LING 333 W4: Psycholinguistic

how do offline methods help?

• measure the final outcome of conscious thought

• no temporal access to millisecond-by-milisceond parsing

what are the offline methods formats

questionnaire, acceptability rates, end-of-sentence comprehension probes

offline methods = worried about final product

what are online methods for?

• tap automatic, unconscious parsing operations in real time

• highly time-senstivie protocols

what are the 2 types of formats online methods collect?

behavioural and neuroimaging

what are the behavioural formats for online methods

reaction times (LDT), reading dwell times (SPR, Eye-tracking)

what are the neuroimaging formats for online methods

ERPs, EEG, fMRI

What are the 3 processes of Garden-pathing?

1. Initial Parse

2. processing breakdown

3. reanalysis (reduced relative clause)

Garden-pathing

incremental structural misinterpretation followed by effortful reanalysis

• think of something but you change direction for analyses

Lexical decision taks (LDT)

• participants see a letter of strings/word on screen and judge whether its a real word

• stimuli can be purely visual or cross-modal (audio + text)

What is LDT measuring

• responses recorded via keyboard or dedicated millisecond response box (how long u take to answer)

What are the core dependent variables of LDT

Accuracy and reaction time

accuracy (LDT)

proportion of correct decisions

reaction time (LDT)

latency from stimulus onset to button press

A

(might be bc their network is repeated so often so its faster realisation)

priming lexical

will affect your rxn time

whats the theoretical takeaway of semantic priming effect

encountering a related prime spreads activation across semantically associated nodes, lowering the retrieval threshold for target recognition

what does the self-paced reading (SPR) use

moving window paradigm

moving window paradigm

• sentence appears masked (with —-)

• pressing a space bar or button reveals one word/region at a time while re-masking the previous one

what does the SPR measure

measures dwell time per individual word/region

• stimulates progressive reading while yielding millisecond-accurate fixation estimates

spillover effect

processing difficulty at a critical word frequency shows up 1-2 words later, bc button-press motor actions lag cognitive reanalysis

Evaluation of SPR PROS

low cost, highly portable, easy to script, sensitive to processing load

evaluation of SPR CONS

lower ecological validity (readers can’t make regressive saccades naturally)

PsyToolkit

• lightweight, browser-based execution

• simple scripting languages for rapid experiment design

• ideal for running remote online participant pools

• completely free and open-access

PsychoPy

• Flexible Python library + Builder GUI

• Sub-millisecond timing precision on local laboratory displays

• integrates seamlessly with eye-trackers and EEG triggers

• supports Pavlovia for browser hosting

eye-tracking in reading

infrared video cameras can track corneal reflections and pupil position

what are the key measures in reading

1. fixations

2. saccades

3. regressions

fixations

pauses on specific words

saccades

rapid jumping b/w fixation points

regressions

backward saccades indicative of parsing breakdown/reanalysis

visual world paradigm

you can measure so many things in 1 trial compared to other

what are the types of results you can get from visual world paradigm?

one-referent context

two-referent context

one-referent context

listeners look to the empty to the empty towel before hearing “in the box”

two-referent context

listeners immediately interpret “on the towel” as a modifier identifying which apple to move

N100

negative peaking post-stimulus onset

N100/P200

phonetic oddball detection

ex: bike, baby, balloon, pike

indexes early sensory detection of acoustic/phonetic deviance

N400

semantic anomaly

• semantic integration difficulty

ex: whipped cream tastes anxious

P600

syntactic anomaly

• structural reanalysis/repair

ex: the spoiled child throw the toy

B

LDT

LDT

SPR

SPR

Eye-tracking

ET

ERPs

ERPs

columns (RStudio)

used ti reference and manipulate data

• numbers, text, dates or logical values

rows (RStudio)

individual observations, cases or records

• w/ a unique index or identifier, typically starting from 1

dataframes (RStudio)

different data types within the same structure, making them versatile for handling diverse datasets

dimensions (RStudio)

numbers of rows and columns

empty dimensions (RStudio)

data.frame()

• function w/ no arguments

  • initialising a data frame before adding data

initialisation (RStudio)

u specify a variable and set its initial value

• variable used as a counter during the loop

iteration (RStudio)

after each iteration of the loop, the counter variable is updated according to a specified increment or decrement

how to display # of rows

head(file_name, #)

when there’s a Freq of 0

means that the word doesn’t exist in the corpus; (typically add 1)

to bring dataset into RStudio

read_csv(file_name)

once you give something a vector/name, you always have to run it again to double-check

how to specific which column

with a $

ex: length(ldf_lt$Length)

how to get specific elements

[#:#]

what to do if you want to change a value

ex: ldt_data$Freq[1] ← 9

OG csv doesn’t change, just in R

why do we use i

because it will help us code/measure from the same row (which R doesn’t normally do; it does it as a list)

logical → integer → numeric → character