Schlechtweg 2024 Slides

Perception of spoken language in noise
The role of paradigmatic and syntagmatic

information
Marcel Schlechtweg
Carl von Ossietzky Universität Oldenburg
Spoken language in noise
Factors affecting the perception of spoken

language in noise:
(1) Characteristics of the noise

➢ Signal-to-noise ratio (SNR)
➢ Type of noise
➢ Background babble: Number of talkers;
language of the babble
e.g., Bent et al. 2010; Garcia Lecumberri & Cooke 2006; Liu & Kewley-Port 2004; Van Engen & Bradlow 2007
Background Slide 1/26

(2) Characteristics of the listener

➢ Native versus non-native
➢ Monolingual versus bilingual
➢ Children versus adults
➢ Individuals with(out) hearing loss
e.g., Bent & Atagi 2015; Garcia Lecumberri & Cooke 2006; Schmidtke 2016; Smiljanic & Sladen 2013

(3) Characteristics of the linguistic materials

➢ Unfamiliar native accent
➢ Non-native accent
➢ Code-switching
➢ (Semantic) context
➢ Frequency of occurrence
➢ Phonetics/phonology of vowels/consonants
➢ Syntactic structure
e.g., Adank et al. 2009; Bent & Atagi 2015; Carroll & Ruigendijk 2013; Cutler et al. 2004; Gross et al. 2021; Smiljanic &
Sladen 2013; Schmidtke 2016

➢ Study 1: Morphosyntactic agreement
(syntagmatic aspects only)
➢ Study 2: Interplay of paradigmatic and

syntagmatic aspects


➢ Study 1: Morphosyntactic agreement
(syntagmatic aspects only)
Schlechtweg, Marcel. 2024. Morphosyntactic agreement in English:

Does it help the listener in noise? English Language and
Linguistics. FirstView.
Background
These cabs broke down.

➢ Determiner-noun agreement
➢ Target (these) precedes controller (cabs)
➢ Target and controller = same phrase
The cabs break down.

➢ Noun-verb agreement -> verb with no "s" because noun in plural
➢ Target (break) follows controller (cabs)
➢ Target and controller ≠ same phrase
The cabs broke down.

➢ No agreement -> verb can be used for singular or plural
Study 1 Slide 5/26

Background
➢ Variation with regard to …
➢ … the presence of agreement

➢ … the type of agreement
➢ … linear precedence
➢ … syntactic structure
Does this variation affect perception in noise?
Does agreement / which type(s) of agreement

support the listener?
Levin 2001; for discussion, see Corbett 2006
Study 1 Slide 6/26

Background
➢ Presence of agreement affects duration of

speech sounds in production
➢ The blue cabs …
➢ These blue cabs …
➢ Higher redundancy = more sensitivity to

agreement violations (written language)
➢ *The cookies tastes …
➢ *Many cookies tastes …
Schlechtweg & Corbett 2023; Tanner & Bulkes 2015
Study 1 Slide 7/26

Background
➢ Redundancy helps in noise when predictability is low

➢ Choguita Rarámuri (Tarahumara), a language of the
Uto-Aztecan language family spoken in northern
Mexico
➢ Multiple marking (agreement) was not beneficial for the

listener
➢ Batsbi, an endangered language of the Nakh-
Dagestanian language family spoken in the Republic
of Georgia
Caballero & Kapatsinski 2015; Harris & Samuel 2011
Study 1 Slide 8/26

Methodology
➢ 48 native speakers of US American English with
normal hearing and no other native language (mean
age: 29.5 years)
➢ 13 test sentences with 8 versions
Study 1 Slide 9/26

Methodology
➢ Forced-choice study (online, Eprime)
➢ Neatly controlled for many potentially

confounding variables
➢ Recordings
➢ Sound-proof booth
➢ Large-diaphragm condenser microphone
➢ Praat
➢ Normalization = 60 dB
➢ Female native speaker of US American English
Boersma & Weenink 2023; Psychology Software Tools 2016
Study 1 Slide 10/26

Methodology
➢ Sentences manipulated using the variables …

➢ Determiner (definite (the), demonstrative
(this/these))
➢ VerbTense (present, past)
➢ Number (singular, plural)
➢ SNR (quiet, 0, -6, -12 dB) (white noise)
➢ All participants tested on all sentences in all

conditions (2 determiners x 2 verb tenses x 2
numbers x 4 SNRs x 13 sentences = 416 trials)
Study 1 Slide 11/26

Methodology
+
(500 ms)
The cabs break down.
cab cabs
= =
Left arrow Right arrow
Study 1 Slide 12/26

Methodology
➢ Binomial logistic regression and Tukey tests

(R, lme4, lmerTest, emmeans packages)
➢ Response variable: Accuracy
➢ Fixed effects: Determiner, VerbTense, Number, SNR,

two-way interactions
➢ Random intercepts by Participant and Item
➢ Analysis I: Only main effects
➢ Analysis II: With interactions

Bates et al. 2015; Kuznetsova et al. 2017; Lenth 2020; R Core Team 2021
Study 1 Slide 13/26

Selected results Analysis I
Study 1 Slide 14/26

Selected results Analysis II
Study 1 Slide 15/26

Discussion
➢ Determiner-noun, but not noun-verb,

agreement supports the listener
➢ Linear precedence and syntactic structure are

possible candidates to explain these findings
➢ Determiner precedes the noun, verb follows
➢ Determiner in the same nominal phrase, verb

more distant syntactically
Study 1 Slide 16/26


➢ Study 2: Interplay of paradigmatic and
syntagmatic information
Schlechtweg, Marcel. 2024. The interaction of paradigmatic and

syntagmatic properties during the perception of spoken English in
noise. Submitted.
Background
➢ Study I = Syntagmatic aspects (Focus on

sentential context)
➢ But what about...

➢ … paradigmatic aspects (Word-internal
aspects)?
➢ … the interaction between paradigmatic
and syntagmatic aspects?
Study 2 Slide 17/26

Methodology
➢ 43 native speakers of US American English with normal
hearing and no other native language (mean age: 30.6
years)
➢ 6 English nouns with different singular-plural variation

➢ Syncretism: sheep
➢ Vowel change I (vowel fronting only): goose/geese
➢ Vowel change II (vowel fronting and lengthening):
foot/feet
➢ Suffixation I (maintaining the number of syllables):
dog(s)
➢ Suffixation II (changing the number of syllables):
fox(es)
➢ Suppletion: child(ren)
Study 2 Slide 18/26

Methodology
Study 2 Slide 19/26

Methodology
➢ Forced-choice study (online, Eprime)
➢ Neatly controlled for many potentially

confounding variables
➢ Recordings
➢ Sound-proof booth
➢ Large-diaphragm condenser microphone
➢ Praat
➢ Normalization = 60 dB
➢ Female native speaker of US American English
Boersma & Weenink 2023; Psychology Software Tools 2016
Study 2 Slide 20/26

Methodology
➢ Sentences manipulated using the variables …

➢ PluralType (sheep, goose/geese, foot/feet,
dog(s), fox(es), child(ren))
➢ Determiner (definite (the), demonstrative
(this/these))
➢ VerbTense (present, past)
➢ Number (singular, plural)
➢ SNR (quiet, -6, -12 dB) (white noise)
➢ All participants tested on all sentences in all

conditions (690 trials)
Study 2 Slide 21/26

Methodology
+
(500 ms)
The dogs freeze.
dog dogs
= =
Left arrow Right arrow
Study 2 Slide 22/26

Methodology
➢ Binomial logistic regression and Tukey tests

(R, lme4, lmerTest, emmeans packages)
➢ Response variable: Accuracy
➢ Fixed effects: PluralType, Determiner, VerbTense,

Number, SNR, PluralType*Determiner,
PluralType*VerbTense, PluralType*SNR
➢ Random intercept by Participant
➢ Analysis I: Only main effects
➢ Analysis II: With interactions

Bates et al. 2015; Kuznetsova et al. 2017; Lenth 2020; R Core Team 2021
Study 2 Slide 23/26

Selected results Analysis I
Study 2 Slide 24/26

Selected results Analysis II
Study 2 Slide 25/26

Discussion
➢ Paradigmatic and syntagmatic aspects

interact in significant ways to support the
listener in noise and compensate the
negative effect of the listening condition
Study 2 Slide 26/26

References
➢ For the full references, see
Schlechtweg, Marcel. 2024. Morphosyntactic

agreement in English: Does it help the listener
in noise? English Language and Linguistics.
FirstView.

Schlechtweg 2024 Slides

Uploaded by

Copyright:

Available Formats

Schlechtweg 2024 Slides

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Schlechtweg 2024 Slides

Uploaded by

Copyright:

Available Formats

Perception of spoken language in noise

The role of paradigmatic and syntagmatic

Factors affecting the perception of spoken

(1) Characteristics of the noise

Background Slide 1/26

(2) Characteristics of the listener

Background Slide 2/26

(3) Characteristics of the linguistic materials

(3) Characteristics of the linguistic materials

➢ Study 2: Interplay of paradigmatic and

Background Slide 4/26

(3) Characteristics of the linguistic materials

Schlechtweg, Marcel. 2024. Morphosyntactic agreement in English:

These cabs broke down.

The cabs break down.

The cabs broke down.

Study 1 Slide 5/26

➢ Variation with regard to …

➢ … the presence of agreement

Does this variation affect perception in noise?

Does agreement / which type(s) of agreement

Levin 2001; for discussion, see Corbett 2006

Study 1 Slide 6/26

➢ Presence of agreement affects duration of

➢ Higher redundancy = more sensitivity to

Schlechtweg & Corbett 2023; Tanner & Bulkes 2015

Study 1 Slide 7/26

➢ Redundancy helps in noise when predictability is low

➢ Multiple marking (agreement) was not beneficial for the

Caballero & Kapatsinski 2015; Harris & Samuel 2011

Study 1 Slide 8/26

➢ 13 test sentences with 8 versions

Study 1 Slide 9/26

➢ Forced-choice study (online, Eprime)

➢ Neatly controlled for many potentially

Boersma & Weenink 2023; Psychology Software Tools 2016

Study 1 Slide 10/26

➢ Sentences manipulated using the variables …

➢ All participants tested on all sentences in all

Study 1 Slide 11/26

The cabs break down.

Study 1 Slide 12/26

➢ Binomial logistic regression and Tukey tests

➢ Response variable: Accuracy

➢ Fixed effects: Determiner, VerbTense, Number, SNR,

➢ Random intercepts by Participant and Item

➢ Analysis I: Only main effects

➢ Analysis II: With interactions

Study 1 Slide 13/26

Study 1 Slide 14/26

Study 1 Slide 15/26

➢ Determiner-noun, but not noun-verb,

➢ Linear precedence and syntactic structure are

➢ Determiner precedes the noun, verb follows

➢ Determiner in the same nominal phrase, verb

Study 1 Slide 16/26

(3) Characteristics of the linguistic materials

Schlechtweg, Marcel. 2024. The interaction of paradigmatic and

➢ Study I = Syntagmatic aspects (Focus on