Word Duration Research Articles

Evaluating the effects of continuous pitch and speech tempo modifications on perceptual speaker verification performance by familiar and unfamiliar listeners

A study was conducted to evaluate the effects of continuous pitch and speech tempo modifications on perceptual speaker verification performance by familiar and unfamiliar naive listeners. Speech recordings made by twelve male, native-French speakers were organised into three groups of four (two in-set, one out-of-set). Two groups of listeners participated, where one group was familiar with one in-set speaker group, while both groups were unfamiliar with the remaining in- and out-of-set speaker groups. Pitch and speech tempo were continuously modified, such that the first 75% of words spoken were modified with percentages of modification beginning at 100% and decaying linearly to 0%. Pitch modifications began at ± 600 cents, while speech tempo modifications started with word durations scaled 1:2 or 3:2. Participants evaluated a series of “go/no-go” task trials, where they were presented a modified speech recording with a face and tasked to respond as quickly as possible if they judged the stimuli to be continuous. The major findings revealed listeners overcame higher percentages of modification when presented familiar speaker stimuli. Familiar listeners outperformed unfamiliar listeners when evaluating continuously modified speech tempo stimuli, however, this effect was speaker-specific for pitch modified stimuli. Contrasting effects of modification direction were also observed. The findings suggest pitch is more useful to listeners when verifying familiar and unfamiliar voices.

Lexical Surprisal Shapes the Time Course of Syntactic Structure Building.

When we understand language, we recognize words and combine them into sentences. In this article, we explore the hypothesis that listeners use probabilistic information about words to build syntactic structure. Recent work has shown that lexical probability and syntactic structure both modulate the delta-band (<4 Hz) neural signal. Here, we investigated whether the neural encoding of syntactic structure changes as a function of the distributional properties of a word. To this end, we analyzed MEG data of 24 native speakers of Dutch who listened to three fairytales with a total duration of 49 min. Using temporal response functions and a cumulative model-comparison approach, we evaluated the contributions of syntactic and distributional features to the variance in the delta-band neural signal. This revealed that lexical surprisal values (a distributional feature), as well as bottom-up node counts (a syntactic feature) positively contributed to the model of the delta-band neural signal. Subsequently, we compared responses to the syntactic feature between words with high- and low-surprisal values. This revealed a delay in the response to the syntactic feature as a consequence of the surprisal value of the word: high-surprisal values were associated with a delayed response to the syntactic feature by 150-190 ms. The delay was not affected by word duration, and did not have a lexical origin. These findings suggest that the brain uses probabilistic information to infer syntactic structure, and highlight an importance for the role of time in this process.

Facial expressions in different communication settings: A case of whispering and speaking with a face mask in Farsi

Abstract This study addresses the importance of orofacial gestures and acoustic cues to execute prosodic patterns under different communicative settings in Farsi. Given that Farsi lacks morpho-syntactic markers for polar questions, we aim to determine whether specific facial movements accompany the prosodic correlates of questionhood in Farsi under conditions of degraded information, that is, whispering and wearing face masks. We hypothesise speakers will employ the most pronounced facial expressions when whispering questions with a face mask to compensate for the absence of F0, reduced intensity and lower face invisibility. To this end, we conducted an experiment with 10 Persian speakers producing 10 pairs of statements and questions in normal and whispered speech modes with and without face masks. Our results provide support to our hypotheses that speakers will intensify their orofacial expressions when confronted with marked conditions. We interpreted our results in terms of the ‘hand in hand’ and ‘trade-off’ hypothesis. In whispered speech, the parallel realisation of longer word duration and orofacial expressions may be a compensatory mechanism for the limited options to convey intonation. Also, the lower face coverage is mutually compensated for by word duration and intensified upper facial expressions, all of which in turn support the trade-off hypothesis.

Rima Aliterasi dan Asonansi sebagai Sumber Ritma Linguistik Surah Fatihah yang Mensinkroni Gelombang Otak

Sound rhythm is one of the factors that make the sound of the Holy Quran rhythmic and synchronous that change the rhythm of human brain waves. However, less studies have been found, and previous studies have focused more on the end rhyme system (end faselah) of the verses of the Qur'an. Thus, we conducted an acoustic phonetic study on the rhyming system in the verses of the Qur'an which is the source of the rhythmic sound in the verses of the Qur'an which contributes to the aesthetics of the sound. Acoustic phonetic analysis was performed on sound waves from Surah al-Fatihah using PRAAT software by comparing its intensity with Arabic News. As a result, we found the existence of a rhyme system of alliteration and assonance in the verses of this surah, that is, in the first, third, fifth and seventh verses which are highlighted by the contours of intensity and time duration of letters, syllables and words that show the greatness of God, giving birth to a feeling of closeness to God in addition to giving birth to the effect of persuasion which is the basis of spiritual therapy. In addition, the end rhyme system found in the Fatihah Chapter is significant (p&lt;0.05) compared to the Arabic News. In conclusion, the Fatihah Chapter also has a rhyme system in its verses, not only at the end of the verse, suggesting that the brain that listens to the sound of the Qur'an is able to synchronize and form rhythmic and harmonious waves that influence mental and physical well-being.

Kesan Modul Terapi Realiti (MTR) Terhadap Ciri-ciri dan Kemurungan Mangsa Buli dari Perspektif Islam

Sound rhythm is one of the factors that make the sound of the Holy Quran rhythmic and synchronous that change the rhythm of human brain waves. However, less studies have been found, and previous studies have focused more on the end rhyme system (end faselah) of the verses of the Qur'an. Thus, we conducted an acoustic phonetic study on the rhyming system in the verses of the Qur'an which is the source of the rhythmic sound in the verses of the Qur'an which contributes to the aesthetics of the sound. Acoustic phonetic analysis was performed on sound waves from Surah al-Fatihah using PRAAT software by comparing its intensity with Arabic News. As a result, we found the existence of a rhyme system of alliteration and assonance in the verses of this surah, that is, in the first, third, fifth and seventh verses which are highlighted by the contours of intensity and time duration of letters, syllables and words that show the greatness of God, giving birth to a feeling of closeness to God in addition to giving birth to the effect of persuasion which is the basis of spiritual therapy. In addition, the end rhyme system found in the Fatihah Chapter is significant (p&lt;0.05) compared to the Arabic News. In conclusion, the Fatihah Chapter also has a rhyme system in its verses, not only at the end of the verse, suggesting that the brain that listens to the sound of the Qur'an is able to synchronize and form rhythmic and harmonious waves that influence mental and physical well-being.

Communicative efficiency in multimodal language directed at children and adults.

The ecology of human communication is face to face. In these contexts, speakers dynamically modify their communication across vocal (e.g., speaking rate) and gestural (e.g., cospeech gestures related in meaning to the content of speech) channels while speaking. What is the function of these adjustments? Here we ask whether speakers dynamically make these adjustments to increase communicative success, and decrease cognitive effort while speaking. We assess whether speakers modulate word durations and produce iconic (i.e., imagistically evoking properties of referents) gestures depending on the predictability of each word they utter. Predictability is operationalized as surprisal and computed from computational language models trained on corpora of child-directed, or adult-directed language. Using data from a novel corpus (Ecological Language Corpus) of naturalistic interactions between adult-child (aged 3-4), and adult-adult, we show that surprisal predicts speakers' multimodal adjustments and that some of these effects are modulated by whether the comprehender is a child or an adult. Thus, communicative efficiency applies generally across vocal and gestural communicative channels not being limited to structural properties of language or vocal modality. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Constructional sources of durational shortening in discourse markers

Abstract Using naturally occurring data from Spanish from Madrid, this study is the first to analyze durations of the Spanish word decir ‘to say, to tell’ both as a verb with prepositional meaning and as part of the reformulating construction [es decir] ‘that is to say’ (N = 388). We show that, although it is neither highly grammaticalized nor frequent, [es decir] undergoes phonological reduction to a significantly greater degree than the more frequent lexical source decir ‘to say’. Results of linear mixed-effects models predicting target duration suggest these durational differences cannot be explained due to conditioning factors of the target context controlled in this analysis (speech rate of the target word context, predictability of following words, number of phones, distance from pause). They do not appear to stem from an accumulation in memory of patterns of likelihood of use in those conditioning environments. We propose that [es decir] is stored as a lexical unit that contains as part of the lexical representation shorter word durations relative to the lexical form decir and that this durational shortening is part of the [reformulator] construction.

Using prosody to produce trust and doubt

Work presented at a 2022 ASA meeting [Thomas, 2022] demonstrated that American English talkers use acoustic prosodic variables to distinguish attitudes of incredulity and trust from a “neutral tone of voice” when asking WH-questions. The present study largely replicates these findings in new data. These data comprise both WH-questions and Yes/No questions, read by a new set of talkers in six imagined scenarios. In the previous study, “doubt” was likely confounded with “authority” due to the imagined scenario used to elicit the stimuli. In the present study, the three attitudes (doubt, neutral, trust) were crossed with dominance (participant was +/− dominant relative to their imagined conversational partner). This study examines additional acoustic variables, including the variability of word duration within utterance and the F0 slope of the longest word in each utterance. Averaging across the + /− dominance conditions and the utterance type (yes/no or WH question), duration-based variables reliably distinguished doubting from neutral utterances (with doubting utterances showing greater duration and greater variability in word duration). In contrast, F0-based variables differentiated trusting from neutral utterances, with trusting utterances showing a higher F0 over the course of the entire utterance.

Listening effort for L2 accents: Adults’ adaptation to Mandarin accented english

Using task-evoked pupillary response (TEPR) and dual-task procedures, prior research has shown that adults recruit additional cognitive resources when processing fully intelligible L2-accented speech compared to speech in their own L1 accent (McLaughlin and Van Engen, 2020; Brown et al., 2020). In this study, we investigated the relationship between three prosodic measures (relative word duration variance, pitch stability, and pitch range) and changes in listening effort for L1 and L2 speakers over the course of an experiment. Using growth curve analysis, we found significant interactions between each of the three measures and trial number, indicating that prosodic factors influence speech adaptation. However, these interactions were in opposite directions for the L1 and L2 speakers. Thus, adult monolingual English listeners rely on prosodic information to adapt to a particular speaker’s voice, but use that information differently based on speaker identity.

Hebbian learning can explain rhythmic neural entrainment to statistical regularities.

In many domains, learners extract recurring units from continuous sequences. For example, in unknown languages, fluent speech is perceived as a continuous signal. Learners need to extract the underlying words from this continuous signal and then memorize them. One prominent candidate mechanism is statistical learning, whereby learners track how predictive syllables (or other items) are of one another. Syllables within the same word predict each other better than syllables straddling word boundaries. But does statistical learning lead to memories of the underlying words-or just to pairwise associations among syllables? Electrophysiological results provide the strongest evidence for the memory view. Electrophysiological responses can be time-locked to statistical word boundaries (e.g., N400s) and show rhythmic activity with a periodicity of word durations. Here, I reproduce such results with a simple Hebbian network. When exposed to statistically structured syllable sequences (and when the underlying words are not excessively long), the network activation is rhythmic with the periodicity of a word duration and activation maxima on word-final syllables. This is because word-final syllables receive more excitation from earlier syllables with which they are associated than less predictable syllables that occur earlier in words. The network is also sensitive to information whose electrophysiological correlates were used to support the encoding of ordinal positions within words. Hebbian learning can thus explain rhythmic neural activity in statistical learning tasks without any memory representations of words. Learners might thus need to rely on cues beyond statistical associations to learn the words of their nativelanguage. RESEARCH HIGHLIGHTS: Statistical learning may be utilized to identify recurring units in continuous sequences (e.g., words in fluent speech) but may not generate explicit memory forwords. Exposure to statistically structured sequences leads to rhythmic activity with a period of the duration of the underlying units (e.g., words). I show that a memory-less Hebbian network model can reproduce this rhythmic neural activity as well as putative encodings of ordinal positions observed in earlier research. Direct tests are needed to establish whether statistical learning leads to declarative memories for words.

Acoustic characteristics of non-native Lombard speech in the DELNN corpus

Lombard speech, speech produced in noise, has been extensively studied in native speakers, while non-native Lombard speech research is limited. This article presents the first corpus of non-native Lombard speech, the Dutch English Lombard Native Non-Native corpus, which includes plain and Lombard read speech from native American-English, non-native English (native Dutch), and native Dutch women. The location of contrastive focus is systematically varied in the sentences. We investigated how intensity, spectral center of gravity, word duration, and VOT varies in the corpus as a function of plain versus Lombard speech and whether it is modulated by the speaker’s nativeness and of the language. We did not find differences in how the native and non-native English speakers adapted their English speech in noise, indicating that the Dutch non-native speakers produced Lombard speech similarly to the native English. The comparison of the native Dutch and non-native English sentences produced by the same participants nevertheless suggests that, for all acoustic measurements except word duration, the Dutch speakers adapted their Lombard speech differently in native Dutch than in non-native English. Combined, this would indicate that, when speaking English, Dutch speakers adapt their way of speaking in noise to the way native English speakers do.

The Effect of Speaking Rate on the Voice Onset Time (VOT) of Bahdini Kurdish Stops

This research examines the effect of speaking rate on the voice onset time (VOT) of stop consonants in Bahdini Kurdish. 30 native speakers are chosen to take part in this study. VOT values are measured for all 10 Kurdish stop consonants produced by participants in carrier words in initial prevocalic onset position. The duration of the carrier word is also measured by using the Web MAUS service (Munich Automatic Segmentation), which is used for word duration measurements. Word duration measurements are compared to the VOT measurements to check whether they are both correlated or not based on laryngeal state and place of stop articulation (POA). Results reveal that there is an effect of speaking rate on VOT based on the three-way laryngeal contrast between the categories of voiced, voiceless aspirated and voiceless unaspirated stops in Bahdini Kurdish. VOT increases as word durations increase, which is inversely related to speaking rate. This means that VOT increases as speaking rate decreases. There is a systematic correlation between VOT and word duration based on different laryngeal states of voiced, voiceless aspirated and voiceless unaspirated stop categories. However, Place of Articulation (POA) does not show such an effect.

Comparison of category and letter fluency tasks through automated analysis.

Category and letter fluency tasks are commonly used neuropsychological tasks to evaluate lexical retrieval. This study used validated automated methods, which allow for more expansive investigation, to analyze speech production of both category ("Animal") and letter ("F") fluency tasks produced by healthy participants (n = 36) on an online platform. Recordings were transcribed and analyzed through automated pipelines, which utilized natural language processing and automatic acoustic processing tools. Automated pipelines calculated overall performance scores, mean inter-word response time, and word start time; errors were excluded from analysis. Each word was rated for age of acquisition (AoA), ambiguity, concreteness, frequency, familiarity, word length, word duration, and phonetic and semantic distance from its previous word. Participants produced significantly more words on the category fluency task relative to the letter fluency task (p < 0.001), which is in line with previous studies. Wilcoxon tests also showed tasks differed on several mean speech measures of words, and category fluency was associated with lower mean AoA (p<0.001), lower frequency (p < 0.001), lower semantic ambiguity (p < 0.001), lower semantic distance (p < 0.001), lower mean inter-word RT (p = 0.03), higher concreteness (p < 0.001), and higher familiarity (p = 0.02), compared to letter fluency. ANOVAs significant interactions for fluency task on total score and lexical measures showed that lower category fluency scores were significantly related to lower AoA and higher prevalence, and this was not observed for letter fluency scores. Finally, word-characteristics changed over time and significant interactions were noted between the tasks, including word familiarity (p = 0.019), semantic ambiguity (p = 0.002), semantic distance (p=0.001), and word duration (p<0.001). These findings showed that certain lexical measures such as AoA, word familiarity, and semantic ambiguity were important for understanding how these tasks differ. Additionally, it found that acoustic measures such as inter-word RT and word duration are also imperative to analyze when comparing the two tasks. By implementing these automated techniques, which are reproducible and scalable, to analyze fluency tasks we were able to quickly detect these differences. In future clinical settings, we expect these methods to expand our knowledge on speech feature differences that impact not only total scores, but many other speech measures among clinical populations.

Children's listening effort for L2-accented speech

Speech perception is a complex task that can become challenging in certain conditions, such as listening to speech in unfamiliar accents. Using task-evoked pupillary response (TEPR) and dual-task procedures, prior research has shown that adults recruit additional cognitive resources when processing fully intelligible L2-accented speech compared to speech in their own L1 accent (McLaughlin and Van Engen, 2020; Brown et al., 2020). In the current study, we used pupillometry to investigate this effect in children (n = 35, ages 5;0 – 8;11) and ask whether prosodic variation can explain any differences in listening effort. Using Growth Curve Analysis, we found that unlike adults, children do not appear to recruit extra cognitive resources to process intelligible L2-accented speech. Like adults, however, pupillary response declined more rapidly for L2-accented speech across the experiment than for L1-accented speech, indicating either rapid adaptation or fatigue. Additionally, we identified three prosodic measures (relative word duration, pitch stability, and pitch range) that were related to speech adaptation/fatigue for both accent conditions, but in opposite directions. Although children do not appear to upregulate listening effort for L2-accented speech like adults, they appear to use prosodic cues differently for L1- versus L2- accented speech as they adapt to speech.

Blockchain Technology Applied to Supply Chain Management: A Systems’ Analysis

Blockchain technology can be used to record real data at various nodes in the entire supply chain and can achieve timely data sharing and full traceability of the whole lifecycle of products in the supply chain. This study conducts a systematic literature review based on the WOS database, focusing on the research topic of the application of blockchain technology in supply chain management. We propose clear search and screening criteria based on 4 research questions and finally obtain 591 target studies. We conduct a detailed study on the distribution of the target literature in this research area, country/region, institution, and journal. Different countries/regions cooperate with each other and, thus, are clustered into five significant countries regions’ cooperation networks. There is a strong coupling relationship between research institutions, forming four major institutional cooperation networks. According to the analysis of those papers with a high number of citations, most of such papers were published in 2019, while the cross-citation phenomenon between papers occurred more frequently in 2021 and 2022. We use VOSviewer for visual coupling analysis of all keywords, which are automatically clustered into three research hotspots. Using CiteSpace to perform timeline-based keyword cooccurrence analysis, we find that scholars have gone through a “discovery-acceptance-question-improvement” process for the application of blockchain technology in supply chain management. According to the intensity and duration of burst detection words in different years, we draw three key points for future research. (i) Blockchain technology should be used to restructure and optimize the high-end food and medical cold chain supply chains. (ii) Blockchain technology should be used to enhance the cooperative relationship between supply chain members and the overall competitiveness of the supply chain. (iii) Innovate blockchain encryption technology that should be used to reduce the risk of information and privacy leakage in the supply chain.

Normative Values for Word Syllable Duration With Interpretation in a Large Sample of Stroke Survivors With Aphasia.

Slow speech rate and abnormal temporal prosody are primary diagnostic criteria for differentiating between people with aphasia who do and do not have apraxia of speech. We sought to identify appropriate cutoff values for abnormal word syllable duration (WSD) in a word repetition task, interpret them relative to a data set of people with chronic aphasia, and evaluate the extent to which manually derived measures could be approximated through an automated process that relied on commercial speech recognition technology. Fifty neurotypical participants produced 49 multisyllabic words during a repetition task. Audio recordings were submitted to an automated speech recognition (ASR) service (IBM Watson) to measure word duration and generate an orthographic transcription. The transcribed words were compared to a lexical database, and the number of syllables was identified. Automatic and manual measures were compared for 50% of the sample. Results were interpreted relative to WSD scores from an existing data set of 195 people with mostly chronic aphasia. ASR correctly identified 83% of target words and 98% of target syllable counts. Automated word duration calculations were longer than manual measures due to imprecise cursor placement. Upon applying regression coefficients to the automated measures and examining the frequency distributions for both manual and estimated measures, a WSD of 303-316 ms was found to indicate longer-than-normal performance (corresponding to the 95th percentile). With this cutoff, 40%-45% of participants with aphasia in our comparison sample had an abnormally long WSD. We recommend using a rounded WSD cutoff score between 303 and 316 ms for manual measures. Future research will focus on customizing automated WSD methods to speech samples from people with aphasia, identifying target words that maximize production and measurement reliability, and developing WSD standard scores based on a large participant sample with and without aphasia.

Cumulative exposure to fast speech conditions duration of content words in English

AbstractThis paper tests the idea that the speech rate with which surrounding words are spoken affects the mental representation of words and conditions production of words. This possibility is operationalized by measuring a word's ratio of occurrence in speaker-relative fast speech. Other variables shown in the literature to influence speech rate are controlled for in a 10,000-iteration bootstrapping procedure of a mixed-effect linear regression model. The results of the analysis of 39,397 tokens of content words from 1,232 word types in English display a significant effect for a word's ratio of conditioning in speaker-relative fast speech, although the effect size is small or very small. Other variables shown in the literature to condition speech rate also significantly condition speech rate here. This paper suggests that in addition to other aspects of the context of use of words, contextual speech rate also influences the mental representation of words.

Dynamic Temporal and Tactile Cueing: Quantifying Speech Motor Changes and Individual Factors That Contribute to Treatment Gains in Childhood Apraxia of Speech.

Speech motor skill is refined over the course of practice, which is commonly reflected by increased accuracy and consistency. This research examined the relationship between auditory-perceptual ratings of word accuracy and measures of speech motor timing and variability at pre- and posttreatment in children with childhood apraxia of speech (CAS). Furthermore, the degree to which individual patterns of baseline probe word accuracy, receptive language, and cognition predicted response to treatment was explored. Probe data were collected from seven children with CAS (aged 2;5-5;0 [years;months]) who received 6 weeks of Dynamic Temporal and Tactile Cueing (DTTC) treatment. Using a multidimensional approach to measuring speech performance, auditory-perceptual (whole-word accuracy), acoustic (whole-word duration), and kinematic (jaw movement variability) analyses were conducted on probe words produced pre- and posttreatment. Standardized tests of receptive language and cognition were administered pretreatment. There was a negative relationship between auditory-perceptual measures of word accuracy and movement variability. Higher word accuracy was associated with lower jaw movement variability following intervention. There was a strong relationship between word accuracy and word duration at baseline, which became less robust posttreatment. Furthermore, baseline word accuracy was the only child-specific factor to predict response to DTTC treatment. Following a period of motor-based intervention, children with CAS appeared to refine speech motor control in conjunction with improvements in word accuracy. Those who demonstrated the poorest performance at treatment onset displayed the greatest degree of gains. Taken together, these results reflect a system-wide change following motor-based intervention.

Application of advanced language technologies in analysis of category naming fluency task in healthy participants

AbstractBackgroundCategory naming fluency tasks have been used frequently to measure patients’ working memory and executive functions. Previous studies show that differences in scores in this task could reflect features of neurodegenerative disease. To better understand decreased performance in patients, it is critical to establish performance in healthy aging populations. Here, we investigated differences in total scores and lexical/speech features of words produced during animal fluency tasks of healthy young and older speakers which can only be derived with novel, automated, computational technologies.MethodData were collected in 52 young participants (26.9% males, mean age = 23.5±3.17) and 46 elderly participants (63.0% males, mean age = 69.9±6.77). Participants were instructed to list as many animals as possible for 60 seconds. Speech was transcribed and automatically force‐aligned. Transcripts were run through our automated Python pipeline to calculate the total number of animals listed and rate each word for the following semantic parameters: age of acquisition (AoA), ambiguity, familiarity, frequency. We also calculated phonetic and semantic distances between each word and its antecedent. Linear regression models tested the association between these automated language measures and total scores. We also examined the differences of the young and elderly groups using t‐tests.ResultElderly participants produced significantly fewer animals than young participants (p&lt;0.001). Higher overall scores were associated with higher mean AoA (beta = 0.22, p = 0.049) and longer word duration (beta = 1.62, p = 0.037), as well as less frequent (beta = ‐0.51, p = 0.021) and familiar (beta = ‐3.96, p&lt;0.001) animal names. We found no interaction with the age groups in any of the models. Elderly participants produced less frequent animal names than young participants (p = 0.029), along with significantly reduced phonetic distance scores (p = 0.008). Elderly participants scored significantly higher articulation rates (p = 0.007) and had a faster speech rate (p&lt;0.001), attributable to producing more non‐animal words.ConclusionThis study demonstrates the applicability of advanced language technologies in analyzing category naming fluency tasks in healthy adults. We developed an automated scoring system and identify novel semantic and phonetic measures from digitally recorded fluency tasks. We also quantified some confounding age effects which could be explained by lexical/speech differences of words produced by the groups.

Effects of utterance length on word and phrase durations

This study examined the effects of utterance length on word and phrase durations to investigate how the planned length of an utterance influences the rate of speech. An experiment was conducted in which the visual stimuli cued the production of sentences with one, two, or three subject noun phrases (NPs)—e.g., “Nine green rhinos and eight red weasels and eight blue llamas live in the zoo.” In addition, a novel, delayed stimuli condition was tested in which the visual stimuli that cued non-initial NPs (“eight red weasels and eight blue llamas”) were presented after participants started production, and thus sentence length changed after utterance initiation. Word and NP durations were analyzed. NP durations were longer in sentences with more NPs, and moreover, durational increases were primarily observed in the rightmost words of the NPs. These findings are important first because they show that the previously observed effects of number of words on word durations extend to phrases, and second because they show that phrase duration effects are localized to phrase ends. When the utterance length changed after the start of production, participants lengthened the end of the initial NP and the following conjunction to plan for the newly presented NPs.