doi
stringlengths 9
10
| chunk-id
stringlengths 1
4
| chunk
stringlengths 1
1.57k
| id
stringlengths 9
10
| title
stringlengths 10
127
| summary
stringlengths 581
1.92k
| source
stringlengths 30
31
| authors
sequence | categories
sequence | comment
stringclasses 28
values | journal_ref
stringclasses 1
value | primary_category
stringclasses 6
values | published
stringlengths 8
8
| updated
stringlengths 8
8
| references
list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2005.00928 | 6 | in the next section, besides input gradients, we
employ an input ablation method, blank-out , to estimate an importance score for each input token.
Blank-out replaces each token in the input, one
by one, with UNK and measures how much it affects the predicted probability of the correct class.
We compute the Spearman’s rank correlation coefficient between the attention weights of the CLS
embedding in the final layer and the importance
scores from blank-out. As shown in the first row
of Table 1, the correlation between raw attention
weights of the CLS token and blank-out scores is
rather low, except for the first layer. As we can see
in Table 2 this is also the case when we compute
the correlations with input gradients.
L1 L2 L3 L4 L5 L6
Raw 0.690.27 0.100.43 -0.110.49 -0.090.52 0.200.45 0.290.39
Rollout 0.320.26 0.380.27 0.510.26 0.620.26 0.700.25 0.710.24 | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 7 | Flow 0.320.26 0.440.29 0.700.25 0.700.22 0.710.22 0.700.22
Table 1: SpearmanR correlation of attention based importance with blank-out scores for 2000 samples from
the test set for the verb number prediction model.
3 Attention Rollout and Attention Flow
Attention rollout and attention flow recursively
compute the token attentions in each layer of a
L1 L2 L3 L4 L5 L6
Raw 0.530.33 0.160.38 -0.060.42 0.000.47 0.240.40 0.460.35
Rollout 0.220.31 0.270.32 0.390.32 0.470.32 0.530.32 0.540.31
Flow 0.220.31 0.310.34 0.540.32 0.610.28 0.600.28 0.610.28
Table 2: SpearmanR correlation of attention based importance with input gradients for 2000 samples from
the test set for the verb number prediction model. | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 8 | Table 2: SpearmanR correlation of attention based importance with input gradients for 2000 samples from
the test set for the verb number prediction model.
given model given the embedding attentions as input. They differ in the assumptions they make
about how attention weights in lower layers affect
the flow of information to the higher layers and
whether to compute the token attentions relative to
each other or independently.
To compute how information propagates from
the input layer to the embeddings in higher layers, it is crucial to take the residual connections
in the model into account as well as the attention
weights. In a Transformer block, both self-attention
and feed-forward networks are wrapped by residual connections, i.e., the input to these modules is
added to their output. When we only use attention
weights to approximate the flow of information in
Transformers, we ignore the residual connections.
But these connections play a significant role in
tying corresponding positions in different layers.
Hence, to compute attention rollout and attention
flow, we augment the attention graph with extra
weights to represent residual connections. Given
the attention module with residual connection, we
compute values in layer l+1asVl+1=Vl+WattVl,
where Wattis the attention matrix. Thus, we have | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 9 | compute values in layer l+1asVl+1=Vl+WattVl,
where Wattis the attention matrix. Thus, we have
Vl+1= (Watt+I)Vl. So, to account for residual
connections, we add an identity matrix to the attention matrix and re-normalize the weights. This
results in A= 0:5Watt+ 0:5I, where Ais the raw
attention updated by residual connections.
Furthermore, analyzing individual heads requires accounting for mixing of information between heads through a position-wise feed-forward
network in Transformer block. Using attention rollout and attention flow, it is also possible to analyze
each head separately. We explain in more details
in Appendix A.1. However, in our analysis in this
paper, for simplicity, we average the attention at
each layer over all heads.
Attention rollout Attention rollout is an intuitive
way of tracking down the information propagated
from the input layer to the embeddings in the higher
layers. Given a Transformer with Llayers, we want
to compute the attention from all positions in layerlito all positions in layer lj, where j < i . In the
attention graph, a path from node vat position k
inli, to node uat position minlj, is a series of | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 10 | attention graph, a path from node vat position k
inli, to node uat position minlj, is a series of
edges that connect these two nodes. If we look
at the weight of each edge as the proportion of
information transferred between two nodes, we
can compute how much of the information at v
is propagated to uthrough a particular path by
multiplying the weights of all edges in that path.
Since there may be more than one path between
two nodes in the attention graph, to compute the
total amount of information propagated from vtou,
we sum over all possible paths between these two
nodes. At the implementation level, to compute the
attentions from litolj, we recursively multiply the
attention weights matrices in all the layers below.
~A(li) =
A(li)~A(li 1)ifi > j
A(li) ifi=j(1)
In this equation, ~Ais attention rollout, Ais raw attention and the multiplication operation is a matrix
multiplication. With this formulation, to compute
input attention we set j= 0.
Attention flow In graph theory, a flow network
is a directed graph with a “capacity” associated | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 11 | input attention we set j= 0.
Attention flow In graph theory, a flow network
is a directed graph with a “capacity” associated
with each edge. Formally, given G= (V; E)is a
graph, where Vis the set of nodes, and Eis the set
of edges in G;C=fcuv2Rj8u; vwhere eu;v2
E^u6=vgdenotes the capacities of the edges and
s; t2Vare the source and target (sink) nodes respectively; flow is a mapping of edges to real numbers,f:E!R, that satisfies two conditions: (a)
capacity constraint : for each edge the flow value
should not exceed its capacity, jfuvcuvj; (b)
flow conservation : for all nodes except sandtthe
input flow should be equal to output flow –sum
of the flow of outgoing edges should be equal to
sum of the flow of incoming edges. Given a flow
network, a maximum flow algorithm finds a flow
which has the maximum possible value between s
andt(Cormen et al., 2009). | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 12 | network, a maximum flow algorithm finds a flow
which has the maximum possible value between s
andt(Cormen et al., 2009).
Treating the attention graph as a flow network,
where the capacities of the edges are attention
weights, using any maximum flow algorithm, we
can compute the maximum attention flow from any
node in any of the layers to any of the input nodes.
We can use this maximum-flow-value as an approximation of the attention to input nodes. In attention
flow, the weight of a single path is the minimum
value of the weights of the edges in the path, instead of the product of the weights. Besides, we
attention rollout
attention flow
Figure 3: Attention maps for the CLS token
.
can not compute the attention for node sto node
tby adding up the weights of all paths between
these two nodes, since there might be an overlap
between the paths and this might result in overflow
in the overlapping edges.
It is noteworthy that both of the proposed methods can be computed in polynomial time. O(dn2)
for attention rollout and O(d2n4)for attention
flow, where dis the depth of the model, and nis | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 13 | for attention rollout and O(d2n4)for attention
flow, where dis the depth of the model, and nis
the number of tokens.
4 Analysis and Discussion
Now, we take a closer look at these three views of
attention. Figure 1 depicts raw attention, attention
rollout and attention flow for a correctly classified
example across different layers. It is noteworthy
that the first layer of attention rollout and attention
flow are the same, and their only difference with
raw attention is the addition of residual connections. As we move to the higher layers, we see that
the residual connections fade away. Moreover, in
contrast to raw attention, the patterns of attention
rollout and attention flow become more distinctive
in the higher layers.
Figures 2 and 3 show the weights from raw attention, attention rollout and attention flow for the
CLS embedding over input tokens (x-axis) in all
6 layers (y-axis) for three examples. The first example is the same as the one in Figure 1. The second example is “ the article on NNP large systems
<?>”. The model correctly classifies this example and changing the subject of the missing verb | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 14 | <?>”. The model correctly classifies this example and changing the subject of the missing verb
from “article” to “article s” flips the decision of the
model. The third example is “ here the NNS differ
in that the female <?>”, which is a miss-classified
example and again changing “NNS” (plural noun)
to “NNP” (singular proper noun) flips the decision
of the model.
For all cases, the raw attention weights are almost uniform above layer three (discussed before).
raw attention
attention rollout
attention flow
(a) “The author talked to Sara about mask book .”
raw attention
attention rollout
attention flow
(b)“Mary convinced John of mask love.”
Figure 4: Bert attention maps. We look at the attention
weights from the mask embedding to the two potential
references for it, e.g. “author” and “Sara” in (a) and
“Mary” and “John” in (b). The bars, at the left, show
the relative predicted probability for the two possible | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 15 | “Mary” and “John” in (b). The bars, at the left, show
the relative predicted probability for the two possible
pronouns, “his” and “her”.
In the case of the correctly classified example, we
observe that both attention rollout and attention
flow assign relatively high weights to both the subject of the verb, “article’ and the attractor, “systems”. For the miss-classified example, both attention rollout and attention flow assign relatively
high scores to the “NNS” token which is not the
subject of the verb. This can explain the wrong
prediction of the model.
The main difference between attention rollout
and attention flow is that attention flow weights are
amortized among the set of most attended tokens,
as expected. Attention flow can indicate a set of
input tokens that are important for the final decision.
Thus we do not get sharp distinctions among them.
On the other hand, attention rollout weights are
more focused compared to attention flow weights,
which is sensible for the third example but not as
much for the second one.
L1 L3 L5 L6 | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 16 | more focused compared to attention flow weights,
which is sensible for the third example but not as
much for the second one.
L1 L3 L5 L6
Raw 0.120.21 0.090.21 0.080.20 0.090.21
Rollout 0.110.19 0.120.21 0.130.21 0.130.20
Flow 0.110.19 0.110.21 0.120.22 0.140.21
Table 3: SpearmanR correlation of attention based importance with input gradients for 100 samples from the
test set for the DistillBERT model fine tuned on SST-2.
Furthermore, as shown in Table 1 and 2 both
attention rollout and attention flow, are better
correlated with blank-out scores and input gradients compared to raw attention, but attention flow
weights are more reliable than attention rollout.
The difference between these two methods is rooted
in their different views of attention weights. Attention flow views them as capacities, and at every
step of the algorithm, it uses as much of the capacity as possible. Hence, attention flow computes the
maximum possibility of token identities to propagate to the higher layers. Whereas attention rollout | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 17 | maximum possibility of token identities to propagate to the higher layers. Whereas attention rollout
views them as proportion factors and at every step,
it allows token identities to be propagated to higher
layers exactly based on this proportion factors. This
makes attention rollout stricter than attention flow,
and so we see that attention rollout provides us with
more focused attention patterns. However, since
we are making many simplifying assumptions, the
strictness of attention rollout does not lead to more
accurate results, and the relaxation of attention flow
seems to be a useful property.
At last, to illustrate the application of attention flow and attention rollout on different tasks
and different models, we examine them on two
pretrained BERT models. We use the models
available at https://github.com/huggingface/
transformers .
Table 3 shows the correlation of the importance
score obtained from raw attention, attention rollout
and attention flow from a DistillBERT (Sanh et al.,
2019) model fine-tuned to solve “SST-2” (Socher
et al., 2013), the sentiment analysis task from the
glue benchmark (Wang et al., 2018). Even though
for this model, all three methods have very low
correlation with the input gradients, we can still see | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 18 | glue benchmark (Wang et al., 2018). Even though
for this model, all three methods have very low
correlation with the input gradients, we can still see
that attention rollout and attention flow are slightly
better than raw attention.
Furthermore, in Figure 4, we show an example
of applying these methods to a pre-trained Bert
to see how it resolves the pronouns in a sentence.
What we do here is to feed the model with a sentence, masking a pronoun. Next, we look at the
prediction of the model for the masked word and
compare the probabilities assigned to “her” and
“his”. Then we look at raw attention, attention rollout and attention flow weights of the embeddings
for the masked pronoun at all the layers. In the
first example, in Figure 4a, attention rollout and
attention flow are consistent with each other and
the prediction of the model. Whereas, the final
layer of raw attention does not seem to be consistent with the prediction of the models, and it varies
a lot across different layers. In the second example, in Figure 4b, only attention flow weights areconsistent with the prediction of the model.
5 Conclusion
Translating embedding attentions to token attentions can provide us with better explanations about | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 19 | 5 Conclusion
Translating embedding attentions to token attentions can provide us with better explanations about
models’ internals. Yet, we should be cautious about
our interpretation of these weights, because, we
are making many simplifying assumptions when
we approximate information flow in a model with
the attention weights. Our ideas are simple and
task/architecture agnostic. In this paper, we insisted on sticking with simple ideas that only require attention weights and can be easily employed
in any task or architecture that uses self-attention.
We should note that all our analysis in this paper is
for a Transformer encoder, with no casual masking.
Since in Transformer decoder, future tokens are
masked, naturally there is more attention toward
initial tokens in the input sequence, and both attention rollout and attention flow will be biased toward
these tokens. Hence, to apply these methods on a
Transformer decoder, we should first normalize
based on the receptive field of attention.
Following this work, we can build the attention
graph with effective attention weights (Brunner
et al., 2020) instead of raw attentions. Furthermore,
we can come up with a new method that adjusts the
attention weights using gradient-based attribution
methods (Ancona et al., 2019).
Acknowledgements | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 20 | we can come up with a new method that adjusts the
attention weights using gradient-based attribution
methods (Ancona et al., 2019).
Acknowledgements
We thank Mostafa Dehghani, Wilker Aziz, and the
anonymous reviewers for their valuable feedback
and comments on this work. The work presented
here was funded by the Netherlands Organization
for Scientific Research (NWO), through a Gravitation Grant 024.001.006 to the Language in Interaction Consortium.
References
Marco Ancona, Enea Ceolini, Cengiz ¨Oztireli, and
Markus Gross. 2019. Gradient-Based Attribution
Methods , pages 169–191. Springer International
Publishing.
Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2015. Neural machine translation by jointly
learning to align and translate. In proceedings of
the 2015 International Conference on Learning Representations .
Gino Brunner, Yang Liu, Damian Pascual, Oliver
Richter, Massimiliano Ciaramita, and Roger Wattenhofer. 2020. On identifiability in transformers. In
International Conference on Learning Representations .
Hanjie Chen and Yangfeng Ji. 2019. Improving the interpretability of neural sentiment classifiers via data | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 21 | International Conference on Learning Representations .
Hanjie Chen and Yangfeng Ji. 2019. Improving the interpretability of neural sentiment classifiers via data
augmentation. arXiv preprint arXiv:1909.04225 .
Kevin Clark, Urvashi Khandelwal, Omer Levy, and
Christopher D. Manning. 2019. What does BERT
look at? an analysis of BERT’s attention. In Proceedings of the 2019 ACL Workshop BlackboxNLP:
Analyzing and Interpreting Neural Networks for
NLP, pages 276–286, Florence, Italy. Association
for Computational Linguistics.
Andy Coenen, Emily Reif, Ann Yuan, Been Kim,
Adam Pearce, Fernanda Vi ´egas, and Martin Wattenberg. 2019. Visualizing and measuring the geometry
of bert. arXiv preprint arXiv:1906.02715 .
Thomas H. Cormen, Charles E. Leiserson, Ronald L.
Rivest, and Clifford Stein. 2009. Introduction to
Algorithms, Third Edition , 3rd edition. The MIT
Press.
Mostafa Dehghani, Stephan Gouws, Oriol Vinyals, | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 22 | Algorithms, Third Edition , 3rd edition. The MIT
Press.
Mostafa Dehghani, Stephan Gouws, Oriol Vinyals,
Jakob Uszkoreit, and Łukasz Kaiser. 2019. Universal transformers. In proceedings of the 2019 International Conference on Learning Representations .
Jacob Devlin, Ming-Wei Chang, Kenton Lee, and
Kristina Toutanova. 2019. BERT: Pre-training of
deep bidirectional transformers for language understanding. In proceedings of the 2019 Conference
of the North American Chapter of the Association
for Computational Linguistics: Human Language
Technologies . Association for Computational Linguistics.
Sarthak Jain and Byron C. Wallace. 2019. Attention
is not Explanation. In proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies , pages 3543–3556. Association
for Computational Linguistics.
Jaesong Lee, Joong-Hwi Shin, and Jun-Seok Kim.
2017. Interactive visualization and manipulation
of attention-based neural machine translation. In
proceedings of the 2017 Conference on Empirical
Methods in Natural Language Processing: System
Demonstrations , pages 121–126.
Tal Linzen, Emmanuel Dupoux, and Yoav Goldberg. | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 23 | Methods in Natural Language Processing: System
Demonstrations , pages 121–126.
Tal Linzen, Emmanuel Dupoux, and Yoav Goldberg.
2016. Assessing the ability of LSTMs to learn
syntax-sensitive dependencies. Transactions of the
Association for Computational Linguistics , 4:521–
535.
Danish Pruthi, Mansi Gupta, Bhuwan Dhingra, Graham Neubig, and Zachary C Lipton. 2019. Learning
to deceive with attention-based explanations. arXiv
preprint arXiv:1909.07913 .Alec Radford, Jeffrey Wu, Rewon Child, David Luan,
Dario Amodei, and Ilya Sutskever. 2019. Language
models are unsupervised multitask learners. OpenAI
Blog , 1(8).
Tim Rockt ¨aschel, Edward Grefenstette, Karl Moritz
Hermann, Tomas Kocisky, and Phil Blunsom. 2016.
Reasoning about entailment with neural attention.
InInternational Conference on Learning Representations (ICLR) .
Victor Sanh, Lysandre Debut, Julien Chaumond, and
Thomas Wolf. 2019. Distilbert, a distilled version of
bert: smaller, faster, cheaper and lighter. | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 24 | Thomas Wolf. 2019. Distilbert, a distilled version of
bert: smaller, faster, cheaper and lighter.
Sofia Serrano and Noah A. Smith. 2019. Is attention
interpretable? In proceedings of the 57th Annual
Meeting of the Association for Computational Linguistics . Association for Computational Linguistics.
Richard Socher, Alex Perelygin, Jean Wu, Jason
Chuang, Christopher D. Manning, Andrew Ng, and
Christopher Potts. 2013. Recursive deep models
for semantic compositionality over a sentiment treebank. In Proceedings of the 2013 Conference on
Empirical Methods in Natural Language Processing ,
pages 1631–1642, Seattle, Washington, USA. Association for Computational Linguistics.
Shikhar Vashishth, Shyam Upadhyay, Gaurav Singh
Tomar, and Manaal Faruqui. 2019. Attention interpretability across nlp tasks. arXiv preprint
arXiv:1909.11218 .
Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob
Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz
Kaiser, and Illia Polosukhin. 2017. Attention is all
you need. In Advances in neural information processing systems , pages 5998–6008. | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 25 | Kaiser, and Illia Polosukhin. 2017. Attention is all
you need. In Advances in neural information processing systems , pages 5998–6008.
Jesse Vig. 2019. Visualizing attention in transformerbased language models. arXiv preprint
arXiv:1904.02679 .
Alex Wang, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel Bowman. 2018.
GLUE: A multi-task benchmark and analysis platform for natural language understanding. In Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP , pages 353–355, Brussels, Belgium.
Association for Computational Linguistics.
Yequan Wang, Minlie Huang, Li Zhao, et al. 2016.
Attention-based lstm for aspect-level sentiment classification. In proceedings of the 2016 conference on
empirical methods in natural language processing ,
pages 606–615.
Sarah Wiegreffe and Yuval Pinter. 2019. Attention is
not not explanation. In proceedings of the 2019 Conference on Empirical Methods in Natural Language
Processing and the 9th International Joint Conference on Natural Language Processing (EMNLPIJCNLP) . Association for Computational Linguistics.
Thomas Wolf, Lysandre Debut, Victor Sanh, Julien | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 26 | Thomas Wolf, Lysandre Debut, Victor Sanh, Julien
Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, R’emi Louf, Morgan Funtowicz, and Jamie Brew. 2019. Huggingface’s transformers: State-of-the-art natural language processing. ArXiv , abs/1910.03771.
Kelvin Xu, Jimmy Ba, Ryan Kiros, Kyunghyun Cho,
Aaron Courville, Ruslan Salakhudinov, Rich Zemel,
and Yoshua Bengio. 2015. Show, attend and tell:
Neural image caption generation with visual attention. In proceedings of International Conference on
Machine Learning , pages 2048–2057.
A Appendices
A.1 Single Head Analysis
For analysing the attention weights, with multihead setup, we could either analyze attention heads
separately, or we could average all heads and have a
single attention graph. However, we should be careful that treating attention heads separately could
potentially mean that we are assuming there is no
mixing of information between heads, which is not
true as we combine information of heads in the
position-wise feed-forward network on top of selfattention in a transformer block. It is possible to
analyse the role of each head in isolation of all other | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.00928 | 27 | position-wise feed-forward network on top of selfattention in a transformer block. It is possible to
analyse the role of each head in isolation of all other
heads using attention rollout and attention flow. To
not make the assumption that there is no mixing
of information between heads, for computing the
“input attention”, we will treat all the layers below
the layer of interest as single head layers, i.e., we
sum the attentions of all heads in the layers below.
For example, we can compute attention rollout for
headkat layer ias~A(i; k) =A(i; k)A(i), where,
A(i)is attention rollout computed for layer iwith
the single head assumption. | 2005.00928 | Quantifying Attention Flow in Transformers | In the Transformer model, "self-attention" combines information from attended
embeddings into the representation of the focal embedding in the next layer.
Thus, across layers of the Transformer, information originating from different
tokens gets increasingly mixed. This makes attention weights unreliable as
explanations probes. In this paper, we consider the problem of quantifying this
flow of information through self-attention. We propose two methods for
approximating the attention to input tokens given attention weights, attention
rollout and attention flow, as post hoc methods when we use attention weights
as the relative relevance of the input tokens. We show that these methods give
complementary views on the flow of information, and compared to raw attention,
both yield higher correlations with importance scores of input tokens obtained
using an ablation method and input gradients. | http://arxiv.org/pdf/2005.00928 | [
"Samira Abnar",
"Willem Zuidema"
] | [
"cs.LG",
"cs.AI",
"cs.CL"
] | null | null | cs.LG | 20200502 | 20200531 | [] |
2005.14050 | 0 | Language (Technology) is Power: A Critical Survey of “Bias” in NLP
Su Lin Blodgett
College of Information and Computer Sciences
University of Massachusetts Amherst
[email protected] Barocas
Microsoft Research
Cornell University
[email protected]
Hal Daumé III
Microsoft Research
University of Maryland
[email protected] Wallach
Microsoft Research
[email protected]
Abstract
We survey 146 papers analyzing “bias” in
NLP systems, finding that their motivations
are often vague, inconsistent, and lacking
in normative reasoning, despite the fact that
analyzing “bias” is an inherently normative
process. We further find that these papers’
proposed quantitative techniques for measuring or mitigating “bias” are poorly matched to
their motivations and do not engage with the
relevant literature outside of NLP. Based on
these findings, we describe the beginnings of a
path forward by proposing three recommendations that should guide work analyzing “bias”
in NLP systems. These recommendations rest
on a greater recognition of the relationships
between language and social hierarchies,
encouraging researchers and practitioners
to articulate their conceptualizations of | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 1 | in NLP systems. These recommendations rest
on a greater recognition of the relationships
between language and social hierarchies,
encouraging researchers and practitioners
to articulate their conceptualizations of
“bias”—i.e., what kinds of system behaviors
are harmful, in what ways, to whom, and why,
as well as the normative reasoning underlying
these statements—and to center work around
the lived experiences of members of communities affected by NLP systems, while interrogating and reimagining the power relations
between technologists and such communities.
1 Introduction
A large body of work analyzing “bias” in natural
language processing (NLP) systems has emerged
in recent years, including work on “bias” in embedding spaces (e.g., Bolukbasi et al., 2016a; Caliskan
et al., 2017; Gonen and Goldberg, 2019; May
et al., 2019) as well as work on “bias” in systems
developed for a breadth of tasks including language
modeling (Lu et al., 2018; Bordia and Bowman,2019), coreference resolution (Rudinger et al.,
2018; Zhao et al., 2018a), machine translation (Vanmassenhove et al., 2018; Stanovsky et al., 2019),
sentiment analysis (Kiritchenko and Mohammad, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 2 | 2018; Zhao et al., 2018a), machine translation (Vanmassenhove et al., 2018; Stanovsky et al., 2019),
sentiment analysis (Kiritchenko and Mohammad,
2018), and hate speech/toxicity detection (e.g.,
Park et al., 2018; Dixon et al., 2018), among others.
Although these papers have laid vital groundwork by illustrating some of the ways that NLP
systems can be harmful, the majority of them fail
to engage critically with what constitutes “bias”
in the first place. Despite the fact that analyzing
“bias” is an inherently normative process—in
which some system behaviors are deemed good
and others harmful—papers on “bias” in NLP
systems are rife with unstated assumptions about
what kinds of system behaviors are harmful, in
what ways, to whom, and why. Indeed, the term
“bias” (or “gender bias” or “racial bias”) is used
to describe a wide range of system behaviors, even
though they may be harmful in different ways, to
different groups, or for different reasons. Even
papers analyzing “bias” in NLP systems developed
for the same task often conceptualize it differently.
For example, the following system behaviors | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 3 | papers analyzing “bias” in NLP systems developed
for the same task often conceptualize it differently.
For example, the following system behaviors
are all understood to be self-evident statements of
“racial bias”: (a) embedding spaces in which embeddings for names associated with African Americans
are closer (compared to names associated with
European Americans) to unpleasant words than
pleasant words (Caliskan et al., 2017); (b) sentiment analysis systems yielding different intensity
scores for sentences containing names associated
with African Americans and sentences containing
names associated with European Americans (Kiritchenko and Mohammad, 2018); and (c) toxicityarXiv:2005.14050v2 [cs.CL] 29 May 2020
detection systems scoring tweets containing features associated with African-American English as
more offensive than tweets without these features
(Davidson et al., 2019; Sap et al., 2019). Moreover,
some of these papers focus on “racial bias”
expressed in written text, while others focus on
“racial bias” against authors. This use of imprecise
terminology obscures these important differences.
We survey 146 papers analyzing “bias” in NLP
systems, finding that their motivations are often
vague and inconsistent. Many lack any normative | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 4 | We survey 146 papers analyzing “bias” in NLP
systems, finding that their motivations are often
vague and inconsistent. Many lack any normative
reasoning for why the system behaviors that are
described as “bias” are harmful, in what ways, and
to whom. Moreover, the vast majority of these
papers do not engage with the relevant literature
outside of NLP to ground normative concerns when
proposing quantitative techniques for measuring
or mitigating “bias.” As a result, we find that many
of these techniques are poorly matched to their
motivations, and are not comparable to one another.
We then describe the beginnings of a path
forward by proposing three recommendations
that should guide work analyzing “bias” in NLP
systems. We argue that such work should examine
the relationships between language and social hierarchies; we call on researchers and practitioners
conducting such work to articulate their conceptualizations of “bias” in order to enable conversations
about what kinds of system behaviors are harmful,
in what ways, to whom, and why; and we recommend deeper engagements between technologists
and communities affected by NLP systems. We
also provide several concrete research questions
that are implied by each of our recommendations.
2 Method
Our survey includes all papers known to us | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 5 | and communities affected by NLP systems. We
also provide several concrete research questions
that are implied by each of our recommendations.
2 Method
Our survey includes all papers known to us
analyzing “bias” in NLP systems—146 papers in
total. We omitted papers about speech, restricting
our survey to papers about written text only. To
identify the 146 papers, we first searched the ACL
Anthology1for all papers with the keywords “bias”
or “fairness” that were made available prior to May
2020. We retained all papers about social “bias,”
and discarded all papers about other definitions of
the keywords (e.g., hypothesis-only bias, inductive
bias, media bias). We also discarded all papers using “bias” in NLP systems to measure social “bias”
in text or the real world (e.g., Garg et al., 2018).
To ensure that we did not exclude any relevant
1https://www.aclweb.org/anthology/NLP task Papers
Embeddings (type-level or contextualized) 54
Coreference resolution 20
Language modeling or dialogue generation 17
Hate-speech detection 17
Sentiment analysis 15
Machine translation 8
Tagging or parsing 5 | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 6 | Coreference resolution 20
Language modeling or dialogue generation 17
Hate-speech detection 17
Sentiment analysis 15
Machine translation 8
Tagging or parsing 5
Surveys, frameworks, and meta-analyses 20
Other 22
Table 1: The NLP tasks covered by the 146 papers.
papers without the keywords “bias” or “fairness,”
we also traversed the citation graph of our initial
set of papers, retaining any papers analyzing “bias”
in NLP systems that are cited by or cite the papers
in our initial set. Finally, we manually inspected
any papers analyzing “bias” in NLP systems from
leading machine learning, human–computer interaction, and web conferences and workshops, such
as ICML, NeurIPS, AIES, FAccT, CHI, and WWW,
along with any relevant papers that were made
available in the “Computation and Language” and
“Computers and Society” categories on arXiv prior
to May 2020, but found that they had already been
identified via our traversal of the citation graph. We
provide a list of all 146 papers in the appendix. In
Table 1, we provide a breakdown of the NLP tasks
covered by the papers. We note that counts do not | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 7 | provide a list of all 146 papers in the appendix. In
Table 1, we provide a breakdown of the NLP tasks
covered by the papers. We note that counts do not
sum to 146, because some papers cover multiple
tasks. For example, a paper might test the efficacy
of a technique for mitigating “bias” in embedding spaces in the context of sentiment analysis.
Once identified, we then read each of the 146 papers with the goal of categorizing their motivations
and their proposed quantitative techniques for measuring or mitigating “bias.” We used a previously
developed taxonomy of harms for this categorization, which differentiates between so-called allocational andrepresentational harms (Barocas et al.,
2017; Crawford, 2017). Allocational harms arise
when an automated system allocates resources (e.g.,
credit) or opportunities (e.g., jobs) unfairly to different social groups; representational harms arise
when a system (e.g., a search engine) represents
some social groups in a less favorable light than
others, demeans them, or fails to recognize their
existence altogether. Adapting and extending this
taxonomy, we categorized the 146 papers’ motivations and techniques into the following categories:
.Allocational harms.
Papers
Category Motivation Technique | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 8 | taxonomy, we categorized the 146 papers’ motivations and techniques into the following categories:
.Allocational harms.
Papers
Category Motivation Technique
Allocational harms 30 4
Stereotyping 50 58
Other representational harms 52 43
Questionable correlations 47 42
Vague/unstated 23 0
Surveys, frameworks, and
meta-analyses20 20
Table 2: The categories into which the 146 papers fall.
.Representational harms:2
.Stereotyping that propagates negative generalizations about particular social groups.
.Differences in system performance for different social groups, language that misrepresents the distribution of different social
groups in the population, or language that
isdenigrating to particular social groups.
.Questionable correlations between system behavior and features of language that are typically associated with particular social groups.
.Vague descriptions of “bias” (or “gender
bias” or “racial bias”) or no description at all.
.Surveys, frameworks, and meta-analyses .
In Table 2 we provide counts for each of the
six categories listed above. (We also provide a
list of the papers that fall into each category in the
appendix.) Again, we note that the counts do not
sum to 146, because some papers state multiple | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 9 | list of the papers that fall into each category in the
appendix.) Again, we note that the counts do not
sum to 146, because some papers state multiple
motivations, propose multiple techniques, or propose a single technique for measuring or mitigating
multiple harms. Table 3, which is in the appendix,
contains examples of the papers’ motivations and
techniques across a range of different NLP tasks.
3 Findings
Categorizing the 146 papers’ motivations and proposed quantitative techniques for measuring or mitigating “bias” into the six categories listed above enabled us to identify several commonalities, which
we present below, along with illustrative quotes.
2We grouped several types of representational harms into
two categories to reflect that the main point of differentiation
between the 146 papers’ motivations and proposed quantitative
techniques for measuring or mitigating “bias” is whether or not
they focus on stereotyping. Among the papers that do not focus on stereotyping, we found that most lack sufficiently clear
motivations and techniques to reliably categorize them further.3.1 Motivations
Papers state a wide range of motivations,
multiple motivations, vague motivations, and
sometimes no motivations at all. We found that
the papers’ motivations span all six categories, with
several papers falling into each one. Appropriately, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 10 | sometimes no motivations at all. We found that
the papers’ motivations span all six categories, with
several papers falling into each one. Appropriately,
papers that provide surveys or frameworks for analyzing “bias” in NLP systems often state multiple
motivations (e.g., Hovy and Spruit, 2016; Bender,
2019; Sun et al., 2019; Rozado, 2020; Shah et al.,
2020). However, as the examples in Table 3 (in the
appendix) illustrate, many other papers (33%) do
so as well. Some papers (16%) state only vague
motivations or no motivations at all. For example,
“[N]o human should be discriminated on the basis
of demographic attributes by an NLP system. ”
—Kaneko and Bollegala (2019)
“[P]rominent word embeddings [...] encode
systematic biases against women and black people
[...] implicating many NLP systems in scaling up
social injustice. ” —May et al. (2019)
These examples leave unstated what it might mean
for an NLP system to “discriminate,” what constitutes “systematic biases,” or how NLP systems
contribute to “social injustice” (itself undefined). | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 11 | contribute to “social injustice” (itself undefined).
Papers’ motivations sometimes include no normative reasoning. We found that some papers
(32%) are not motivated by any apparent normative
concerns, often focusing instead on concerns about
system performance. For example, the first quote
below includes normative reasoning—namely that
models should not use demographic information
to make predictions—while the other focuses on
learned correlations impairing system performance.
“In [text classification], models are expected to
make predictions with the semantic information
rather than with the demographic group identity
information ( e.g., ‘gay’, ‘black’) contained in the
sentences. ” —Zhang et al. (2020a)
“An over-prevalence of some gendered forms in the
training data leads to translations with identifiable
errors. Translations are better for sentences
involving men and for sentences containing
stereotypical gender roles. ”
—Saunders and Byrne (2020)
Even when papers do state clear motivations,
they are often unclear about why the system behaviors that are described as “bias” are harmful, in what ways, and to whom. We found that | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 12 | they are often unclear about why the system behaviors that are described as “bias” are harmful, in what ways, and to whom. We found that
even papers with clear motivations often fail to explain what kinds of system behaviors are harmful,
in what ways, to whom, and why. For example,
“Deploying these word embedding algorithms in
practice, for example in automated translation
systems or as hiring aids, runs the serious risk of
perpetuating problematic biases in important
societal contexts. ” —Brunet et al. (2019)
“[I]f the systems show discriminatory behaviors in
the interactions, the user experience will be
adversely affected. ” —Liu et al. (2019)
These examples leave unstated what “problematic
biases” or non-ideal user experiences might look
like, how the system behaviors might result in
these things, and who the relevant stakeholders
or users might be. In contrast, we find that papers
that provide surveys or frameworks for analyzing
“bias” in NLP systems often name who is harmed,
acknowledging that different social groups may
experience these systems differently due to their
different relationships with NLP systems or
different social positions. For example, Ruane
et al. (2019) argue for a “deep understanding of | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 13 | different relationships with NLP systems or
different social positions. For example, Ruane
et al. (2019) argue for a “deep understanding of
the user groups [sic] characteristics, contexts, and
interests” when designing conversational agents.
Papers about NLP systems developed for the
same task often conceptualize “bias” differently. Even papers that cover the same NLP task
often conceptualize “bias” in ways that differ substantially and are sometimes inconsistent. Rows 3
and 4 of Table 3 (in the appendix) contain machine
translation papers with different conceptualizations
of “bias,” leading to different proposed techniques,
while rows 5 and 6 contain papers on “bias” in embedding spaces that state different motivations, but
propose techniques for quantifying stereotyping.
Papers’ motivations conflate allocational and
representational harms. We found that the papers’ motivations sometimes (16%) name immediate representational harms, such as stereotyping,
alongside more distant allocational harms, which,
in the case of stereotyping, are usually imagined as
downstream effects of stereotypes on résumé filtering. Many of these papers use the imagined downstream effects to justify focusing on particular system behaviors, even when the downstream effects
are not measured. Papers on “bias” in embedding | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 14 | are not measured. Papers on “bias” in embedding
spaces are especially likely to do this because embeddings are often used as input to other systems:
“However, none of these papers [on embeddings]
have recognized how blatantly sexist the
embeddings are and hence risk introducing biases
of various types into real-world systems. ”
—Bolukbasi et al. (2016a)“It is essential to quantify and mitigate gender bias
in these embeddings to avoid them from affecting
downstream applications. ” —Zhou et al. (2019)
In contrast, papers that provide surveys or frameworks for analyzing “bias” in NLP systems treat
representational harms as harmful in their own
right. For example, Mayfield et al. (2019) and
Ruane et al. (2019) cite the harmful reproduction
of dominant linguistic norms by NLP systems (a
point to which we return in section 4), while Bender
(2019) outlines a range of harms, including seeing
stereotypes in search results and being made invisible to search engines due to language practices.
3.2 Techniques
Papers’ techniques are not well grounded in the
relevant literature outside of NLP. Perhaps unsurprisingly given that the papers’ motivations are
often vague, inconsistent, and lacking in normative | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 15 | Papers’ techniques are not well grounded in the
relevant literature outside of NLP. Perhaps unsurprisingly given that the papers’ motivations are
often vague, inconsistent, and lacking in normative
reasoning, we also found that the papers’ proposed
quantitative techniques for measuring or mitigating
“bias” do not effectively engage with the relevant
literature outside of NLP. Papers on stereotyping
are a notable exception: the Word Embedding
Association Test (Caliskan et al., 2017) draws on
the Implicit Association Test (Greenwald et al.,
1998) from the social psychology literature, while
several techniques operationalize the well-studied
“Angry Black Woman” stereotype (Kiritchenko
and Mohammad, 2018; May et al., 2019; Tan
and Celis, 2019) and the “double bind” faced by
women (May et al., 2019; Tan and Celis, 2019), in
which women who succeed at stereotypically male
tasks are perceived to be less likable than similarly
successful men (Heilman et al., 2004). Tan and
Celis (2019) also examine the compounding effects
of race and gender, drawing on Black feminist
scholarship on intersectionality (Crenshaw, 1989).
Papers’ techniques are poorly matched to their | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 16 | of race and gender, drawing on Black feminist
scholarship on intersectionality (Crenshaw, 1989).
Papers’ techniques are poorly matched to their
motivations. We found that although 21% of the
papers include allocational harms in their motivations, only four papers actually propose techniques
for measuring or mitigating allocational harms.
Papers focus on a narrow range of potential
sources of “bias.” We found that nearly all of the
papers focus on system predictions as the potential
sources of “bias,” with many additionally focusing
on “bias” in datasets (e.g., differences in the
number of gendered pronouns in the training data
(Zhao et al., 2019)). Most papers do not interrogate
the normative implications of other decisions made
during the development and deployment lifecycle—
perhaps unsurprising given that their motivations
sometimes include no normative reasoning. A
few papers are exceptions, illustrating the impacts
of task definitions, annotation guidelines, and
evaluation metrics: Cao and Daumé (2019) study
how folk conceptions of gender (Keyes, 2018) are
reproduced in coreference resolution systems that
assume a strict gender dichotomy, thereby maintaining cisnormativity; Sap et al. (2019) focus on
the effect of priming annotators with information | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 17 | assume a strict gender dichotomy, thereby maintaining cisnormativity; Sap et al. (2019) focus on
the effect of priming annotators with information
about possible dialectal differences when asking
them to apply toxicity labels to sample tweets, finding that annotators who are primed are significantly
less likely to label tweets containing features associated with African-American English as offensive.
4 A path forward
We now describe how researchers and practitioners
conducting work analyzing “bias” in NLP systems
might avoid the pitfalls presented in the previous
section—the beginnings of a path forward. We
propose three recommendations that should guide
such work, and, for each, provide several concrete
research questions. We emphasize that these questions are not comprehensive, and are intended to
generate further questions and lines of engagement.
Our three recommendations are as follows:
(R1) Ground work analyzing “bias” in NLP systems in the relevant literature outside of NLP
that explores the relationships between language and social hierarchies. Treat representational harms as harmful in their own right.
(R2) Provide explicit statements of why the
system behaviors that are described as “bias”
are harmful, in what ways, and to whom.
Be forthright about the normative reasoning
(Green, 2019) underlying these statements. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 18 | are harmful, in what ways, and to whom.
Be forthright about the normative reasoning
(Green, 2019) underlying these statements.
(R3) Examine language use in practice by engaging with the lived experiences of members of
communities affected by NLP systems. Interrogate and reimagine the power relations between technologists and such communities.
4.1 Language and social hierarchies
Turning first to (R1) , we argue that work analyzing
“bias” in NLP systems will paint a much fuller picture if it engages with the relevant literature outside
of NLP that explores the relationships betweenlanguage and social hierarchies. Many disciplines,
including sociolinguistics, linguistic anthropology,
sociology, and social psychology, study how
language takes on social meaning and the role that
language plays in maintaining social hierarchies.
For example, language is the means through which
social groups are labeled and one way that beliefs
about social groups are transmitted (e.g., Maass,
1999; Beukeboom and Burgers, 2019). Group
labels can serve as the basis of stereotypes and thus
reinforce social inequalities: “[T]he label content
functions to identify a given category of people,
and thereby conveys category boundaries and a
position in a hierarchical taxonomy” (Beukeboom | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 19 | functions to identify a given category of people,
and thereby conveys category boundaries and a
position in a hierarchical taxonomy” (Beukeboom
and Burgers, 2019). Similarly, “controlling
images,” such as stereotypes of Black women,
which are linguistically and visually transmitted
through literature, news media, television, and so
forth, provide “ideological justification” for their
continued oppression (Collins, 2000, Chapter 4).
As a result, many groups have sought to bring
about social changes through changes in language,
disrupting patterns of oppression and marginalization via so-called “gender-fair” language
(Sczesny et al., 2016; Menegatti and Rubini, 2017),
language that is more inclusive to people with
disabilities (ADA, 2018), and language that is less
dehumanizing (e.g., abandoning the use of the term
“illegal” in everyday discourse on immigration in
the U.S. (Rosa, 2019)). The fact that group labels
are so contested is evidence of how deeply intertwined language and social hierarchies are. Taking
“gender-fair” language as an example, the hope
is that reducing asymmetries in language about
women and men will reduce asymmetries in their | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 20 | “gender-fair” language as an example, the hope
is that reducing asymmetries in language about
women and men will reduce asymmetries in their
social standing. Meanwhile, struggles over language use often arise from dominant social groups’
desire to “control both material and symbolic
resources”—i.e., “the right to decide what words
will mean and to control those meanings”—as was
the case in some white speakers’ insistence on
using offensive place names against the objections
of Indigenous speakers (Hill, 2008, Chapter 3).
Sociolinguists and linguistic anthropologists
have also examined language attitudes and language ideologies, or people’s metalinguistic beliefs
about language: Which language varieties or practices are taken as standard, ordinary, or unmarked?
Which are considered correct, prestigious, or appropriate for public use, and which are considered
incorrect, uneducated, or offensive (e.g., CampbellKibler, 2009; Preston, 2009; Loudermilk, 2015;
Lanehart and Malik, 2018)? Which are rendered invisible (Roche, 2019)?3Language ideologies play
a vital role in reinforcing and justifying social hierarchies because beliefs about language varieties
or practices often translate into beliefs about their
speakers (e.g. Alim et al., 2016; Rosa and Flores, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 21 | or practices often translate into beliefs about their
speakers (e.g. Alim et al., 2016; Rosa and Flores,
2017; Craft et al., 2020). For example, in the U.S.,
the portrayal of non-white speakers’ language
varieties and practices as linguistically deficient
helped to justify violent European colonialism, and
today continues to justify enduring racial hierarchies by maintaining views of non-white speakers
as lacking the language “required for complex
thinking processes and successful engagement
in the global economy” (Rosa and Flores, 2017).
Recognizing the role that language plays in
maintaining social hierarchies is critical to the
future of work analyzing “bias” in NLP systems.
First, it helps to explain why representational
harms are harmful in their own right. Second, the
complexity of the relationships between language
and social hierarchies illustrates why studying
“bias” in NLP systems is so challenging, suggesting
that researchers and practitioners will need to move
beyond existing algorithmic fairness techniques.
We argue that work must be grounded in the
relevant literature outside of NLP that examines
the relationships between language and social
hierarchies; without this grounding, researchers
and practitioners risk measuring or mitigating
only what is convenient to measure or mitigate, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 22 | the relationships between language and social
hierarchies; without this grounding, researchers
and practitioners risk measuring or mitigating
only what is convenient to measure or mitigate,
rather than what is most normatively concerning.
More specifically, we recommend that work
analyzing “bias” in NLP systems be reoriented
around the following question: How are social
hierarchies, language ideologies, and NLP systems
coproduced? This question mirrors Benjamin’s
(2020) call to examine how “race and technology
are coproduced”—i.e., how racial hierarchies, and
the ideologies and discourses that maintain them,
create and are re-created by technology. We recommend that researchers and practitioners similarly
ask how existing social hierarchies and language
ideologies drive the development and deployment
of NLP systems, and how these systems therefore
reproduce these hierarchies and ideologies. As
a starting point for reorienting work analyzing
“bias” in NLP systems around this question, we
3Language ideologies encompass much more than this; see,
e.g., Lippi-Green (2012), Alim et al. (2016), Rosa and Flores
(2017), Rosa and Burdick (2017), and Charity Hudley (2017).provide the following concrete research questions: | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 23 | (2017), Rosa and Burdick (2017), and Charity Hudley (2017).provide the following concrete research questions:
.How do social hierarchies and language
ideologies influence the decisions made during
the development and deployment lifecycle?
What kinds of NLP systems do these decisions
result in, and what kinds do they foreclose?
General assumptions: To which linguistic
norms do NLP systems adhere (Bender,
2019; Ruane et al., 2019)? Which language
practices are implicitly assumed to be
standard, ordinary, correct, or appropriate?
Task definition: For which speakers
are NLP systems (and NLP resources)
developed? (See Joshi et al. (2020) for
a discussion.) How do task definitions
discretize the world? For example, how
are social groups delineated when defining
demographic attribute prediction tasks
(e.g., Koppel et al., 2002; Rosenthal and
McKeown, 2011; Nguyen et al., 2013)?
What about languages in native language
prediction tasks (Tetreault et al., 2013)?
Data: How are datasets collected, preprocessed, and labeled or annotated? What are
the impacts of annotation guidelines, annotator assumptions and perceptions (Olteanu | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 24 | Data: How are datasets collected, preprocessed, and labeled or annotated? What are
the impacts of annotation guidelines, annotator assumptions and perceptions (Olteanu
et al., 2019; Sap et al., 2019; Geiger et al.,
2020), and annotation aggregation processes (Pavlick and Kwiatkowski, 2019)?
Evaluation: How are NLP systems evaluated? What are the impacts of evaluation
metrics (Olteanu et al., 2017)? Are any
non-quantitative evaluations performed?
.How do NLP systems reproduce or transform
language ideologies? Which language varieties
or practices come to be deemed good or bad?
Might “good” language simply mean language
that is easily handled by existing NLP systems? For example, linguistic phenomena arising from many language practices (Eisenstein,
2013) are described as “noisy text” and often
viewed as a target for “normalization.” How
do the language ideologies that are reproduced
by NLP systems maintain social hierarchies?
.Which representational harms are being
measured or mitigated? Are these the most
normatively concerning harms, or merely
those that are well handled by existing algorithmic fairness techniques? Are there other
representational harms that might be analyzed?
4.2 Conceptualizations of “bias” | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 25 | those that are well handled by existing algorithmic fairness techniques? Are there other
representational harms that might be analyzed?
4.2 Conceptualizations of “bias”
Turning now to (R2) , we argue that work analyzing
“bias” in NLP systems should provide explicit
statements of why the system behaviors that are
described as “bias” are harmful, in what ways,
and to whom, as well as the normative reasoning
underlying these statements. In other words,
researchers and practitioners should articulate their
conceptualizations of “bias.” As we described
above, papers often contain descriptions of system
behaviors that are understood to be self-evident
statements of “bias.” This use of imprecise
terminology has led to papers all claiming to
analyze “bias” in NLP systems, sometimes even
in systems developed for the same task, but with
different or even inconsistent conceptualizations of
“bias,” and no explanations for these differences.
Yet analyzing “bias” is an inherently normative
process—in which some system behaviors are
deemed good and others harmful—even if assumptions about what kinds of system behaviors are
harmful, in what ways, for whom, and why are | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 26 | process—in which some system behaviors are
deemed good and others harmful—even if assumptions about what kinds of system behaviors are
harmful, in what ways, for whom, and why are
not stated. We therefore echo calls by Bardzell and
Bardzell (2011), Keyes et al. (2019), and Green
(2019) for researchers and practitioners to make
their normative reasoning explicit by articulating
the social values that underpin their decisions to
deem some system behaviors as harmful, no matter
how obvious such values appear to be. We further
argue that this reasoning should take into account
the relationships between language and social
hierarchies that we described above. First, these
relationships provide a foundation from which to
approach the normative reasoning that we recommend making explicit. For example, some system
behaviors might be harmful precisely because
they maintain social hierarchies. Second, if work
analyzing “bias” in NLP systems is reoriented
to understand how social hierarchies, language
ideologies, and NLP systems are coproduced, then
this work will be incomplete if we fail to account
for the ways that social hierarchies and language
ideologies determine what we mean by “bias” in
the first place. As a starting point, we therefore
provide the following concrete research questions: | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 27 | ideologies determine what we mean by “bias” in
the first place. As a starting point, we therefore
provide the following concrete research questions:
.What kinds of system behaviors are described
as “bias”? What are their potential sources (e.g.,
general assumptions, task definition, data)?
.In what ways are these system behaviors harmful, to whom are they harmful, and why?
.What are the social values (obvious or not) thatunderpin this conceptualization of “bias?”
4.3 Language use in practice
Finally, we turn to (R3). Our perspective, which
rests on a greater recognition of the relationships
between language and social hierarchies, suggests
several directions for examining language use in
practice. Here, we focus on two. First, because language is necessarily situated, and because different
social groups have different lived experiences due
to their different social positions (Hanna et al.,
2020)—particularly groups at the intersections
of multiple axes of oppression—we recommend
that researchers and practitioners center work
analyzing “bias” in NLP systems around the lived
experiences of members of communities affected
by these systems. Second, we recommend that
the power relations between technologists and
such communities be interrogated and reimagined. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 28 | experiences of members of communities affected
by these systems. Second, we recommend that
the power relations between technologists and
such communities be interrogated and reimagined.
Researchers have pointed out that algorithmic
fairness techniques, by proposing incremental
technical mitigations—e.g., collecting new datasets
or training better models—maintain these power
relations by (a) assuming that automated systems
should continue to exist, rather than asking
whether they should be built at all, and (b) keeping
development and deployment decisions in the
hands of technologists (Bennett and Keyes, 2019;
Cifor et al., 2019; Green, 2019; Katell et al., 2020).
There are many disciplines for researchers and
practitioners to draw on when pursuing these
directions. For example, in human–computer
interaction, Hamidi et al. (2018) study transgender
people’s experiences with automated gender
recognition systems in order to uncover how
these systems reproduce structures of transgender
exclusion by redefining what it means to perform
gender “normally.” Value-sensitive design provides
a framework for accounting for the values of different stakeholders in the design of technology (e.g.,
Friedman et al., 2006; Friedman and Hendry, 2019;
Le Dantec et al., 2009; Yoo et al., 2019), while | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 29 | Friedman et al., 2006; Friedman and Hendry, 2019;
Le Dantec et al., 2009; Yoo et al., 2019), while
participatory design seeks to involve stakeholders
in the design process itself (Sanders, 2002; Muller,
2007; Simonsen and Robertson, 2013; DiSalvo
et al., 2013). Participatory action research in education (Kemmis, 2006) and in language documentation and reclamation (Junker, 2018) is also relevant.
In particular, work on language reclamation to
support decolonization and tribal sovereignty
(Leonard, 2012) and work in sociolinguistics focusing on developing co-equal research relationships
with community members and supporting linguistic justice efforts (e.g., Bucholtz et al., 2014, 2016,
2019) provide examples of more emancipatory relationships with communities. Finally, several workshops and events have begun to explore how to empower stakeholders in the development and deployment of technology (Vaccaro et al., 2019; Givens
and Morris, 2020; Sassaman et al., 2020)4and how
to help researchers and practitioners consider when
not to build systems at all (Barocas et al., 2020).
As a starting point for engaging with communities affected by NLP systems, we therefore
provide the following concrete research questions:
.How do communities become aware of NLP | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 30 | As a starting point for engaging with communities affected by NLP systems, we therefore
provide the following concrete research questions:
.How do communities become aware of NLP
systems? Do they resist them, and if so, how?
.What additional costs are borne by communities for whom NLP systems do not work well?
.Do NLP systems shift power toward oppressive
institutions (e.g., by enabling predictions that
communities do not want made, linguistically
based unfair allocation of resources or opportunities (Rosa and Flores, 2017), surveillance,
or censorship), or away from such institutions?
.Who is involved in the development and
deployment of NLP systems? How do
decision-making processes maintain power relations between technologists and communities
affected by NLP systems? Can these processes be changed to reimagine these relations?
5 Case study
To illustrate our recommendations, we present a
case study covering work on African-American
English (AAE).5Work analyzing “bias” in the context of AAE has shown that part-of-speech taggers,
language identification systems, and dependency
parsers all work less well on text containing
features associated with AAE than on text without
these features (Jørgensen et al., 2015, 2016; Blodgett et al., 2016, 2018), and that toxicity detection | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 31 | features associated with AAE than on text without
these features (Jørgensen et al., 2015, 2016; Blodgett et al., 2016, 2018), and that toxicity detection
systems score tweets containing features associated
with AAE as more offensive than tweets without them (Davidson et al., 2019; Sap et al., 2019).
These papers have been critical for highlighting
AAE as a language variety for which existing NLP
4Alsohttps://participatoryml.github.io/
5This language variety has had many different names
over the years, but is now generally called AfricanAmerican English (AAE), African-American Vernacular English (AA VE), or African-American Language (AAL) (Green,
2002; Wolfram and Schilling, 2015; Rickford and King, 2016).systems may not work, illustrating their limitations.
However, they do not conceptualize “racial bias” in
the same way. The first four of these papers simply
focus on system performance differences between
text containing features associated with AAE and
text without these features. In contrast, the last
two papers also focus on such system performance
differences, but motivate this focus with the following additional reasoning: If tweets containing
features associated with AAE are scored as more
offensive than tweets without these features, then
this might (a) yield negative perceptions of AAE; | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 32 | features associated with AAE are scored as more
offensive than tweets without these features, then
this might (a) yield negative perceptions of AAE;
(b) result in disproportionate removal of tweets
containing these features, impeding participation
in online platforms and reducing the space available online in which speakers can use AAE freely;
and (c) cause AAE speakers to incur additional
costs if they have to change their language practices
to avoid negative perceptions or tweet removal.
More importantly, none of these papers engage
with the literature on AAE, racial hierarchies in the
U.S., and raciolinguistic ideologies. By failing to
engage with this literature—thereby treating AAE
simply as one of many non-Penn Treebank varieties of English or perhaps as another challenging
domain—work analyzing “bias” in NLP systems
in the context of AAE fails to situate these systems
in the world. Who are the speakers of AAE? How
are they viewed? We argue that AAE as a language
variety cannot be separated from its speakers—
primarily Black people in the U.S., who experience
systemic anti-Black racism—and the language ideologies that reinforce and justify racial hierarchies.
Even after decades of sociolinguistic efforts to
legitimize AAE, it continues to be viewed as “bad” | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 33 | Even after decades of sociolinguistic efforts to
legitimize AAE, it continues to be viewed as “bad”
English and its speakers continue to be viewed as
linguistically inadequate—a view called the deficit
perspective (Alim et al., 2016; Rosa and Flores,
2017). This perspective persists despite demonstrations that AAE is rule-bound and grammatical
(Mufwene et al., 1998; Green, 2002), in addition
to ample evidence of its speakers’ linguistic adroitness (e.g., Alim, 2004; Rickford and King, 2016).
This perspective belongs to a broader set of raciolinguistic ideologies (Rosa and Flores, 2017), which
also produce allocational harms; speakers of AAE
are frequently penalized for not adhering to dominant language practices, including in the education
system (Alim, 2004; Terry et al., 2010), when
seeking housing (Baugh, 2018), and in the judicial
system, where their testimony is misunderstood or,
worse yet, disbelieved (Rickford and King, 2016;
Jones et al., 2019). These raciolinguistic ideologies
position racialized communities as needing
linguistic intervention, such as language education
programs, in which these and other harms can be
reduced if communities accommodate to dominant language practices (Rosa and Flores, 2017). | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 34 | linguistic intervention, such as language education
programs, in which these and other harms can be
reduced if communities accommodate to dominant language practices (Rosa and Flores, 2017).
In the technology industry, speakers of AAE are
often not considered consumers who matter. For
example, Benjamin (2019) recounts an Apple employee who worked on speech recognition for Siri:
“As they worked on different English dialects —
Australian, Singaporean, and Indian English — [the
employee] asked his boss: ‘What about African
American English?’ To this his boss responded:
‘Well, Apple products are for the premium market. ”’
The reality, of course, is that speakers of AAE tend
not to represent the “premium market” precisely because of institutions and policies that help to maintain racial hierarchies by systematically denying
them the opportunities to develop wealth that are
available to white Americans (Rothstein, 2017)—
an exclusion that is reproduced in technology by
countless decisions like the one described above.
Engaging with the literature outlined above
situates the system behaviors that are described
as “bias,” providing a foundation for normative
reasoning. Researchers and practitioners should
be concerned about “racial bias” in toxicity
detection systems not only because performance
differences impair system performance, but | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 35 | reasoning. Researchers and practitioners should
be concerned about “racial bias” in toxicity
detection systems not only because performance
differences impair system performance, but
because they reproduce longstanding injustices of
stigmatization and disenfranchisement for speakers
of AAE. In re-stigmatizing AAE, they reproduce
language ideologies in which AAE is viewed as
ungrammatical, uneducated, and offensive. These
ideologies, in turn, enable linguistic discrimination
and justify enduring racial hierarchies (Rosa and
Flores, 2017). Our perspective, which understands
racial hierarchies and raciolinguistic ideologies as
structural conditions that govern the development
and deployment of technology, implies that
techniques for measuring or mitigating “bias”
in NLP systems will necessarily be incomplete
unless they interrogate and dismantle these
structural conditions, including the power relations
between technologists and racialized communities.
We emphasize that engaging with the literature
on AAE, racial hierarchies in the U.S., and
raciolinguistic ideologies can generate new lines of
engagement. These lines include work on the ways
that the decisions made during the developmentand deployment of NLP systems produce stigmatization and disenfranchisement, and work on AAE
use in practice, such as the ways that speakers
of AAE interact with NLP systems that were not | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 36 | use in practice, such as the ways that speakers
of AAE interact with NLP systems that were not
designed for them. This literature can also help researchers and practitioners address the allocational
harms that may be produced by NLP systems, and
ensure that even well-intentioned NLP systems
do not position racialized communities as needing
linguistic intervention or accommodation to
dominant language practices. Finally, researchers
and practitioners wishing to design better systems
can also draw on a growing body of work on
anti-racist language pedagogy that challenges the
deficit perspective of AAE and other racialized
language practices (e.g. Flores and Chaparro, 2018;
Baker-Bell, 2019; Martínez and Mejía, 2019), as
well as the work that we described in section 4.3
on reimagining the power relations between technologists and communities affected by technology.
6 Conclusion
By surveying 146 papers analyzing “bias” in NLP
systems, we found that (a) their motivations are
often vague, inconsistent, and lacking in normative reasoning; and (b) their proposed quantitative
techniques for measuring or mitigating “bias” are
poorly matched to their motivations and do not engage with the relevant literature outside of NLP.
To help researchers and practitioners avoid these | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 37 | poorly matched to their motivations and do not engage with the relevant literature outside of NLP.
To help researchers and practitioners avoid these
pitfalls, we proposed three recommendations that
should guide work analyzing “bias” in NLP systems, and, for each, provided several concrete research questions. These recommendations rest on
a greater recognition of the relationships between
language and social hierarchies—a step that we
see as paramount to establishing a path forward.
Acknowledgments
This paper is based upon work supported by the
National Science Foundation Graduate Research
Fellowship under Grant No. 1451512. Any opinion, findings, and conclusions or recommendations
expressed in this material are those of the authors
and do not necessarily reflect the views of the National Science Foundation. We thank the reviewers
for their useful feedback, especially the suggestion to include additional details about our method.
References
Artem Abzaliev. 2019. On GAP coreference resolution shared task: insights from the 3rd place solution.
InProceedings of the Workshop on Gender Bias in
Natural Language Processing , pages 107–112, Florence, Italy.
ADA. 2018. Guidelines for Writing About People With Disabilities. ADA National Network.
https://bit.ly/2KREbkB.
Oshin Agarwal, Funda Durupinar, Norman I. Badler, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 38 | https://bit.ly/2KREbkB.
Oshin Agarwal, Funda Durupinar, Norman I. Badler,
and Ani Nenkova. 2019. Word embeddings (also)
encode human personality stereotypes. In Proceedings of the Joint Conference on Lexical and Computational Semantics , pages 205–211, Minneapolis,
MN.
H. Samy Alim. 2004. You Know My Steez: An Ethnographic and Sociolinguistic Study of Styleshifting in
a Black American Speech Community . American Dialect Society.
H. Samy Alim, John R. Rickford, and Arnetha F. Ball,
editors. 2016. Raciolinguistics: How Language
Shapes Our Ideas About Race . Oxford University
Press.
Sandeep Attree. 2019. Gendered ambiguous pronouns
shared task: Boosting model confidence by evidence
pooling. In Proceedings of the Workshop on Gender Bias in Natural Language Processing , Florence,
Italy.
Pinkesh Badjatiya, Manish Gupta, and Vasudeva
Varma. 2019. Stereotypical bias removal for hate
speech detection task using knowledge-based generalizations. In Proceedings of the International
World Wide Web Conference , pages 49–59, San Francisco, CA. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 39 | speech detection task using knowledge-based generalizations. In Proceedings of the International
World Wide Web Conference , pages 49–59, San Francisco, CA.
Eugene Bagdasaryan, Omid Poursaeed, and Vitaly
Shmatikov. 2019. Differential Privacy Has Disparate Impact on Model Accuracy. In Proceedings
of the Conference on Neural Information Processing
Systems , Vancouver, Canada.
April Baker-Bell. 2019. Dismantling anti-black linguistic racism in English language arts classrooms:
Toward an anti-racist black language pedagogy. Theory Into Practice .
David Bamman, Sejal Popat, and Sheng Shen. 2019.
An annotated dataset of literary entities. In Proceedings of the North American Association for Computational Linguistics (NAACL) , pages 2138–2144,
Minneapolis, MN.
Xingce Bao and Qianqian Qiao. 2019. Transfer Learning from Pre-trained BERT for Pronoun Resolution.
InProceedings of the Workshop on Gender Bias
in Natural Language Processing , pages 82–88, Florence, Italy.Shaowen Bardzell and Jeffrey Bardzell. 2011. Towards
a Feminist HCI Methodology: Social Science, Feminism, and HCI. In Proceedings of the Conference on
Human Factors in Computing Systems (CHI) , pages | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 40 | a Feminist HCI Methodology: Social Science, Feminism, and HCI. In Proceedings of the Conference on
Human Factors in Computing Systems (CHI) , pages
675–684, Vancouver, Canada.
Solon Barocas, Asia J. Biega, Benjamin Fish, J˛ edrzej
Niklas, and Luke Stark. 2020. When Not to Design, Build, or Deploy. In Proceedings of the Conference on Fairness, Accountability, and Transparency ,
Barcelona, Spain.
Solon Barocas, Kate Crawford, Aaron Shapiro, and
Hanna Wallach. 2017. The Problem With Bias: Allocative Versus Representational Harms in Machine
Learning. In Proceedings of SIGCIS , Philadelphia,
PA.
Christine Basta, Marta R. Costa-jussà, and Noe Casas.
2019. Evaluating the underlying gender bias in contextualized word embeddings. In Proceedings of
the Workshop on Gender Bias for Natural Language
Processing , pages 33–39, Florence, Italy.
John Baugh. 2018. Linguistics in Pursuit of Justice .
Cambridge University Press.
Emily M. Bender. 2019. A typology of ethical risks
in language technology with an eye towards where
transparent documentation can help. Presented at
The Future of Artificial Intelligence: Language, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 41 | in language technology with an eye towards where
transparent documentation can help. Presented at
The Future of Artificial Intelligence: Language,
Ethics, Technology Workshop. https://bit.ly/
2P9t9M6 .
Ruha Benjamin. 2019. Race After Technology: Abolitionist Tools for the New Jim Code . John Wiley &
Sons.
Ruha Benjamin. 2020. 2020 Vision: Reimagining the
Default Settings of Technology & Society. Keynote
at ICLR.
Cynthia L. Bennett and Os Keyes. 2019. What is the
Point of Fairness? Disability, AI, and The Complexity of Justice. In Proceedings of the ASSETS
Workshop on AI Fairness for People with Disabilities, Pittsburgh, PA.
Camiel J. Beukeboom and Christian Burgers. 2019.
How Stereotypes Are Shared Through Language: A
Review and Introduction of the Social Categories
and Stereotypes Communication (SCSC) Framework. Review of Communication Research , 7:1–37.
Shruti Bhargava and David Forsyth. 2019. Exposing and Correcting the Gender Bias in Image
Captioning Datasets and Models. arXiv preprint
arXiv:1912.00578 .
Jayadev Bhaskaran and Isha Bhallamudi. 2019. Good | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 42 | arXiv:1912.00578 .
Jayadev Bhaskaran and Isha Bhallamudi. 2019. Good
Secretaries, Bad Truck Drivers? Occupational Gender Stereotypes in Sentiment Analysis. In Proceedings of the Workshop on Gender Bias in Natural Language Processing , pages 62–68, Florence, Italy.
Su Lin Blodgett, Lisa Green, and Brendan O’Connor.
2016. Demographic Dialectal Variation in Social
Media: A Case Study of African-American English.
InProceedings of Empirical Methods in Natural
Language Processing (EMNLP) , pages 1119–1130,
Austin, TX.
Su Lin Blodgett and Brendan O’Connor. 2017. Racial
Disparity in Natural Language Processing: A Case
Study of Social Media African-American English.
InProceedings of the Workshop on Fairness, Accountability, and Transparency in Machine Learning
(FAT/ML) , Halifax, Canada.
Su Lin Blodgett, Johnny Wei, and Brendan O’Connor.
2018. Twitter Universal Dependency Parsing for
African-American and Mainstream American English. In Proceedings of the Association for Computational Linguistics (ACL) , pages 1415–1425, Melbourne, Australia.
Tolga Bolukbasi, Kai-Wei Chang, James Zou, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 43 | Tolga Bolukbasi, Kai-Wei Chang, James Zou,
Venkatesh Saligrama, and Adam Kalai. 2016a.
Man is to Computer Programmer as Woman is to
Homemaker? Debiasing Word Embeddings. In Proceedings of the Conference on Neural Information
Processing Systems , pages 4349–4357, Barcelona,
Spain.
Tolga Bolukbasi, Kai-Wei Chang, James Zou,
Venkatesh Saligrama, and Adam Kalai. 2016b.
Quantifying and reducing stereotypes in word
embeddings. In Proceedings of the ICML Workshop
on #Data4Good: Machine Learning in Social Good
Applications , pages 41–45, New York, NY .
Shikha Bordia and Samuel R. Bowman. 2019. Identifying and reducing gender bias in word-level language
models. In Proceedings of the NAACL Student Research Workshop , pages 7–15, Minneapolis, MN.
Marc-Etienne Brunet, Colleen Alkalay-Houlihan, Ashton Anderson, and Richard Zemel. 2019. Understanding the Origins of Bias in Word Embeddings.
InProceedings of the International Conference on
Machine Learning , pages 803–811, Long Beach,
CA.
Mary Bucholtz, Dolores Inés Casillas, and Jin Sook | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 44 | Machine Learning , pages 803–811, Long Beach,
CA.
Mary Bucholtz, Dolores Inés Casillas, and Jin Sook
Lee. 2016. Beyond Empowerment: Accompaniment and Sociolinguistic Justice in a Youth Research
Program. In Robert Lawson and Dave Sayers, editors, Sociolinguistic Research: Application and Impact, pages 25–44. Routledge.
Mary Bucholtz, Dolores Inés Casillas, and Jin Sook
Lee. 2019. California Latinx Youth as Agents of
Sociolinguistic Justice. In Netta Avineri, Laura R.
Graham, Eric J. Johnson, Robin Conley Riner, and
Jonathan Rosa, editors, Language and Social Justice
in Practice , pages 166–175. Routledge.
Mary Bucholtz, Audrey Lopez, Allina Mojarro, Elena
Skapoulli, Chris VanderStouwe, and Shawn WarnerGarcia. 2014. Sociolinguistic Justice in the Schools:Student Researchers as Linguistic Experts. Language and Linguistics Compass , 8:144–157.
Kaylee Burns, Lisa Anne Hendricks, Kate Saenko,
Trevor Darrell, and Anna Rohrbach. 2018. Women
also Snowboard: Overcoming Bias in Captioning
Models. In Procedings of the European Conference | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 45 | Trevor Darrell, and Anna Rohrbach. 2018. Women
also Snowboard: Overcoming Bias in Captioning
Models. In Procedings of the European Conference
on Computer Vision (ECCV) , pages 793–811, Munich, Germany.
Aylin Caliskan, Joanna J. Bryson, and Arvind
Narayanan. 2017. Semantics derived automatically
from language corpora contain human-like biases.
Science , 356(6334).
Kathryn Campbell-Kibler. 2009. The nature of sociolinguistic perception. Language Variation and
Change , 21(1):135–156.
Yang Trista Cao and Hal Daumé, III. 2019. Toward gender-inclusive coreference resolution. arXiv
preprint arXiv:1910.13913 .
Rakesh Chada. 2019. Gendered pronoun resolution using bert and an extractive question answering formulation. In Proceedings of the Workshop on Gender
Bias in Natural Language Processing , pages 126–
133, Florence, Italy.
Kaytlin Chaloner and Alfredo Maldonado. 2019. Measuring Gender Bias in Word Embedding across Domains and Discovering New Gender Bias Word Categories. In Proceedings of the Workshop on Gender
Bias in Natural Language Processing , pages 25–32,
Florence, Italy. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 46 | Bias in Natural Language Processing , pages 25–32,
Florence, Italy.
Anne H. Charity Hudley. 2017. Language and Racialization. In Ofelia García, Nelson Flores, and Massimiliano Spotti, editors, The Oxford Handbook of
Language and Society . Oxford University Press.
Won Ik Cho, Ji Won Kim, Seok Min Kim, and
Nam Soo Kim. 2019. On measuring gender bias in
translation of gender-neutral pronouns. In Proceedings of the Workshop on Gender Bias in Natural Language Processing , pages 173–181, Florence, Italy.
Shivang Chopra, Ramit Sawhney, Puneet Mathur, and
Rajiv Ratn Shah. 2020. Hindi-English Hate Speech
Detection: Author Profiling, Debiasing, and Practical Perspectives. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI) , New York,
NY .
Marika Cifor, Patricia Garcia, T.L. Cowan, Jasmine
Rault, Tonia Sutherland, Anita Say Chan, Jennifer
Rode, Anna Lauren Hoffmann, Niloufar Salehi, and
Lisa Nakamura. 2019. Feminist Data ManifestNo. Retrieved from https://www.manifestno.
com/ .
Patricia Hill Collins. 2000. Black Feminist Thought: | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 47 | Lisa Nakamura. 2019. Feminist Data ManifestNo. Retrieved from https://www.manifestno.
com/ .
Patricia Hill Collins. 2000. Black Feminist Thought:
Knowledge, Consciousness, and the Politics of Empowerment . Routledge.
Justin T. Craft, Kelly E. Wright, Rachel Elizabeth
Weissler, and Robin M. Queen. 2020. Language
and Discrimination: Generating Meaning, Perceiving Identities, and Discriminating Outcomes. Annual Review of Linguistics , 6(1).
Kate Crawford. 2017. The Trouble with Bias. Keynote
at NeurIPS.
Kimberle Crenshaw. 1989. Demarginalizing the Intersection of Race and Sex: A Black Feminist Critique
of Antidiscrmination Doctrine, Feminist Theory and
Antiracist Politics. University of Chicago Legal Forum.
Amanda Cercas Curry and Verena Rieser. 2018.
#MeToo: How Conversational Systems Respond to
Sexual Harassment. In Proceedings of the Workshop
on Ethics in Natural Language Processing , pages 7–
14, New Orleans, LA.
Karan Dabas, Nishtha Madaan, Gautam Singh, Vijay Arya, Sameep Mehta, and Tanmoy Chakraborty.
2020. Fair Transfer of Multiple Style Attributes in | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 48 | 2020. Fair Transfer of Multiple Style Attributes in
Text. arXiv preprint arXiv:2001.06693 .
Thomas Davidson, Debasmita Bhattacharya, and Ingmar Weber. 2019. Racial bias in hate speech and
abusive language detection datasets. In Proceedings
of the Workshop on Abusive Language Online , pages
25–35, Florence, Italy.
Maria De-Arteaga, Alexey Romanov, Hanna Wallach, Jennifer Chayes, Christian Borgs, Alexandra
Chouldechova, Sahin Geyik, Krishnaram Kenthapadi, and Adam Tauman Kalai. 2019. Bias in bios:
A case study of semantic representation bias in a
high-stakes setting. In Proceedings of the Conference on Fairness, Accountability, and Transparency ,
pages 120–128, Atlanta, GA.
Sunipa Dev, Tao Li, Jeff Phillips, and Vivek Srikumar. 2019. On Measuring and Mitigating Biased
Inferences of Word Embeddings. arXiv preprint
arXiv:1908.09369 .
Sunipa Dev and Jeff Phillips. 2019. Attenuating Bias in
Word Vectors. In Proceedings of the International
Conference on Artificial Intelligence and Statistics ,
pages 879–887, Naha, Japan. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 49 | Word Vectors. In Proceedings of the International
Conference on Artificial Intelligence and Statistics ,
pages 879–887, Naha, Japan.
Mark Díaz, Isaac Johnson, Amanda Lazar, Anne Marie
Piper, and Darren Gergle. 2018. Addressing agerelated bias in sentiment analysis. In Proceedings
of the Conference on Human Factors in Computing
Systems (CHI) , Montréal, Canada.
Emily Dinan, Angela Fan, Adina Williams, Jack Urbanek, Douwe Kiela, and Jason Weston. 2019.
Queens are Powerful too: Mitigating Gender
Bias in Dialogue Generation. arXiv preprint
arXiv:1911.03842 .
Carl DiSalvo, Andrew Clement, and V olkmar Pipek.
2013. Communities: Participatory Design for, with
and by communities. In Jesper Simonsen and ToniRobertson, editors, Routledge International Handbook of Participatory Design , pages 182–209. Routledge.
Lucas Dixon, John Li, Jeffrey Sorensen, Nithum Thain,
and Lucy Vasserman. 2018. Measuring and mitigating unintended bias in text classification. In Proceedings of the Conference on Artificial Intelligence,
Ethics, and Society (AIES) , New Orleans, LA. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 50 | Ethics, and Society (AIES) , New Orleans, LA.
Jacob Eisenstein. 2013. What to do about bad language on the Internet. In Proceedings of the North
American Association for Computational Linguistics
(NAACL) , pages 359–369.
Kawin Ethayarajh. 2020. Is Your Classifier Actually
Biased? Measuring Fairness under Uncertainty with
Bernstein Bounds. In Proceedings of the Association for Computational Linguistics (ACL) .
Kawin Ethayarajh, David Duvenaud, and Graeme Hirst.
2019. Understanding Undesirable Word Embedding
Assocations. In Proceedings of the Association for
Computational Linguistics (ACL) , pages 1696–1705,
Florence, Italy.
Joseph Fisher. 2019. Measuring social bias in
knowledge graph embeddings. arXiv preprint
arXiv:1912.02761 .
Nelson Flores and Sofia Chaparro. 2018. What counts
as language education policy? Developing a materialist Anti-racist approach to language activism. Language Policy , 17(3):365–384.
Omar U. Florez. 2019. On the Unintended Social Bias
of Training Language Generation Models with Data | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 51 | Omar U. Florez. 2019. On the Unintended Social Bias
of Training Language Generation Models with Data
from Local Media. In Proceedings of the NeurIPS
Workshop on Human-Centric Machine Learning ,
Vancouver, Canada.
Joel Escudé Font and Marta R. Costa-jussà. 2019.
Equalizing gender biases in neural machine translation with word embeddings techniques. In Proceedings of the Workshop on Gender Bias for Natural Language Processing , pages 147–154, Florence,
Italy.
Batya Friedman and David G. Hendry. 2019. Value
Sensitive Design: Shaping Technology with Moral
Imagination . MIT Press.
Batya Friedman, Peter H. Kahn Jr., and Alan Borning.
2006. Value Sensitive Design and Information Systems. In Dennis Galletta and Ping Zhang, editors,
Human-Computer Interaction in Management Information Systems: Foundations , pages 348–372. M.E.
Sharpe.
Nikhil Garg, Londa Schiebinger, Dan Jurafsky, and
James Zou. 2018. Word Embeddings Quantify 100
Years of Gender and Ethnic Stereotypes. Proceedings of the National Academy of Sciences , 115(16).
Sahaj Garg, Vincent Perot, Nicole Limtiaco, Ankur | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 52 | Years of Gender and Ethnic Stereotypes. Proceedings of the National Academy of Sciences , 115(16).
Sahaj Garg, Vincent Perot, Nicole Limtiaco, Ankur
Taly, Ed H. Chi, and Alex Beutel. 2019. Counterfactual fairness in text classification through robustness. In Proceedings of the Conference on Artificial
Intelligence, Ethics, and Society (AIES) , Honolulu,
HI.
Aparna Garimella, Carmen Banea, Dirk Hovy, and
Rada Mihalcea. 2019. Women’s syntactic resilience
and men’s grammatical luck: Gender bias in part-ofspeech tagging and dependency parsing data. In Proceedings of the Association for Computational Linguistics (ACL) , pages 3493–3498, Florence, Italy.
Andrew Gaut, Tony Sun, Shirlyn Tang, Yuxin Huang,
Jing Qian, Mai ElSherief, Jieyu Zhao, Diba
Mirza, Elizabeth Belding, Kai-Wei Chang, and
William Yang Wang. 2020. Towards Understanding Gender Bias in Relation Extraction. In Proceedings of the Association for Computational Linguistics (ACL) .
R. Stuart Geiger, Kevin Yu, Yanlai Yang, Mindy Dai, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 53 | R. Stuart Geiger, Kevin Yu, Yanlai Yang, Mindy Dai,
Jie Qiu, Rebekah Tang, and Jenny Huang. 2020.
Garbage In, Garbage Out? Do Machine Learning Application Papers in Social Computing Report
Where Human-Labeled Training Data Comes From?
InProceedings of the Conference on Fairness, Accountability, and Transparency , pages 325–336.
Oguzhan Gencoglu. 2020. Cyberbullying Detection with Fairness Constraints. arXiv preprint
arXiv:2005.06625 .
Alexandra Reeve Givens and Meredith Ringel Morris.
2020. Centering Disability Perspecives in Algorithmic Fairness, Accountability, and Transparency. In
Proceedings of the Conference on Fairness, Accountability, and Transparency , Barcelona, Spain.
Hila Gonen and Yoav Goldberg. 2019. Lipstick on a
Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove
Them. In Proceedings of the North American Association for Computational Linguistics (NAACL) ,
pages 609–614, Minneapolis, MN.
Hila Gonen and Kellie Webster. 2020. Automatically Identifying Gender Issues in Machine
Translation using Perturbations. arXiv preprint
arXiv:2004.14065 . | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 54 | Translation using Perturbations. arXiv preprint
arXiv:2004.14065 .
Ben Green. 2019. “Good” isn’t good enough. In Proceedings of the AI for Social Good Workshop , Vancouver, Canada.
Lisa J. Green. 2002. African American English: A Linguistic Introduction . Cambridge University Press.
Anthony G. Greenwald, Debbie E. McGhee, and Jordan L.K. Schwartz. 1998. Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74(6):1464–1480.Enoch Opanin Gyamfi, Yunbo Rao, Miao Gou, and
Yanhua Shao. 2020. deb2viz: Debiasing gender in
word embedding data using subspace visualization.
InProceedings of the International Conference on
Graphics and Image Processing .
Foad Hamidi, Morgan Klaus Scheuerman, and Stacy M.
Branham. 2018. Gender Recognition or Gender Reductionism? The Social Implications of Automatic
Gender Recognition Systems. In Proceedings of the
Conference on Human Factors in Computing Systems (CHI) , Montréal, Canada.
Alex Hanna, Emily Denton, Andrew Smart, and Jamila
Smith-Loud. 2020. Towards a Critical Race Methodology in Algorithmic Fairness. In Proceedings of the | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 55 | Alex Hanna, Emily Denton, Andrew Smart, and Jamila
Smith-Loud. 2020. Towards a Critical Race Methodology in Algorithmic Fairness. In Proceedings of the
Conference on Fairness, Accountability, and Transparency , pages 501–512, Barcelona, Spain.
Madeline E. Heilman, Aaaron S. Wallen, Daniella
Fuchs, and Melinda M. Tamkins. 2004. Penalties
for Success: Reactions to Women Who Succeed at
Male Gender-Typed Tasks. Journal of Applied Psychology , 89(3):416–427.
Jane H. Hill. 2008. The Everyday Language of White
Racism . Wiley-Blackwell.
Dirk Hovy, Federico Bianchi, and Tommaso Fornaciari.
2020. Can You Translate that into Man? Commercial Machine Translation Systems Include Stylistic
Biases. In Proceedings of the Association for Computational Linguistics (ACL) .
Dirk Hovy and Anders Søgaard. 2015. Tagging Performance Correlates with Author Age. In Proceedings of the Association for Computational Linguistics and the International Joint Conference on Natural Language Processing , pages 483–488, Beijing,
China.
Dirk Hovy and Shannon L. Spruit. 2016. The social | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 56 | China.
Dirk Hovy and Shannon L. Spruit. 2016. The social
impact of natural language processing. In Proceedings of the Association for Computational Linguistics (ACL) , pages 591–598, Berlin, Germany.
Po-Sen Huang, Huan Zhang, Ray Jiang, Robert
Stanforth, Johannes Welbl, Jack W. Rae, Vishal
Maini, Dani Yogatama, and Pushmeet Kohli. 2019.
Reducing Sentiment Bias in Language Models
via Counterfactual Evaluation. arXiv preprint
arXiv:1911.03064 .
Xiaolei Huang, Linzi Xing, Franck Dernoncourt, and
Michael J. Paul. 2020. Multilingual Twitter Corpus and Baselines for Evaluating Demographic Bias
in Hate Speech Recognition. In Proceedings of
the Language Resources and Evaluation Conference
(LREC) , Marseille, France.
Christoph Hube, Maximilian Idahl, and Besnik Fetahu.
2020. Debiasing Word Embeddings from Sentiment
Associations in Names. In Proceedings of the International Conference on Web Search and Data Mining, pages 259–267, Houston, TX.
Ben Hutchinson, Vinodkumar Prabhakaran, Emily | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 57 | Ben Hutchinson, Vinodkumar Prabhakaran, Emily
Denton, Kellie Webster, Yu Zhong, and Stephen Denuyl. 2020. Social Biases in NLP Models as Barriers
for Persons with Disabilities. In Proceedings of the
Association for Computational Linguistics (ACL) .
Matei Ionita, Yury Kashnitsky, Ken Krige, Vladimir
Larin, Dennis Logvinenko, and Atanas Atanasov.
2019. Resolving gendered ambiguous pronouns
with BERT. In Proceedings of the Workshop on
Gender Bias in Natural Language Processing , pages
113–119, Florence, Italy.
Hailey James-Sorenson and David Alvarez-Melis.
2019. Probabilistic Bias Mitigation in Word Embeddings. In Proceedings of the Workshop on HumanCentric Machine Learning , Vancouver, Canada.
Shengyu Jia, Tao Meng, Jieyu Zhao, and Kai-Wei
Chang. 2020. Mitigating Gender Bias Amplification
in Distribution by Posterior Regularization. In Proceedings of the Association for Computational Linguistics (ACL) .
Taylor Jones, Jessica Rose Kalbfeld, Ryan Hancock,
and Robin Clark. 2019. Testifying while black: An
experimental study of court reporter accuracy in transcription of African American English. Language ,
95(2). | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 58 | and Robin Clark. 2019. Testifying while black: An
experimental study of court reporter accuracy in transcription of African American English. Language ,
95(2).
Anna Jørgensen, Dirk Hovy, and Anders Søgaard. 2015.
Challenges of studying and processing dialects in
social media. In Proceedings of the Workshop on
Noisy User-Generated Text , pages 9–18, Beijing,
China.
Anna Jørgensen, Dirk Hovy, and Anders Søgaard. 2016.
Learning a POS tagger for AA VE-like language. In
Proceedings of the North American Association for
Computational Linguistics (NAACL) , pages 1115–
1120, San Diego, CA.
Pratik Joshi, Sebastian Santy, Amar Budhiraja, Kalika
Bali, and Monojit Choudhury. 2020. The State and
Fate of Linguistic Diversity and Inclusion in the
NLP World. In Proceedings of the Association for
Computational Linguistics (ACL) .
Jaap Jumelet, Willem Zuidema, and Dieuwke Hupkes.
2019. Analysing Neural Language Models: Contextual Decomposition Reveals Default Reasoning in
Number and Gender Assignment. In Proceedings
of the Conference on Natural Language Learning , | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 59 | 2019. Analysing Neural Language Models: Contextual Decomposition Reveals Default Reasoning in
Number and Gender Assignment. In Proceedings
of the Conference on Natural Language Learning ,
Hong Kong, China.
Marie-Odile Junker. 2018. Participatory action research for Indigenous linguistics in the digital age.
In Shannon T. Bischoff and Carmen Jany, editors,
Insights from Practices in Community-Based Research , pages 164–175. De Gruyter Mouton.
David Jurgens, Yulia Tsvetkov, and Dan Jurafsky. 2017.
Incorporating Dialectal Variability for Socially Equitable Language Identification. In Proceedings of the
Association for Computational Linguistics (ACL) ,
pages 51–57, Vancouver, Canada.Masahiro Kaneko and Danushka Bollegala. 2019.
Gender-preserving debiasing for pre-trained word
embeddings. In Proceedings of the Association for
Computational Linguistics (ACL) , pages 1641–1650,
Florence, Italy.
Saket Karve, Lyle Ungar, and João Sedoc. 2019. Conceptor debiasing of word representations evaluated
on WEAT. In Proceedings of the Workshop on Gender Bias in Natural Language Processing , pages 40– | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 60 | on WEAT. In Proceedings of the Workshop on Gender Bias in Natural Language Processing , pages 40–
48, Florence, Italy.
Michael Katell, Meg Young, Dharma Dailey, Bernease
Herman, Vivian Guetler, Aaron Tam, Corinne Bintz,
Danielle Raz, and P.M. Krafft. 2020. Toward situated interventions for algorithmic equity: lessons
from the field. In Proceedings of the Conference on
Fairness, Accountability, and Transparency , pages
45–55, Barcelona, Spain.
Stephen Kemmis. 2006. Participatory action research
and the public sphere. Educational Action Research ,
14(4):459–476.
Os Keyes. 2018. The Misgendering Machines:
Trans/HCI Implications of Automatic Gender
Recognition. Proceedings of the ACM on HumanComputer Interaction , 2(CSCW).
Os Keyes, Josephine Hoy, and Margaret Drouhard.
2019. Human-Computer Insurrection: Notes on an
Anarchist HCI. In Proceedings of the Conference on
Human Factors in Computing Systems (CHI) , Glasgow, Scotland, UK.
Jae Yeon Kim, Carlos Ortiz, Sarah Nam, Sarah Santiago, and Vivek Datta. 2020. Intersectional Bias in
Hate Speech and Abusive Language Datasets. In | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 61 | Hate Speech and Abusive Language Datasets. In
Proceedings of the Association for Computational
Linguistics (ACL) .
Svetlana Kiritchenko and Saif M. Mohammad. 2018.
Examining Gender and Race Bias in Two Hundred
Sentiment Analysis Systems. In Proceedings of the
Joint Conference on Lexical and Computational Semantics , pages 43–53, New Orleans, LA.
Moshe Koppel, Shlomo Argamon, and Anat Rachel
Shimoni. 2002. Automatically Categorizing Written Texts by Author Gender. Literary and Linguistic
Computing , 17(4):401–412.
Keita Kurita, Nidhi Vyas, Ayush Pareek, Alan W.
Black, and Yulia Tsvetkov. 2019. Measuring bias
in contextualized word representations. In Proceedings of the Workshop on Gender Bias for Natural Language Processing , pages 166–172, Florence,
Italy.
Sonja L. Lanehart and Ayesha M. Malik. 2018. Black
Is, Black Isn’t: Perceptions of Language and Blackness. In Jeffrey Reaser, Eric Wilbanks, Karissa Wojcik, and Walt Wolfram, editors, Language Variety in
the New South . University of North Carolina Press. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 62 | the New South . University of North Carolina Press.
Brian N. Larson. 2017. Gender as a variable in naturallanguage processing: Ethical considerations. In Proceedings of the Workshop on Ethics in Natural Language Processing , pages 30–40, Valencia, Spain.
Anne Lauscher and Goran Glavaš. 2019. Are We Consistently Biased? Multidimensional Analysis of Biases in Distributional Word Vectors. In Proceedings
of the Joint Conference on Lexical and Computational Semantics , pages 85–91, Minneapolis, MN.
Anne Lauscher, Goran Glavaš, Simone Paolo Ponzetto,
and Ivan Vuli ´c. 2019. A General Framework for Implicit and Explicit Debiasing of Distributional Word
Vector Spaces. arXiv preprint arXiv:1909.06092 .
Christopher A. Le Dantec, Erika Shehan Poole, and Susan P. Wyche. 2009. Values as Lived Experience:
Evolving Value Sensitive Design in Support of Value
Discovery. In Proceedings of the Conference on Human Factors in Computing Systems (CHI) , Boston,
MA.
Nayeon Lee, Andrea Madotto, and Pascale Fung. 2019.
Exploring Social Bias in Chatbots using Stereotype
Knowledge. In Proceedings of the Workshop on | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 63 | Exploring Social Bias in Chatbots using Stereotype
Knowledge. In Proceedings of the Workshop on
Widening NLP , pages 177–180, Florence, Italy.
Wesley Y . Leonard. 2012. Reframing language reclamation programmes for everybody’s empowerment.
Gender and Language , 6(2):339–367.
Paul Pu Liang, Irene Li, Emily Zheng, Yao Chong Lim,
Ruslan Salakhutdinov, and Louis-Philippe Morency.
2019. Towards Debiasing Sentence Representations.
InProceedings of the NeurIPS Workshop on HumanCentric Machine Learning , Vancouver, Canada.
Rosina Lippi-Green. 2012. English with an Accent: Language, Ideology, and Discrimination in the
United States . Routledge.
Bo Liu. 2019. Anonymized BERT: An Augmentation
Approach to the Gendered Pronoun Resolution Challenge. In Proceedings of the Workshop on Gender
Bias in Natural Language Processing , pages 120–
125, Florence, Italy.
Haochen Liu, Jamell Dacon, Wenqi Fan, Hui Liu, Zitao Liu, and Jiliang Tang. 2019. Does Gender Matter? Towards Fairness in Dialogue Systems. arXiv
preprint arXiv:1910.10486 . | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 64 | preprint arXiv:1910.10486 .
Felipe Alfaro Lois, José A.R. Fonollosa, and Costa-jà.
2019. BERT Masked Language Modeling for Coreference Resolution. In Proceedings of the Workshop on Gender Bias in Natural Language Processing, pages 76–81, Florence, Italy.
Brandon C. Loudermilk. 2015. Implicit attitudes and
the perception of sociolinguistic variation. In Alexei
Prikhodkine and Dennis R. Preston, editors, Responses to Language Varieties: Variability, processes and outcomes , pages 137–156.Anastassia Loukina, Nitin Madnani, and Klaus Zechner. 2019. The many dimensions of algorithmic fairness in educational applications. In Proceedings of
the Workshop on Innovative Use of NLP for Building Educational Applications , pages 1–10, Florence,
Italy.
Kaiji Lu, Peter Mardziel, Fangjing Wu, Preetam Amancharla, and Anupam Datta. 2018. Gender bias in
neural natural language processing. arXiv preprint
arXiv:1807.11714 .
Anne Maass. 1999. Linguistic intergroup bias: Stereotype perpetuation through language. Advances in
Experimental Social Psychology , 31:79–121. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 65 | Anne Maass. 1999. Linguistic intergroup bias: Stereotype perpetuation through language. Advances in
Experimental Social Psychology , 31:79–121.
Nitin Madnani, Anastassia Loukina, Alina von Davier,
Jill Burstein, and Aoife Cahill. 2017. Building Better Open-Source Tools to Support Fairness in Automated Scoring. In Proceedings of the Workshop on
Ethics in Natural Language Processing , pages 41–
52, Valencia, Spain.
Thomas Manzini, Yao Chong Lim, Yulia Tsvetkov, and
Alan W. Black. 2019. Black is to Criminal as Caucasian is to Police: Detecting and Removing Multiclass Bias in Word Embeddings. In Proceedings of
the North American Association for Computational
Linguistics (NAACL) , pages 801–809, Minneapolis,
MN.
Ramón Antonio Martínez and Alexander Feliciano
Mejía. 2019. Looking closely and listening carefully: A sociocultural approach to understanding
the complexity of Latina/o/x students’ everyday language. Theory Into Practice .
Rowan Hall Maudslay, Hila Gonen, Ryan Cotterell, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 66 | the complexity of Latina/o/x students’ everyday language. Theory Into Practice .
Rowan Hall Maudslay, Hila Gonen, Ryan Cotterell,
and Simone Teufel. 2019. It’s All in the Name: Mitigating Gender Bias with Name-Based Counterfactual Data Substitution. In Proceedings of Empirical
Methods in Natural Language Processing (EMNLP) ,
pages 5270–5278, Hong Kong, China.
Chandler May, Alex Wang, Shikha Bordia, Samuel R.
Bowman, and Rachel Rudinger. 2019. On Measuring Social Biases in Sentence Encoders. In Proceedings of the North American Association for Computational Linguistics (NAACL) , pages 629–634, Minneapolis, MN.
Elijah Mayfield, Michael Madaio, Shrimai Prabhumoye, David Gerritsen, Brittany McLaughlin,
Ezekiel Dixon-Roman, and Alan W. Black. 2019.
Equity Beyond Bias in Language Technologies for
Education. In Proceedings of the Workshop on Innovative Use of NLP for Building Educational Applications , Florence, Italy.
Katherine McCurdy and O ˘guz Serbetçi. 2017. Grammatical gender associations outweigh topical gender | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 67 | Katherine McCurdy and O ˘guz Serbetçi. 2017. Grammatical gender associations outweigh topical gender
bias in crosslinguistic word embeddings. In Proceedings of the Workshop for Women & Underrepresented Minorities in Natural Language Processing ,
Vancouver, Canada.
Ninareh Mehrabi, Thamme Gowda, Fred Morstatter,
Nanyun Peng, and Aram Galstyan. 2019. Man is to
Person as Woman is to Location: Measuring Gender
Bias in Named Entity Recognition. arXiv preprint
arXiv:1910.10872 .
Michela Menegatti and Monica Rubini. 2017. Gender
bias and sexism in language. In Oxford Research
Encyclopedia of Communication . Oxford University
Press.
Inom Mirzaev, Anthony Schulte, Michael Conover, and
Sam Shah. 2019. Considerations for the interpretation of bias measures of word embeddings. arXiv
preprint arXiv:1906.08379 .
Salikoko S. Mufwene, Guy Bailey, and John R. Rickford, editors. 1998. African-American English:
Structure, History, and Use . Routledge.
Michael J. Muller. 2007. Participatory Design: The | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 68 | Structure, History, and Use . Routledge.
Michael J. Muller. 2007. Participatory Design: The
Third Space in HCI. In The Human-Computer Interaction Handbook , pages 1087–1108. CRC Press.
Moin Nadeem, Anna Bethke, and Siva Reddy.
2020. StereoSet: Measuring stereotypical bias
in pretrained language models. arXiv preprint
arXiv:2004.09456 .
Dong Nguyen, Rilana Gravel, Dolf Trieschnigg, and
Theo Meder. 2013. “How Old Do You Think I
Am?”: A Study of Language and Age in Twitter. In
Proceedings of the Conference on Web and Social
Media (ICWSM) , pages 439–448, Boston, MA.
Malvina Nissim, Rik van Noord, and Rob van der Goot.
2020. Fair is better than sensational: Man is to doctor as woman is to doctor. Computational Linguistics.
Debora Nozza, Claudia V olpetti, and Elisabetta Fersini.
2019. Unintended Bias in Misogyny Detection. In
Proceedings of the Conference on Web Intelligence ,
pages 149–155.
Alexandra Olteanu, Carlos Castillo, Fernando Diaz, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 69 | Proceedings of the Conference on Web Intelligence ,
pages 149–155.
Alexandra Olteanu, Carlos Castillo, Fernando Diaz,
and Emre Kıcıman. 2019. Social Data: Biases,
Methodological Pitfalls, and Ethical Boundaries.
Frontiers in Big Data , 2.
Alexandra Olteanu, Kartik Talamadupula, and Kush R.
Varshney. 2017. The Limits of Abstract Evaluation
Metrics: The Case of Hate Speech Detection. In
Proceedings of the ACM Web Science Conference ,
Troy, NY .
Orestis Papakyriakopoulos, Simon Hegelich, Juan Carlos Medina Serrano, and Fabienne Marco. 2020.
Bias in word embeddings. In Proceedings of the
Conference on Fairness, Accountability, and Transparency , pages 446–457, Barcelona, Spain.
Ji Ho Park, Jamin Shin, and Pascale Fung. 2018. Reducing Gender Bias in Abusive Language Detection.
InProceedings of Empirical Methods in Natural
Language Processing (EMNLP) , pages 2799–2804,
Brussels, Belgium.Ellie Pavlick and Tom Kwiatkowski. 2019. Inherent
Disagreements in Human Textual Inferences. Transactions of the Association for Computational Linguistics , 7:677–694. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 70 | Disagreements in Human Textual Inferences. Transactions of the Association for Computational Linguistics , 7:677–694.
Xiangyu Peng, Siyan Li, Spencer Frazier, and Mark
Riedl. 2020. Fine-Tuning a Transformer-Based Language Model to Avoid Generating Non-Normative
Text. arXiv preprint arXiv:2001.08764 .
Radomir Popovi ´c, Florian Lemmerich, and Markus
Strohmaier. 2020. Joint Multiclass Debiasing of
Word Embeddings. In Proceedings of the International Symposium on Intelligent Systems , Graz, Austria.
Vinodkumar Prabhakaran, Ben Hutchinson, and Margaret Mitchell. 2019. Perturbation Sensitivity Analysis to Detect Unintended Model Biases. In Proceedings of Empirical Methods in Natural Language
Processing (EMNLP) , pages 5744–5749, Hong
Kong, China.
Shrimai Prabhumoye, Elijah Mayfield, and Alan W.
Black. 2019. Principled Frameworks for Evaluating
Ethics in NLP Systems. In Proceedings of the Workshop on Innovative Use of NLP for Building Educational Applications , Florence, Italy.
Marcelo Prates, Pedro Avelar, and Luis C. Lamb. 2019. | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 71 | Marcelo Prates, Pedro Avelar, and Luis C. Lamb. 2019.
Assessing gender bias in machine translation: A
case study with google translate. Neural Computing
and Applications .
Rasmus Précenth. 2019. Word embeddings and gender
stereotypes in Swedish and English. Master’s thesis,
Uppsala University.
Dennis R. Preston. 2009. Are you really smart (or
stupid, or cute, or ugly, or cool)? Or do you just talk
that way? Language attitudes, standardization and
language change. Oslo: Novus forlag , pages 105–
129.
Flavien Prost, Nithum Thain, and Tolga Bolukbasi.
2019. Debiasing Embeddings for Reduced Gender
Bias in Text Classification. In Proceedings of the
Workshop on Gender Bias in Natural Language Processing , pages 69–75, Florence, Italy.
Reid Pryzant, Richard Diehl Martinez, Nathan Dass,
Sadao Kurohashi, Dan Jurafsky, and Diyi Yang.
2020. Automatically Neutralizing Subjective Bias
in Text. In Proceedings of the AAAI Conference on
Artificial Intelligence (AAAI) , New York, NY . | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 72 | 2020. Automatically Neutralizing Subjective Bias
in Text. In Proceedings of the AAAI Conference on
Artificial Intelligence (AAAI) , New York, NY .
Arun K. Pujari, Ansh Mittal, Anshuman Padhi, Anshul Jain, Mukesh Jadon, and Vikas Kumar. 2019.
Debiasing Gender biased Hindi Words with Wordembedding. In Proceedings of the International
Conference on Algorithms, Computing and Artificial
Intelligence , pages 450–456.
Yusu Qian, Urwa Muaz, Ben Zhang, and Jae Won
Hyun. 2019. Reducing gender bias in word-level
language models with a gender-equalizing loss function. In Proceedings of the ACL Student Research
Workshop , pages 223–228, Florence, Italy.
Shauli Ravfogel, Yanai Elazar, Hila Gonen, Michael
Twiton, and Yoav Goldberg. 2020. Null It Out:
Guarding Protected Attributes by Iterative Nullspace
Projection. In Proceedings of the Association for
Computational Linguistics (ACL) .
John R. Rickford and Sharese King. 2016. Language
and linguistics on trial: Hearing Rachel Jeantel (and | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 73 | Computational Linguistics (ACL) .
John R. Rickford and Sharese King. 2016. Language
and linguistics on trial: Hearing Rachel Jeantel (and
other vernacular speakers) in the courtroom and beyond. Language , 92(4):948–988.
Anthony Rios. 2020. FuzzE: Fuzzy Fairness Evaluation of Offensive Language Classifiers on AfricanAmerican English. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI) , New York,
NY .
Gerald Roche. 2019. Articulating language oppression: colonialism, coloniality and the erasure of Tibetâ˘A´Zs minority languages. Patterns of Prejudice .
Alexey Romanov, Maria De-Arteaga, Hanna Wallach, Jennifer Chayes, Christian Borgs, Alexandra
Chouldechova, Sahin Geyik, Krishnaram Kenthapadi, Anna Rumshisky, and Adam Tauman Kalai.
2019. What’s in a Name? Reducing Bias in Bios
without Access to Protected Attributes. In Proceedings of the North American Association for Computational Linguistics (NAACL) , pages 4187–4195,
Minneapolis, MN.
Jonathan Rosa. 2019. Contesting Representations of | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 74 | Minneapolis, MN.
Jonathan Rosa. 2019. Contesting Representations of
Migrant “Illegality” through the Drop the I-Word
Campaign: Rethinking Language Change and Social Change. In Netta Avineri, Laura R. Graham,
Eric J. Johnson, Robin Conley Riner, and Jonathan
Rosa, editors, Language and Social Justice in Practice. Routledge.
Jonathan Rosa and Christa Burdick. 2017. Language
Ideologies. In Ofelia García, Nelson Flores, and
Massimiliano Spotti, editors, The Oxford Handbook
of Language and Society . Oxford University Press.
Jonathan Rosa and Nelson Flores. 2017. Unsettling
race and language: Toward a raciolinguistic perspective. Language in Society , 46:621–647.
Sara Rosenthal and Kathleen McKeown. 2011. Age
Prediction in Blogs: A Study of Style, Content, and
Online Behavior in Pre- and Post-Social Media Generations. In Proceedings of the North American Association for Computational Linguistics (NAACL) ,
pages 763–772, Portland, OR.
Candace Ross, Boris Katz, and Andrei Barbu.
2020. Measuring Social Biases in Grounded Vision and Language Embeddings. arXiv preprint
arXiv:2002.08911 . | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 75 | 2020. Measuring Social Biases in Grounded Vision and Language Embeddings. arXiv preprint
arXiv:2002.08911 .
Richard Rothstein. 2017. The Color of Law: A Forgotten History of How Our Government Segregated
America . Liveright Publishing.David Rozado. 2020. Wide range screening of algorithmic bias in word embedding models using large
sentiment lexicons reveals underreported bias types.
PLOS One .
Elayne Ruane, Abeba Birhane, and Anthony Ventresque. 2019. Conversational AI: Social and Ethical Considerations. In Proceedings of the Irish Conference on Artificial Intelligence and Cognitive Science, Galway, Ireland.
Rachel Rudinger, Chandler May, and Benjamin
Van Durme. 2017. Social bias in elicited natural language inferences. In Proceedings of the Workshop
on Ethics in Natural Language Processing , pages
74–79, Valencia, Spain.
Rachel Rudinger, Jason Naradowsky, Brian Leonard,
and Benjamin Van Durme. 2018. Gender Bias
in Coreference Resolution. In Proceedings of the
North American Association for Computational Linguistics (NAACL) , pages 8–14, New Orleans, LA.
Elizabeth B.N. Sanders. 2002. From user-centered to
participatory design approaches. In Jorge Frascara, | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 76 | Elizabeth B.N. Sanders. 2002. From user-centered to
participatory design approaches. In Jorge Frascara,
editor, Design and the Social Sciences: Making Connections , pages 18–25. CRC Press.
Brenda Salenave Santana, Vinicius Woloszyn, and Leandro Krug Wives. 2018. Is there gender bias and
stereotype in Portuguese word embeddings? In
Proceedings of the International Conference on the
Computational Processing of Portuguese Student Research Workshop , Canela, Brazil.
Maarten Sap, Dallas Card, Saadia Gabriel, Yejin Choi,
and Noah A. Smith. 2019. The risk of racial bias in
hate speech detection. In Proceedings of the Association for Computational Linguistics (ACL) , pages
1668–1678, Florence, Italy.
Maarten Sap, Saadia Gabriel, Lianhui Qin, Dan Jurafsky, Noah A. Smith, and Yejin Choi. 2020. Social
Bias Frames: Reasoning about Social and Power Implications of Language. In Proceedings of the Association for Computational Linguistics (ACL) .
Hanna Sassaman, Jennifer Lee, Jenessa Irvine, and
Shankar Narayan. 2020. Creating CommunityBased Tech Policy: Case Studies, Lessons Learned,
and What Technologists and Communities Can Do | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
2005.14050 | 77 | Shankar Narayan. 2020. Creating CommunityBased Tech Policy: Case Studies, Lessons Learned,
and What Technologists and Communities Can Do
Together. In Proceedings of the Conference on Fairness, Accountability, and Transparency , Barcelona,
Spain.
Danielle Saunders and Bill Byrne. 2020. Reducing
Gender Bias in Neural Machine Translation as a Domain Adaptation Problem. In Proceedings of the Association for Computational Linguistics (ACL) .
Tyler Schnoebelen. 2017. Goal-Oriented Design for
Ethical Machine Learning and NLP. In Proceedings
of the Workshop on Ethics in Natural Language Processing , pages 88–93, Valencia, Spain.
Sabine Sczesny, Magda Formanowicz, and Franziska
Moser. 2016. Can gender-fair language reduce gender stereotyping and discrimination? Frontiers in
Psychology , 7.
João Sedoc and Lyle Ungar. 2019. The Role of Protected Class Word Lists in Bias Identification of Contextualized Word Representations. In Proceedings
of the Workshop on Gender Bias in Natural Language Processing , pages 55–61, Florence, Italy.
Procheta Sen and Debasis Ganguly. 2020. Towards Socially Responsible AI: Cognitive Bias-Aware MultiObjective Learning. In Proceedings of the AAAI | 2005.14050 | Language (Technology) is Power: A Critical Survey of "Bias" in NLP | We survey 146 papers analyzing "bias" in NLP systems, finding that their
motivations are often vague, inconsistent, and lacking in normative reasoning,
despite the fact that analyzing "bias" is an inherently normative process. We
further find that these papers' proposed quantitative techniques for measuring
or mitigating "bias" are poorly matched to their motivations and do not engage
with the relevant literature outside of NLP. Based on these findings, we
describe the beginnings of a path forward by proposing three recommendations
that should guide work analyzing "bias" in NLP systems. These recommendations
rest on a greater recognition of the relationships between language and social
hierarchies, encouraging researchers and practitioners to articulate their
conceptualizations of "bias"---i.e., what kinds of system behaviors are
harmful, in what ways, to whom, and why, as well as the normative reasoning
underlying these statements---and to center work around the lived experiences
of members of communities affected by NLP systems, while interrogating and
reimagining the power relations between technologists and such communities. | http://arxiv.org/pdf/2005.14050 | [
"Su Lin Blodgett",
"Solon Barocas",
"Hal Daumé III",
"Hanna Wallach"
] | [
"cs.CL",
"cs.CY"
] | null | null | cs.CL | 20200528 | 20200529 | [] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.