Copyright © 2003 by K. Stromswold

 

 

Lab 10: Auditory Sentence Processing

 

 

Background

 

[For this lab, you will be a subject in a PC-based experiment of auditory sentence processing.  The experiment is a gating experiment that runs under the Windows program E-prime.  This handout describes the nature of the experiment and the questions you need to answer for the lab.  Your TA will give you separate instructions on how to start E-prime, run the experiment, and save your data.]

 

In English, some sentences are temporarily structurally ambiguous.  For example, the active sentence given in (a) and the passive sentence given in (b) are syntactically ambiguous until one has read the verb inflection (the inflection -ing in the active and the inflection –ed in the passive)

 

a.  The girl was pushing the boy.

 

b.  The girl was pushed by the boy.

 

In this experiment, we explore whether adults can predict whether a sentence is active or passive prior to hearing the verbal inflection (i.e., prior to the point of syntactic disambiguation).  In the experiment, you will listen to sentence fragments that are either taken from an active sentence or a passive sentence.  Sentences are either truncated after the was (as shown in example c) or after the verb stem (as shown in example d).  Your job is to guess how the sentence ends.  In other words, you must decide whether the sentence fragment was taken from an active sentence or a passive sentence.

 

c.  Was termination:  The girl was ….

d.  Verb stem termination:  The girl was push ….

 

The goals of this lab are 1) to investigate whether non-syntactic factors affect sentence processing, 2) to understand how psycholinguistic experiments are designed such that one can systematically test hypotheses and 3) to understand some key statistical concepts such as null hypothesis, main effect, interaction, and significance level

 

 

Design

 

One common way of designing experiments is to have a factorial design in each factor is orthogonal to the others.   In this experiment , there are 96 trials. The two main factors are (1) passive/active and (2) terminates after aux/terminates after verb stem.  These two factors are crossed (orthogonal), yielding distinct sentence types.  A third factor is verb.  Each of the 6 verbs appear in each sentence condition an equal number of times. There are two sets of noun phrases (NPs) that are always paired with each another.  Specifically, the man is always paired with the woman and the boy is always paired with the girl.  Each NP appears as the subject (i.e., first NP) and object (i.e., second NP) of each sentence type an equal number of times. Each NP appears as the subject (the first NP) with each verb in each sentence condition an equal number of times.

 

Based on the above description, answer the following questions:

 

1.  How many sentence types are there?

 

2.  How many examples of each sentence type do subjects do?

 

3:  How many times do the subjects hear a given verb in a sentence type

 

4:  How many times does man appear as the subject (first NP) of the following sentence type "The man was VERB(ing the woman).

 

5:  How many times does man appear as the subject (first NP) in the following sentence:  "The man was PUSH(ing the woman).

 

Analysis

 

Using the Analysis of Variance (ANOVA) table given at the end of this handout, answer the following questions

 

6a.   How many subjects do we have data from?

 

6b.  What 3 factors are analyzed in this ANOVA?

 

6c.  How many levels are there for each of the 3 factors?

 

 

 

7a.  On average, what percentage of the time did these subjects guess correctly for actives?

 

7b.  On average, what percentage of the time did they guess correctly for passives?

 

7c.  What is the null hypothesis with respect to the factor active/passive?

 

7d.  How likely was the difference in performance on actives and passives due to chance along?  (Hint:  look for the p value)

 

7e.  How confident can we feel in rejecting the null hypothesis?

 

 

 

8a.  On average, what percentage of the time did these subjects guess correctly for sentences that were truncated after the auxiliary?

 

8b.  On average, what percentage of the time did these subjects guess correctly for sentences that were truncated after the verb stem?

 

8c.  What is the null hypothesis with respect to the point at which the sentences were truncated?

 

8d.  How likely was the difference on was truncations and verb stem truncations due to chance alone?

 

8e.  How confident can we feel in rejecting the null hypothesis?

 

 

 

9a.  On average, which verb(s) did subjects do the worst on?_______ What percentage of the time did they guess correctly for this/these verb(s)?

 

9b.  On average, which verb(s) did subjects do the best on?_______ What percentage of the time did they guess correctly for this/these verb(s)?

 

9c.  What is the null hypothesis with respect to the verbs used in the sentences?

 

9d.  How likely was the difference among the verbs due to chance alone?

 

9e.  How confident can we feel in rejecting the null hypothesis?

 

 

 

10.  There was a significant interaction between the choice of verb and performance on active versus. passive sentences (F(5, 105) = 3.465, p = .006).  Your TA will explain what is meant by an interaction, but intuitively, in this case, it means that the particular verb used in a sentence affected the pattern of response for actives and passives.  Using the ANOVA table, can you find any verbs that displayed a different pattern from the other verbs.   (HINT:  look to see if the difference in percentage correct for actives and passives was approximately same for all 6 verbs or if the difference was a lot greater or lesser for some verb than others.)

 

 

 

 

 

 

11.  The same sets pictures were used for push and shove and for touch and tickle.  Given this, the difference in performance between push and shove and between touch and tickle could not be due to the pictures per se. 

 

 

What might account for the difference between push and shove? 

 

 

 

 

What might account for the difference between touch and tickle?

 

 

12.  Are any of the other interactions among the 3 factors significant? 

 

 

 

 

 


ANOVA TABLE

 

SOURCE: grand mean

verb    psvact  auxstem    N       MEAN         SD         SE

                         528     0.6608     0.3302     0.0144

 

SOURCE: verb

verb    psvact  auxstem    N       MEAN         SD         SE

tickle                    88     0.6250     0.3515     0.0375

touch                     88     0.6903     0.3216     0.0343

push                      88     0.6712     0.3498     0.0373

sniff                     88     0.6903     0.2984     0.0318

kiss                      88     0.6659     0.3228     0.0344

shove                     88     0.6222     0.3362     0.0358

 

SOURCE: psvact

verb    psvact  auxstem    N       MEAN         SD         SE

        Act              264     0.7396     0.3012     0.0185

        Pass             264     0.5820     0.3395     0.0209

 

SOURCE: verb psvact

verb    psvact  auxstem    N       MEAN         SD         SE

tickle  Act               44     0.7102     0.3321     0.0501

tickle  Pass              44     0.5398     0.3533     0.0533

touch   Act               44     0.7159     0.2727     0.0411

touch   Pass              44     0.6648     0.3655     0.0551

push    Act               44     0.6799     0.3752     0.0566

push    Pass              44     0.6625     0.3266     0.0492

sniff   Act               44     0.7955     0.2487     0.0375

sniff   Pass              44     0.5852     0.3095     0.0467

kiss    Act               44     0.7920     0.2715     0.0409

kiss    Pass              44     0.5398     0.3232     0.0487

shove   Act               44     0.7443     0.2878     0.0434

shove   Pass              44     0.5000     0.3396     0.0512

 

SOURCE: auxstem

verb    psvact  auxstem    N       MEAN         SD         SE

                Stem     264     0.8330     0.2595     0.0160

                Aux      264     0.4886     0.3027     0.0186

 

SOURCE: verb auxstem

verb    psvact  auxstem    N       MEAN         SD         SE

tickle          Stem      44     0.7727     0.3135     0.0473

tickle          Aux       44     0.4773     0.3272     0.0493

touch           Stem      44     0.8977     0.1732     0.0261

touch           Aux       44     0.4830     0.3021     0.0455

push            Stem      44     0.8652     0.2343     0.0353

push            Aux       44     0.4773     0.3402     0.0513

sniff           Stem      44     0.8523     0.2244     0.0338

sniff           Aux       44     0.5284     0.2761     0.0416

kiss            Stem      44     0.8261     0.2693     0.0406

kiss            Aux       44     0.5057     0.2928     0.0441

shove           Stem      44     0.7841     0.3062     0.0462

shove           Aux       44     0.4602     0.2850     0.0430

 

SOURCE: psvact auxstem

verb    psvact  auxstem    N       MEAN         SD         SE

        Act     Stem     132     0.8922     0.2202     0.0192

        Act     Aux      132     0.5871     0.2946     0.0256

        Pass    Stem     132     0.7739     0.2822     0.0246

        Pass    Aux      132     0.3902     0.2785     0.0242

 

SOURCE: verb psvact auxstem

verb    psvact  auxstem    N       MEAN         SD         SE

tickle  Act     Stem      22     0.8182     0.3379     0.0720

tickle  Act     Aux       22     0.6023     0.2954     0.0630

tickle  Pass    Stem      22     0.7273     0.2877     0.0613

tickle  Pass    Aux       22     0.3523     0.3149     0.0671

touch   Act     Stem      22     0.8864     0.1490     0.0318

touch   Act     Aux       22     0.5455     0.2632     0.0561

touch   Pass    Stem      22     0.9091     0.1974     0.0421

touch   Pass    Aux       22     0.4205     0.3308     0.0705

push    Act     Stem      22     0.8485     0.2668     0.0569

push    Act     Aux       22     0.5114     0.3970     0.0846

push    Pass    Stem      22     0.8818     0.2015     0.0430

push    Pass    Aux       22     0.4432     0.2774     0.0592

sniff   Act     Stem      22     0.9318     0.1376     0.0293

sniff   Act     Aux       22     0.6591     0.2622     0.0559

sniff   Pass    Stem      22     0.7727     0.2662     0.0568

sniff   Pass    Aux       22     0.3977     0.2270     0.0484

kiss    Act     Stem      22     0.9364     0.1787     0.0381

kiss    Act     Aux       22     0.6477     0.2745     0.0585

kiss    Pass    Stem      22     0.7159     0.3017     0.0643

kiss    Pass    Aux       22     0.3636     0.2406     0.0513

shove   Act     Stem      22     0.9318     0.1756     0.0374

shove   Act     Aux       22     0.5568     0.2551     0.0544

shove   Pass    Stem      22     0.6364     0.3398     0.0724

shove   Pass    Aux       22     0.3636     0.2858     0.0609

 

FACTOR  :       subj       verb     psvact    auxstem    correct

LEVELS  :         22          6          2          2        528

TYPE    :     RANDOM     WITHIN     WITHIN     WITHIN       DATA

 

SOURCE                SS     df             MS         F      p

===============================================================

mean            230.5736      1       230.5736  2080.267  0.000 ***

s/                2.3276     21         0.1108

 

verb              0.4096      5         0.0819     1.838  0.112

vs/               4.6807    105         0.0446

 

psvact            3.2802      1         3.2802     7.081  0.015 *

ps/               9.7283     21         0.4633

 

vp                1.1088      5         0.2218     3.465  0.006 **

vps/              6.7200    105         0.0640

 

auxstem          15.6550      1        15.6550   196.395  0.000 ***

as/               1.6739     21         0.0797

 

va                0.2344      5         0.0469     1.265  0.284

vas/              3.8906    105         0.0371

 

pa                0.2042      1         0.2042     1.886  0.184

pas/              2.2737     21         0.1083

 

vpa               0.2490      5         0.0498     1.043  0.396

vpas/             5.0145    105         0.0478