-áà/ [bèl-áà], *[bɛ ̀ l-áà] (9) (9) Low/high vowel non-triggers [bèl-áà] *[bɛ ̀ l-áà] ‘edible leaves (cooked)’-sg. [òb-áà] *[ɔ ̀ b-áà] ‘�exible liana branch’-sg. [ʤóŋ-ónʤ-ójì] ‘heal’-imperf.-1.pl. [sém-ánʤ-ójì] *[sɛ ́ m-ánʤ-ójì] ‘slaughter’-imperf.-1.pl. ☞ The marked value is [RTR] in Bondu-so i.e. [RTR] /ɛ, ɔ/ vs. (non-RTR) /e, o/ Bondu-so harmony is active [RTR]-spreading Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 23 / 46

e.g. /bèl-áà/ [bèl-áà], *[bɛ ̀ l-áà] (9) ☞ The marked value is [RTR] in Bondu-so i.e. [RTR] /ɛ, ɔ/ vs. (non-RTR) /e, o/ Bondu-so harmony is active [RTR]-spreading Transparent segments (e.g. /i, u, a/) co-occur with non-RTR /e, o/ vowels (9) Low/high vowel non-triggers [bèl-áà] *[bɛ ̀ l-áà] ‘edible leaves (cooked)’-sg. [òb-áà] *[ɔ ̀ b-áà] ‘�exible liana branch’-sg. [ʤóŋ-ónʤ-ójì] ‘heal’-imperf.-1.pl. [sém-ánʤ-ójì] *[sɛ ́ m-ánʤ-ójì] ‘slaughter’-imperf.-1.pl. Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 23 / 46

☞ The marked value is [RTR] in Bondu-so i.e. [RTR] /ɛ, ɔ/ vs. (non-RTR) /e, o/ Bondu-so harmony is active [RTR]-spreading Transparent segments (e.g. /i, u, a/) co-occur with non-RTR /e, o/ vowels ▶ e.g. /bèl-áà/ → [bèl-áà], *[bɛ ̀ l-áà] (9) (9) Low/high vowel non-triggers [bèl-áà] *[bɛ ̀ l-áà] ‘edible leaves (cooked)’-sg. [òb-áà] *[ɔ ̀ b-áà] ‘�exible liana branch’-sg. [ʤóŋ-ónʤ-ójì] ‘heal’-imperf.-1.pl. [sém-ánʤ-ójì] *[sɛ ́ m-ánʤ-ójì] ‘slaughter’-imperf.-1.pl. Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 23 / 46

Bondu-so harmony is active [RTR]-spreading Transparent segments (e.g. /i, u, a/) co-occur with non-RTR /e, o/ vowels ▶ e.g. /bèl-áà/ → [bèl-áà], *[bɛ ̀ l-áà] (9) (9) Low/high vowel non-triggers [bèl-áà] *[bɛ ̀ l-áà] ‘edible leaves (cooked)’-sg. [òb-áà] *[ɔ ̀ b-áà] ‘�exible liana branch’-sg. [ʤóŋ-ónʤ-ójì] ‘heal’-imperf.-1.pl. [sém-ánʤ-ójì] *[sɛ ́ m-ánʤ-ójì] ‘slaughter’-imperf.-1.pl. ☞ The marked value is [RTR] in Bondu-so ▶ i.e. [RTR] /ɛ, ɔ/ vs. (non-RTR) /e, o/ Jade J. Sandstedt (Humboldt University of Berlin) Bondu-so harmony and abstract contrasts 9. May 2�19 23 / 46